Principia Humanitatis, London, November 2026

Humans Are Not Estuaries

A disputation on anthro-complexity, reflexivity, and the observer who will not stay outside
A river shapes a city, and the city's planners argue about the river. The planners also argue with one another, and someone is drawing the map.
Mark DownhamAuthor
253,279Words of argument
63Chapters and apparatus
440Turnable explainer cards
Begin reading
Opening

The Argument in Brief, the Theses, and the Method

4,580 words, about 20 minutes of reading

The Argument in Brief

On 13 November 2026 Dave Snowden published a post called Not a branch of the tree. It continues his anthro-complexity series, and its subject is the systems-changes programme that David Ing has developed with Gary Metcalf and the Systems Changes Learning Circle in Toronto. The post is generous where generosity is owed and firm where firmness is due. It contests one move in particular: placing complexity science inside systems thinking, as one tradition among several on a map of traditions. Against that move Snowden proposes a different picture, the one in the photograph at the head of his post. Roots grip an isolated rock in front of a waterfall. Tree, rock and water are three actants, each constraining and enabling the others, and none contains or descends from any other. Neighbours, he says, not branches.

This disputation accepts that picture and asks what happens when it is turned round to face its author.

The post does more than defend complexity's independence from systems thinking. In the course of the argument it sets out, more clearly than any single earlier text, the elements that make anthro-complexity a body of thought in its own right rather than a brand. Those elements are: coherence found rather than wholes presumed; actants rather than variables; constraints and constructors; dispositional states and vectors; propensity rather than causality; retrospective coherence; kairos rather than chronos; Peircean pragmatism; trialectic rather than dyad; behaviour as the shadow of conditions; distributed sensing; and a Pelagian rather than an Augustinian stance towards intervention. Above all it names the move that gives the school its name. Human systems are complex, but they differ from other complex systems, including other living ones, because humans hold several identities at once, act with intention, and create abstractions and shared narratives that change the constraints they operate within. Humans are not ants. A river shapes a city, and the city's planners argue about the river; both are actants, but only one of them can change the story it tells about the other. That asymmetry, Snowden writes, is where anthro-complexity starts.

The thesis of this book can be put in one sentence:

ANTHRO-COMPLEXITY IS RIGHT THAT HUMAN COMPLEXITY DIFFERS IN KIND —

AND WRONG ABOUT WHERE THE DIFFERENCE LIES —

AND BECAUSE IT IS WRONG ABOUT WHERE, IT BUILDS INSTRUMENTS THAT TREAT HUMANS AS THE NON-HUMAN COMPLEXITY IT SAYS THEY ARE NOT.

The argument runs in three movements.

The first movement is reconstruction. Before anything is contested it must be stated at its strongest. Book One lays out the claim, assembling the elements of anthro-complexity from the November post and from the wider work it summarises: the Cynefin framework, the work on constraints with Alicia Juarrero, estuarine mapping, SenseMaker®, the trialectics series and the abductive warrant. Book Two places anthro-complexity among its neighbours. The first is the complex adaptive systems tradition of Holland, Gell-Mann, Kauffman and the Santa Fe Institute, from which anthro-complexity must be distinguished if its name is to mean anything. The second is the systems-thinking lineage that Snowden has just declined to join. The third is a neighbour he does not name in the post but whose shadow falls across all of it: the complex responsive processes tradition of Stacey, Griffin, Shaw and Mowles, which reached many of the same conclusions about human complexity from the opposite direction. Book Three asks why anthro-complexity treats human complexity differently. It finds five distinct reasons, which are scientific, reflexive, anti-computational, ethical and positional. It then argues that the distinction does three jobs at once, and that these jobs work against one another.

The second movement is challenge. Book Four advances eight theses against anthro-complexity, each tested against the strongest available reply:

  1. The possessive fallacy. Human distinctiveness is placed in properties that persons have (identities, intentions, abstractions) rather than in the relational processes through which persons are formed. Having refused to presume wholes, anthro-complexity presumes parts.
  2. The asymmetry principle applies to itself. If the distinctive human capacity is to tell stories that change constraints, then Cynefin, SenseMaker and estuarine mapping are such stories. The theory exempts its own narrative from the asymmetry it calls foundational.
  3. Humans are not estuaries. The theory that forbids modelling people on ants models practice on tides, rivers, waterfalls, surfing and sailing. Those images share two properties that human systems lack: the medium is not reflexive, and the reader stands outside the reading.
  4. The under-socialised constraint. Governing and enabling constraints are joined by constructors borrowed from physics. But the constraints that make a human community what it is are constitutive, reproduced in the very act of complying with or mapping them. And no human constructor is left unchanged by its work.
  5. The missing term is power. The three marks of the human are capacities, and none of them is power. Snowden applies a power critique to Ing's eurhythmia ("talk of restoring balance tends to favour whoever defines it"), but not to the design of his own signifiers, the destination of his own patterns, or the gradient down which distributed risk flows.
  6. An anti-computational ontology with a computational practice. The brain is not a programme, yet the method chain converts human noticing into machine-readable signal, and recursive training erases the tails where weak signals live.
  7. The triad is a simplex. A barycentric triangle whose values sum to one builds a zero-sum trade-off into the instrument. In that respect it is more dichotomising than two independent scales. And the Peircean thirdness it claims to honour, the interpretant, sits with the analyst, not the narrator.
  8. Reflexivity predicts that the instruments will degrade. Starlings do not learn the camera. Humans learn the signifier set. The more true the human distinction is, the faster narrative instruments become counter-performative: Goodhart, Lucas, Campbell and the social Zeno effect.

Book Five brings the wider corpus to bear: Mead's conversation of gestures, Elias's figuration and his critique of the closed person, Juarrero's three kinds of constraint, Ingold's meshwork, Hutchins's distributed cognition, Peirce's thirdness and community of inquirers, Bourdieu's habitus, Giddens's double hermeneutic, Hacking's looping kinds, and the literature on performativity from MacKenzie onwards. Book Six reads the whole question through the triad that runs through all of Principia Humanitatis, 道理勢: Dào, the way; Lǐ, the grain; Shì, the propensity. It argues that anthro-complexity has grasped the third term, lost the other two, and therefore lost the strategist's place inside the configuration, which Jullien, Snowden's own source, insisted on. Book Seven takes up the observer: Wigner's friend and the claim to stand "neither"; question-order effects, Zeno and decoherence as disciplines for the asker; Kipling's Mariner inside the whale and the grating he fits to its throat; and the crowds and Leviathans that only human collectives become.

The third movement is reconstruction of a different kind. Book Eight does not ask anthro-complexity to give up anything it has rightly won. It asks it to finish its own thought. A reflexive anthro-complexity would keep propensity, retrospective coherence, constraints and distributed sensing. It would add five things: identity as a position in a figuration rather than a possession; constitutive constraints; the reading of the reader; the tracking of power gradients; and consent to the harvest. It would replace alignment with recognition, the test being whether people can say we did this ourselves. The Tidal Loom of The Compass and the Loom is offered as one working form of that reconstruction, and its principles are applied here to SenseMaker and estuarine mapping. An illustrative case from the receiver's desk shows what changes when a fixed-charge receivership is read as reflexive human complexity rather than as a field of actants mapped from above. The Coda gathers the argument into the formula of the title:

ANTHRO-COMPLEXITY SAYS HUMANS ARE NOT ANTS.

THE HARDER CLAIM IS THAT HUMANS ARE NOT ESTUARIES EITHER —

AND NEITHER IS THE PERSON READING THE MAP.

The Twenty-Four Theses

The theses are numbered for reference throughout. Each is tested somewhere in the body of the work, and the Table of Verdicts in the Apparatus records whether it holds, holds in part, or fails. Following the Additive Covenant, later editions may add theses or amend them in place. None is ever silently removed.

Thesis 1. Anthro-complexity is a body of thought with its own ontology and not a brand or a method. It deserves to be argued with on its own terms, as Pepper's maxims require.

Thesis 2. Snowden is right that complexity is not a kind of systems thinking. The taxonomic move that folds Prigogine, Kauffman and Holling into a systems lineage buys coverage at the cost of distinction.

Thesis 3. Snowden is right that human systems differ in kind from other complex systems. The common practice of modelling organisations on ant colonies, flocks and agent-based simulations misleads.

Thesis 4. The reason anthro-complexity gives for that difference is misplaced. It puts the difference in properties persons have, not in the relational processes through which persons, identities and intentions are formed.

Thesis 5. Having refused to presume wholes, anthro-complexity presumes parts: discrete persons, each the bearer of identities, intentions and abstractions before relation. This is Elias's closed person in a complexity costume.

Thesis 6. The asymmetry principle (only humans can change the story they tell about the other) applies to the theory that states it. Cynefin, SenseMaker and estuarine mapping are stories that change constraints.

Thesis 7. Anthro-complexity exempts its own narrative-making from the asymmetry it calls foundational. In Stacey's terms, it keeps a formative teleology for the system and a rationalist teleology for the practitioner.

Thesis 8. The governing metaphors of anthro-complexity's practice are drawn from non-reflexive natural systems (tide, estuary, river, wave, waterfall). Each places the reader outside the reading.

Thesis 9. Jullien's 勢, the source of anthro-complexity's propensity, includes the strategist's own disposition in the configuration. A propensity mapped from outside is a landscape, not a 勢.

Thesis 10. The constraints that make a human community what it is are constitutive. They are reproduced in the act of complying with them, and in the act of mapping them.

Thesis 11. Constructor theory, imported to supply a class of actants, is a physics of transformation without change to the transformer. In human affairs everyone is a reagent, so the import cancels the human distinction it was meant to serve.

Thesis 12. The three marks of the human (multiple identities, intention, abstraction) are capacities. None of them is power. A theory of human complexity without power is CAS with better adjectives.

Thesis 13. Snowden's power critique of eurhythmia ("talk of restoring balance tends to favour whoever defines it") applies with equal force to whoever defines the signifier set, receives the pattern, and holds the residual risk.

Thesis 14. Self-signification is a real advance on survey and expert diagnosis. It moves interpretation partly to the narrator, but only partly, and on a frame the narrator did not choose.

Thesis 15. Anthro-complexity's ontology is anti-computational and its practice is increasingly computational. The harvest of narrative into machine-readable signal is where the contradiction bites.

Thesis 16. Recursive learning on harvested narrative erases the tails of the distribution, which are where weak signals live. A weak-signal instrument that feeds a model can starve itself.

Thesis 17. The barycentric triad builds a zero-sum trade-off into its geometry. It cannot tell strong-in-all from weak-in-all, and in that respect it dichotomises more than independent scales do.

Thesis 18. Peirce's thirdness is the interpretant. In narrative sensing the interpretant role is split between narrator and analyst, and the mediation the trialectic celebrates is where the designer sits.

Thesis 19. Human reflexivity predicts the degradation of any stable sensing instrument: order effects, the social Zeno effect, decoherence into cliché, and Goodhart's law. Non-human complex systems do not learn the camera; human systems do.

Thesis 20. Kairos, the turn of the tide, is a real quality of time in human affairs. But in human affairs the tide turns partly because someone visible was seen to wait for it.

Thesis 21. The distinction between human and other complexity does three jobs at once (ontological, methodological, institutional), and taking the first seriously undermines the other two.

Thesis 22. The drift from ontology to epistemology, energetics, decision and finally product is the path along which reflexivity is lost.

Thesis 23. A reflexive anthro-complexity is possible. It keeps everything Snowden has rightly won and adds identity as position, constitutive constraint, the reading of the reader, the power gradient, and consent to the harvest.

Thesis 24. Its test is recognition, not alignment: whether those inside can say we did this ourselves (我自然).

The Method of the Disputation

This book is a disputation in the scholastic sense. That is a practice with its own discipline, older than the modern essay, and a reader coming to it from management literature may find it strange. Its rules are stated here so the strangeness does not get in the way.

The adversary wins half

The disputation's first rule is that the opposing case is stated at its strongest before it is answered. In the medieval schools the respondent had to restate the opponent's argument to the opponent's satisfaction before replying. A reply to a weaker version was not a reply. This book honours that rule in the boxes marked The Case for the Other Side, which concede to the adversary everything that can honestly be conceded. Where a thesis is defeated by the adversary's case it is marked as failing, and the failure is kept in the text. Chapter 43 gathers the adversary's whole case into a single argument that Snowden himself might sign, and grants it the ground it wins.

There is a reason for this beyond courtesy. A critique of a theory of complex human systems that treats its object as simply mistaken has already failed. It has adopted the outside position it accuses its object of adopting. The disputation is therefore not a tribunal. It is a conversation of gestures in Mead's sense, a gesture whose meaning will be completed by the response it calls out. This book is itself a move in the field it describes, and it should be read that way.

Obligationes

The medieval grant-and-follow exercise was a structured exercise. A proposition was put forward, and the respondent had to grant, deny or doubt each subsequent proposition consistently with what had already been granted. The point was not to win but to discover what a commitment commits you to. The boxes marked Grant-and-follow exercise do this. They take a proposition anthro-complexity holds, grant it, and follow what else must then be granted. Several of the eight theses are built this way. They do not show that anthro-complexity is false. They show that its premises commit it to conclusions its practice does not draw.

Verdicts

Claims are given one of three verdicts, shown as coloured pills: holds, partial, fails. A verdict is a judgement made in this book at this time, and it can be revised. Under the Additive Covenant a revised verdict is amended in place with its date, and the earlier judgement stays visible.

Deeper

Boxes marked Deeper ("higher" or "deeper") hold technical excursions: the physics of quantum measurement, the geometry of the simplex, the biology of quorum sensing, the mathematics of performativity. They are not ornament. A disputation about complexity that borrows from physics and biology owes the reader an honest account of what is being borrowed. The rule for every crossing is the one stated in The Compass and the Loom: a structure may be carried from a source domain to the social only when the borrowed structure is real in the source, and the inference does not depend on the source's physics. Correlated sensing beats independent averaging passes that test. Intention bends random numbers does not.

The Additive Covenant

All the work of Principia Humanitatis follows a covenant: nothing is deleted, only inserted. Text may be amended, extended, qualified and contradicted, but always openly, in place and with a date. The build system for this volume enforces the covenant mechanically. It records the word count of every chapter and refuses to build if any chapter has shrunk. The covenant is more than a scruple. It is the method's own reflexivity: a book about systems that cannot be described from outside should not pretend to have been written from outside its own history.

A Note on Sources

The primary text under disputation is Dave Snowden's post Not a branch of the tree, published on 13 November 2026 as part of the anthro-complexity series begun in March of that year. Quotations from it are short and exact. Where the argument paraphrases Snowden, it tries to do so in terms he would recognise. The secondary primary text is the author's own The Compass and the Loom, a disputation on two Cynefin Company instruments, SwarmCompass and Weaving Risk and Strategy. This book develops several of its arguments and does not merely repeat them. Where the two books disagree, the disagreement is noted.

The wider literature appears through paraphrase and attribution, not quotation. Mead, Elias, Juarrero, Ingold, Hutchins, Peirce, Bourdieu, Giddens, Hacking, MacKenzie, Stacey, Griffin, Shaw, Mowles, Jullien, Kauffman, Cilliers, Holland, Gell-Mann, Prigogine, Boisot and Ashby are discussed in their own terms as far as the author's reading allows. The bibliography lists the works drawn on. No page numbers are given where the author has not checked them; a disputation about honest instruments should not begin with dishonest apparatus. Readers who find a source misrepresented are asked to say so. Snowden ends his post the same way: if I have misread either source, I would welcome the correction. The same invitation stands here, and it is meant.

Two further limits should be stated plainly. First, Snowden's post is one text, and anthro-complexity is a programme spread across years of writing, teaching and practice. Where the argument reaches beyond the post, the reach is flagged. Snowden has written at length elsewhere about the limits of agent-based modelling, about the multiple identities he calls crews and roles, and about the dangers of quantification, and some objections raised here may already be answered somewhere this book has not looked. Second, the illustrative case in Chapter 46 is a composite. It describes no real receivership, borrower, lender or property. Its details are invented for argument, and any resemblance to a particular case is accidental.

How to Read This Book

The book is long because the subject is large, not because the argument is complicated. The argument can be grasped from three places: the Argument in Brief above; the eight thesis chapters of Book Four, each of which opens with its thesis and closes with its verdict; and the Coda. A reader with an afternoon should read those. A reader with a week should read Books I–IV in order and then whichever of Books V–VII matches their interest: the social theory of Book Five, the Chinese philosophy of Book Six, or the physics and political theory of Book Seven. A practitioner who wants to know what to do differently on Monday should start with Book Eight and work backwards. The search box in the contents panel searches the whole text, and a reading-progress bar runs along the top of the page.

The book is not a hostile reading of Dave Snowden. It would not exist without him. Cynefin was one of the few ideas from management in the past quarter-century that told managers the truth about their situation: that much of what they face cannot be known in advance, and that pretending otherwise is a choice with costs. The anthro-complexity series is one of the most serious attempts by any practitioner-theorist to give that truth a rigorous foundation. Snowden takes on his opponents' strongest work, as he has just done with Ing, and changes his mind in public when the reading requires it, as he also does in the November post. This book tries to pay him the same compliment. Its claim is not that he is wrong about human complexity. It is that he is more right than his instruments allow, and that the remaining work is to bring the instruments up to the theory.

NOT A BRANCH OF THE TREE —

BUT A ROOT IN THE SAME ROCK —

AND THE WATER FALLING ON BOTH.

Paths Through the Book

The disputation is long, and readers come to it with different purposes. The following paths are suggestions. Each lists the chapters to read in order, with an estimate of the reading involved.

The argument in an afternoon

For a reader who wants the argument and its verdicts without the apparatus.

  1. The Argument in Brief and the Twenty-Four Theses (this Preliminaries)
  2. Chapter 4: The Human Cluster
  3. Chapter 14: The Reflexive Reason
  4. Chapter 19: Thesis I, The Possessive Fallacy
  5. Chapter 21: Thesis III, Humans Are Not Estuaries
  6. Chapter 43: The Case for the Other Side
  7. Chapter 47: Coda

For practitioners of anthro-complexity

For a reader who uses Cynefin, SenseMaker or estuarine mapping and wants to know what to do differently.

  1. Chapter 5: Method and Practice (to check the disputation's reading of the practice)
  2. Chapter 6: Genealogy (and its addition on Stage One)
  3. Chapter 20: Thesis II, the asymmetry applied to itself
  4. Chapter 23: Thesis V, power (and its addition, step by step)
  5. Chapter 25: Thesis VII, the simplex (and its addition, three stories)
  6. Chapter 26: Thesis VIII, degradation (and its addition, a research design)
  7. Chapter 44: Principles
  8. Chapter 45: Protocols
  9. Chapter 47, afterword: What Would Change the Author's Mind

For critics and scholars of complexity

For a reader interested in the theoretical argument and its sources.

  1. Chapters 2–3: the ontology and causality clusters
  2. Chapters 7–8: CAS and the twelve dimensions
  3. Chapter 10: systems thinking and the taxonomic move
  4. Chapter 11: complex responsive processes (and its addition, the baggy and the boxed)
  5. Chapters 13–15: the scientific, reflexive and anti-computational reasons
  6. Chapter 18: the three jobs of a distinction (and its addition)
  7. Chapters 27–34: the corpus
  8. Chapter 43: the adversary's case

For readers of classical Chinese thought

For a reader interested in 道理勢 and its bearing on complexity.

  1. Chapter 3: propensity, retrospect and the turn of the tide (for Jullien)
  2. Chapter 21: humans are not estuaries (Jullien against the surfer)
  3. Chapter 35: 勢
  4. Chapter 36: 理 (and its addition, the grain of a loan book)
  5. Chapter 37: 道
  6. Chapter 38: the dyad revisited
  7. Chapter 19, the section on Confucian role ethics

For readers interested in power and politics

  1. Chapter 12: critical systems thinking (and its addition, boundary critique applied)
  2. Chapter 16: the ethical reason (and its addition, a casuistry of probes)
  3. Chapter 17: the positional reason
  4. Chapter 23: Thesis V, power
  5. Chapter 28: Elias
  6. Chapter 42: crowds and Leviathan

For readers interested in physics and measurement

  1. Chapter 14: the reflexive reason
  2. Chapter 26: Thesis VIII, degradation
  3. Chapter 39: Wigner's friend (and its addition, a relational dashboard)
  4. Chapter 40: order, Zeno, decoherence
  5. Read with the analogical discipline in mind (Preliminaries, Deeper)

For lenders, receivers and those in property and credit

  1. Chapter 4: the human cluster
  2. Chapter 21: humans are not estuaries
  3. Chapter 36, addition: the grain of a loan book
  4. Chapter 41: the Mariner inside the whale
  5. Chapter 46: from the receiver's desk
  6. Chapter 16, addition: a casuistry of probes (Case 5)

The Additiones

Under the Additive Covenant, chapters may be extended by additiones, sections added after the chapter's first composition, marked with the heading Addition. They are part of the chapter and are kept with it. They are listed here so that readers can find them.

Chapter Addition
2 Archer's objection, and the limits of relationism
3 Lewin recovered
6 Stage One as a resource
7 Complexity economics and the economy in formation
9 The alpha wolf and the captive observer
10 Ing's programme given its due
11 The baggy and the boxed, revisited
12 Boundary critique applied to estuarine mapping and SwarmCompass
13 Weick and the sensemaking tradition
14 Luhmann and the blind spot
15 The anti-computational reason after large language models
16 A casuistry of probes
18 Three jobs in other schools
20 A ledger of anthro-complexity's performatives
21 Eyes on the organisation
23 Power in a narrative study, step by step
25 Three stories on one triangle
26 A research design for testing Thesis VIII
29 Constraint and power, a synthesis
31 A bridge for eighty thousand
34 The performativity of "complexity" itself
36 The grain of a loan book
39 Sketch of a relational dashboard
41 Gratings already in the world
42 The six thresholds as a convener's check
44 The reconstruction on one page
46 What the receiver's position teaches the consultant
47 Afterword: what would change the author's mind

The Boxes

The disputation's apparatus of boxes is used consistently:

  • Thesis boxes (dark blue, gold border) state a thesis and, where reached, its verdict.
  • The Case for the Other Side boxes* (parchment) concede the adversary's case where it holds.
  • Grant-and-follow exercise boxes* (white, dashed border) follow what a granted proposition commits one to.
  • Deeper boxes* (pale green, sea-coloured border) give technical excursions.
  • Capital formulas (centred, between gold rules) gather each chapter's argument into a few lines.

Verdict pills mark judgements: holds, partial, fails.

Addition: How This Edition Is Organised

This edition has grown by addition, and a reader needs a plan of it.

Nine books. Book One sets out anthro-complexity in twenty-two elements, each followed by a slow reading with a worked case and by a step-by-step account of the methods. Book Two places it among its neighbours. Book Three asks why it treats human complexity differently. Book Four states eight theses against it. Book Five brings a body of social theory to bear. Book Six reads it through three Chinese terms. Book Seven treats the observer inside. Book Eight reconstructs. Book Nine was added after the coda. It answers two further posts by Snowden, applies Rahel Jaeggi's account of criticism, draws on A Web of Relations, Flinn's book and the Zen Garden study, challenges the doctrine of incompressibility, and gathers fourteen dialectics.

The cards. Every blue term opens an explainer. Each chapter ends with a deck of the cards for its terms. Argument decks set out each thesis in three cards: claim and premises, evidence with the strongest objection and reply, and consequence with test and verdict. The Lexicon at the end holds every card.

The labels. Earlier editions used Latin for the books and for the devices of the disputation. This edition uses English: Book, addition, for the opposing side, deeper water, and grant and follow.

Three shorter paths. A reader who wants the theory explained should read Chapters 2 to 5 with their slow readings. A reader who wants the argument should read Chapters 19 to 26 and then 50, 52 and 56. A reader who wants the outcome should read Chapters 44, 45 and 55.

The seat. The Garden study of Chapter 53 holds that every view is from a seat. This book is written by one author who has not worked inside the practice it examines. Article 10 of the adversary's case says so, and it belongs at the front as well as at the back.

Chapter 1

The Occasion: A Tree Gripping a Rock

3,700 words, about 16 minutes of reading

A Post in a Series

Some texts are occasions, and some occasions become texts. Not a branch of the tree is both. It is the latest instalment in a series Snowden began in March 2026 "to better understand anthro-complexity by setting it against other bodies of thought". That is a method of definition by contrast, and an old one. The scholastics defined a thing by genus and difference. Saussure held that in language there are only differences without positive terms. Snowden's series defines anthro-complexity by its neighbours: semiotics, the new materialisms, the ecological anthropology of Ingold, and now the systems-changes programme of David Ing. Each post marks a border. Taken together, the borders outline a territory.

This chapter does three things. It describes the occasion: who Ing is, what his programme claims, and why Snowden chose it. It follows the shape of Snowden's argument, which moves through shared ground to a single contested move and on to three areas of real conflict. And it reads the image Snowden chose for the head of his post, a tree's roots gripping a rock before a waterfall, as a theory in its own right, because that is what Snowden asks us to do with it.

The Interlocutor

David Ing spent twenty-eight years at IBM; Snowden spent seven. Both taught at Hull, the university that gave British systems thinking much of its institutional home through Checkland's soft systems methodology, Flood and Jackson's critical systems thinking, and the work of the Centre for Systems Studies. Both draw on Emery and Trist, on the pragmatists, and on the ecological turn in psychology and anthropology. The shared history matters, because the disagreement that follows is between people who have read the same books and worked in the same kind of institution. It is not a quarrel between strangers. Snowden says plainly that there is more common ground than he expected.

Snowden works from two of Ing's texts. The first is the 2023 paper with Gary Metcalf in the Journal of the International Society for the Systems Sciences, Reframing Systems Thinking for Systems Changes: Sciencing and Philosophizing from Pragmatism towards Processes as Rhythms. Snowden calls it a careful piece of intellectual history. It traces the Trist–Emery and Churchman–Ackoff partnerships back through Pepper and Singer to the Metaphysical Club, and then proposes a new world hypothesis, (con)texturalism-dyadicism, drawing on Classical Chinese Medicine via Keekok Lee. The second is Ing's 2022 Barcelona lecture, Knowing Better via Systems Thinking. It takes its map of traditions from Ramage and Shipp's Systems Thinkers and favours three of them: general systems theory, soft and critical systems, and learning systems.

It helps to state Ing's programme on its own terms before following Snowden's response. It is a serious programme. Ing wants a systems thinking adequate to a world in which systems change, not one that describes systems as stable objects. He finds the resources for it in a reading of the pragmatist inheritance that runs from James through Perry to Pepper's root metaphors, and from Singer to Churchman and Ackoff. Following Pepper, he holds that there are a small number of "world hypotheses", each grounded in a root metaphor: formism, mechanism, organicism, contextualism, and (Pepper's late addition) selectivism. He argues that systems thinking has moved through several of them. To these he adds a new hypothesis drawn from Chinese thought. Its root metaphor is texture, the weave of strands over time. Its logic is dyadic: yin and yang as inseparable, mutually transforming modes, not opposed poles. Its practical aim is eurhythmia, a healthy rhythm neither too fast nor too slow, relative to its context. Its model of the practitioner is the Classical Chinese physician who reads the pulse against the season.

This is not a straw man, and Snowden does not treat it as one. He finds in it a critique of Lewin's mythical three-step model, a scepticism about the systems-change industry, a theory of propensity drawn from Jullien, a retrospective theory of truth drawn from Robert Rosen, an ecological turn through Gibson and Ingold, a recovery of Emery and Trist's turbulent field, and a preference for kairos over chronos. All of these he can endorse. The disagreement comes later, and in one place.

The Shape of the Argument

The post has a clear architecture, worth setting out because the architecture is itself an argument.

First, shared ground. Snowden lists seven areas of agreement:

  • Lewin never wrote "refreeze". The three-step model is a later reconstruction, and the refreezing stage first appears in a 1950 conference paper by Festinger.
  • The systems-change industry applies its label to almost anything. Most of what it sells carries an "Augustinian" assumption: the system lacks something that an outside intermediary must supply.
  • Propensity matters more than causality. You work with the dispositional state of a complex system and move with its propensity, as a surfer reads the wave, rather than being pulled, as a water-skier is pulled by the boat.
  • Looking back is more reliable than looking forward. Rosen's entailment allows an outcome to be traced back through what entailed it, but does not support forward-looking causality. Cynefin has always said something close to this about the complex domain: cause and effect are coherent only in retrospect.
  • The ecological turn through Gibson's affordances and Ingold's anthropology. Estuarine mapping began life as affordance mapping.
  • Emery and Trist's turbulent field, where the dynamics arise from the ground and not only from the interaction of identifiable systems.
  • Practical wisdom and kairos: phronesis over episteme and techne, opportune time over clock time.

Second, the origins in pragmatism, put to different uses. Ing's pragmatism, in Snowden's reading, is a pragmatism of worldviews and of planning. Snowden's is Peircean: a pragmatism of signs, abduction and habit, grounded in a commitment to the natural sciences that the original Metaphysical Club shared. The difference matters later, when the argument turns to dyads and triads.

Third, the taxonomic move. This is the move Snowden contests directly. In the lecture, complexity theory appears as one tradition on Ramage and Shipp's map and is then set aside in favour of three others. In the paper, the eras of systems thinking run from mechanism through social systems to ecology. Resilience thinking, panarchy, Holling, Ostrom and Rosen are absorbed into the systems lineage, while complexity science as such, with its origins in physics, chemistry and theoretical biology and its specific claims about emergence, constraints and dispositionality, does not appear. Snowden puts the consequence sharply: complexity is subsumed in two ways, placed on the map and then set aside, and absorbed piecemeal into a systems narrative. He answers with three arguments. One is from origins: complexity comes from the natural sciences, not from operational research or social intervention. One is from scope: a definition of systems thinking broad enough to include anything concerned with relationships cannot make distinctions. One is from Pepper's own maxims: each world hypothesis is autonomous and should stand on its own evidence, and eclecticism is confusing. He adds an institutional story. Systems thinking offers structure, which corporations like, and that may say something about which body of thought an institution finds easy to host.

Fourth, three areas of conflict. Snowden sets these out under three headings:

  • Wholes, presumed or found. Systems thinking presumes wholes and infers parts. Anthro-complexity presumes neither, and finds coherence in practice.
  • Humans, animals and intention. Ing's programme deliberately downplays what is distinctive about humans ("attention trumps intention"). Anthro-complexity makes exactly the opposite move.
  • Dyads and triads. Ing's dyadicism, however carefully distinguished from dualism, will be turned into dichotomies by people who think in pairs. Snowden's trialectic, grounded in Peirce's thirdness, keeps the system open.

A fourth section, on eurhythmia, records where Snowden revised his first reading. He had feared the medical language put the practitioner in the Augustinian position, but the paper favours self-healing. Two concerns remain: the norm of health itself, and the question of who does the reading.

Fifth, the conclusion: neighbours, not branches. Ing's paper ends with the line "Textures can be systems; not all systems are textures." Snowden offers a parallel: complex systems can be looked at systemically, but "complexity is not a kind of systems thinking".

Sixth, a postscript of honest limits. He has worked from one paper and one lecture deck. The map of traditions is Ramage and Shipp's, not Ing's. Reading the paper in full changed his view on two points. Systems thinking is not monolithic, and Jackson's critical systems thinking engages with complexity far more directly. The dyad-and-triad point is the one he holds most loosely.

What the Architecture Says

Read as a whole, the architecture tells us something before any single claim is examined. Snowden's method is differential. He defines anthro-complexity by marking where it departs from a neighbour who shares much of its ground. The shared ground is not a concession made before the attack. It is part of the definition: these are the commitments anthro-complexity holds in common with the most sophisticated contemporary systems thinking. The differences then mark what is proper to anthro-complexity alone.

It follows that the three areas of conflict are where anthro-complexity's identity is concentrated. If we want to know what anthro-complexity is, as distinct from what it shares, we should look at what it says about wholes, about humans, and about triads. This book will argue that the second of these, the human, is the deepest, and that the other two depend on it. The question of wholes becomes the question of what a human whole is, and therefore of who finds coherence and how. The question of triads becomes the question of how humans think and what structures change their thinking. The question of the human is where the name comes from. Snowden says so himself: that asymmetry is where anthro-complexity starts.

There is a second thing the architecture says, and it is easier to miss. Snowden's critique of Ing is a critique of a position. Ing's systems thinking keeps the synthetic mode of reasoning, in which wholes are presumed. It keeps, through eurhythmia, a norm against which disorder is diagnosed. And it keeps, through the physician metaphor, a reader who takes the pulse while the body does not. In each case Snowden objects to the place where the theorist or practitioner stands: outside the system, presuming its whole, holding its norm, reading its rhythm. This is the right objection. It is also the objection this book will make to Snowden, and it is a measure of the post's seriousness that it supplies the terms of its own critique. In the paragraph on eurhythmia Snowden asks who does the reading? That question, asked of anthro-complexity's own instruments, is the hinge on which this book turns.

The Image at the Head of the Post

Snowden ends his post by explaining its photograph. Cropped from an original by Zach Reiner, it shows the roots of a tree gripping an isolated rock in front of a waterfall. Darwin's tree, the tree of descent that serves him so well for living things, is set against this other tree, which is not a branch of anything:

In Snowden's account, the tree is where it is because a seed happened to lodge there, its roots followed whatever cracks the rock offered, and the falling water behind it supplied the conditions of its life. Nothing in that history was designed.

Over time, Snowden says, the roots will widen the cracks and change the rock, as the water has already shaped it. Tree, rock and water are three actants, each constraining and enabling the others; none contains the others, and none is descended from them. This, he suggests, is a better image of how complexity and systems thinking stand to each other than the family tree: different things formed by different flows, which have come to share a place and now shape one another.

The image is good, and the use of it is careful. It deserves the kind of reading Snowden gives his sources, because it is not just an illustration. It is a small model of anthro-complexity's ontology:

  • Coherence is found, not presumed. The tree did not plan its position. A seed lodged where it could, and the arrangement is the trace of a history no one designed.
  • Constraints, not causes. The rock does not cause the tree. It allows certain cracks, and the roots find them. The constraint is enabling as well as limiting.
  • Co-evolution. The roots widen the cracks and the water shapes the rock. Each actant modulates the conditions of the others.
  • A triad, not a dyad. Three actants, none reducible to a relation between the other two. The water mediates between tree and rock, the rock mediates between water and tree, and the tree mediates between rock and water.
  • No containment and no descent. None of the three is a part of a whole formed by the others, and none is the ancestor of another. They are neighbours.

All of this is true, and it is a strong picture. Now ask what the picture leaves out.

It contains no human. The image of how two bodies of human thought stand to each other is an image of non-human nature. Tree, rock and water do not tell stories about one another. The tree does not have several identities. The rock does not act with intention. The water does not form abstractions that change the constraints it works within. Each of the properties Snowden will say makes human complexity distinctive is absent from the image chosen to represent the relation between two human traditions.

It has a viewer who is not in it. The photograph was taken by someone standing somewhere, and the crop was made by someone else for a purpose. The meaning was supplied by a third person, in a post that is itself a move in an argument. This is not pedantry. In the image, the three actants are visible as three because there is a viewer outside the frame to whom the arrangement appears. Inside the arrangement there is no "three". There is the pressure of water, the resistance of stone and the growth of root. The triad is a fact relative to an observer, not of the waterfall alone.

It shows no conversation. The relation Snowden wants to describe, between complexity and systems thinking, is conducted in papers, lectures, posts, citations, conferences and consulting proposals. In that relation each party changes the story it tells about the other, and the stories change the field. Ing's map places complexity among the systems traditions. Snowden's post places it outside. Each placement is a gesture that calls out a response. The waterfall image has no gestures in it. It is an image of mutual constraint without mutual interpretation.

None of this is a fault in the photograph. It is a fine photograph, and the use made of it is honest and illuminating. But the reading shows, in miniature, the pattern this book will trace through anthro-complexity's theory and instruments. The theory insists that human complexity differs in kind because humans narrate, intend and abstract. The images through which the theory pictures complexity, and through which it pictures its own relations, are drawn from non-reflexive nature, viewed from outside. Tree, rock and water; tide and estuary; wave and surfer; river and city. In each case the human, if present at all, is the reader of the scene, not one of the things in the scene being read.

Proposition granted: Complexity and systems thinking stand to each other as tree, rock and water: three actants, each constraining and enabling the others, none containing or descended from them.

Then it must be granted that the relation between two bodies of human thought is pictured adequately by an arrangement of non-human actants. — Granted, as a picture.

Then it must be granted that whatever is distinctive about human complexity, namely narrative, intention and abstraction, is not needed to picture the relation between two human traditions. — Doubted. The relation between complexity and systems thinking is carried on almost entirely in narrative, intention and abstraction: maps, lineages, definitions, posts. If these are not needed to picture it, they are not doing the work anthro-complexity says they do.

Then the respondent must choose. Either the image is a picture of the non-human aspect of the relation (the institutional, material and economic constraints that shape two traditions) and is silent on the human aspect; or the human aspect is not distinctive after all. — The first horn is the honest one, and it is the one this book takes. The waterfall shows how two traditions constrain one another. It does not show how they read one another. And reading is where anthro-complexity says human complexity begins.

A Second Image

Set beside Snowden's photograph a different image, not to replace it but to complete it. Two people stand at the edge of a pool below a waterfall. A rock rises from the water, and a tree's roots grip it. One person is photographing the tree. The other is reading a map of the river, and the map was drawn by a third person who is not present. The photographer frames the scene with tree, rock and water and calls it three actants. The map-reader sees the rock as a hazard to navigation and the tree as evidence of the rock's stability. The absent cartographer drew the river as a channel, its rocks as soundings, its trees as markers. Each of the three human figures is reading the same arrangement. Each is reading it differently, according to what they are there to do, who they are with, and whose map they carry. Each reading changes what the arrangement will become. The photograph will be published. The map will guide a boat that may strike the rock or avoid it. The cartographer's channel will be dredged, or not, by an authority that relied on it.

This second image contains everything in the first, and it adds what anthro-complexity says is distinctive about human complexity: multiple identities (photographer, navigator, cartographer), intention (to frame, to pass, to chart), and abstractions that change constraints (the photograph, the map, the channel). It also adds three things the first image cannot show:

  • Positions. Each reader stands somewhere, and the standing shapes the reading.
  • Power. The cartographer, who is absent, has more influence over what happens to the rock than either person present, because the cartographer's abstraction has been adopted by an authority.
  • Reflexivity. If the photographer publishes the image with a caption calling it a picture of how two traditions stand to each other, the photograph becomes an actant in the relation it depicts.

This book argues that anthro-complexity's theory is about the second image and its practice is organised around the first.

THREE ACTANTS IN THE FRAME —

THREE READERS OUTSIDE IT —

AND THE READERS ARE WHERE THE HUMAN BEGINS.

Why Ing, and Why Now

One more question about the occasion is worth asking: why this interlocutor at this moment? Snowden says the series aims to understand anthro-complexity by setting it against other bodies of thought, and Ing's programme is an obvious candidate. It is the most philosophically developed contemporary attempt to rebuild systems thinking for a world of change. It draws on many of the sources Snowden draws on: pragmatism, Gibson, Ingold, Jullien, Rosen. It makes a move, the absorption of complexity into a systems lineage, that Snowden has spent much of his career resisting.

There is a further reason, and Snowden gives it himself in the institutional story about IBM. Systems thinking, he says, was always more at home in large organisations than complexity science. It offers structure: models, boundaries, purposes, and a method that can be taught and rolled out. The absorption of complexity into systems thinking is therefore not only an intellectual error. It is an institutional pressure. It is what happens to complexity when organisations take it up: it is turned into a method that fits the structures organisations already have. Snowden's resistance to the taxonomic move is a resistance to that pressure.

Here the argument of this book has to begin carefully, because the institutional pressure Snowden describes does not act only on systems thinkers. It acts on anyone whose ideas are taken up by organisations. That includes the Cynefin Company, whose instruments, SenseMaker®, estuarine mapping, SwarmCompass and Weaving Risk and Strategy, are sold to organisations, taught in accredited programmes, and rolled out. The Compass and the Loom traced the drift by which the Cynefin lineage moves from ontology (what kind of system is this?) through epistemology, energetics and decision towards product, each step closer to the client's question. If Snowden is right that institutions find structure easier to host than complexity, then anthro-complexity faces the same pressure systems thinking faced. The question is whether it has resisted it.

That question cannot be answered by looking at the theory alone. It has to be answered by looking at where the theory puts the human, and then at where the instruments put the human. The next four chapters assemble the theory, element by element, from the November post and the wider work it summarises. They do so as sympathetically as possible, because the critique that follows has value only if what it criticises has first been shown to be worth criticising.

Addition: Cross-Lights From Book Nine

The post that occasioned this book is one of a series, and three other posts in Snowden's body of work bear on it. They are treated in Book Nine and should be named here.

"Knotty problems", of January 2024, contains two sentences on which Chapter 56 turns: that a knot comes undone easily if its structure is understood, and that all diagnosis is intervention. "A Trialectic for organisational practice & coaching", of May 2026, adds a third position to Augustine and Pelagius, the formative encounter, and Chapter 48 credits it with conceding much of what the theses ask. "Irritation is not an alibi", of 14 November 2026, appeared the day after the tree post. It states that nobody stands outside, the practitioner least of all, and Chapter 52 examines it.

A reader of the present chapter should therefore hold its criticisms as provisional. The disputation was written against a target that was moving, and moving its way.

Chapter 2

What Kind of Thing a Complex System Is

8,335 words, about 36 minutes of reading

The Order of Exposition

This chapter and the three that follow assemble the elements of anthro-complexity into a single account. They group the elements into four clusters. This chapter covers ontology: what kind of thing a complex system is. Chapter 3 covers causality and time. Chapter 4 covers the human. Chapter 5 covers method and practice. The grouping is the author's. Snowden does not present his position as a system, and in the November post he would resist the suggestion that it is one. Any account of a body of thought is a compression, though, and the honest course is to state the compression and its basis. The basis here is the order of dependence. The human cluster depends on the ontological, because what is distinctive about humans is distinctive as a kind of complex system. The practical cluster depends on all three, because what one does follows from what one takes the situation to be.

Each element is given in four parts:

  • the claim;
  • its sources, in the November post and in the wider corpus;
  • its function, meaning what it does in the architecture of the whole;
  • a flag, meaning a note of where later chapters will press on it.

The flags are not objections. They are placeholders, so that when the objections come the reader can see they were anticipated in the reading and not invented afterwards.

Element 1: Complexity Is a Claim About the World

The claim

The foundational claim of anthro-complexity concerns complexity itself, before anything is said about the human. Complexity is not a perspective one adopts. It is a claim about the nature of certain systems, with its own commitments. Snowden makes the point against Ing's broad definition of systems thinking as "a perspective on parts, wholes, and their relations". A perspective can be taken up or put down, combined with other perspectives, or chosen for convenience. A claim about the nature of a system is either true of that system or false. If some systems really are complex (if their behaviour arises from interactions that cannot be decomposed and recomposed predictably, if their causal structure is coherent only in retrospect, if their states are dispositional and not determinate), then that is a fact about them. Treating them as ordered is not a different perspective. It is a mistake.

The sources

The claim has two sources. The first is the history of complexity science, which Snowden places in the natural sciences: physics, chemistry and theoretical biology. Prigogine's far-from-equilibrium thermodynamics, Kauffman's work on autocatalytic sets and the origins of order, the study of self-organised criticality, and the mathematics of nonlinear dynamics are not perspectives on chemical reactions or ecosystems. They are claims about how certain physical and biological systems behave, tested against how those systems actually behave. When Snowden says complexity "comes from the natural sciences", he is not making a sociological point about where its founders worked. He is making an epistemological point about what kind of claim it is.

The second source is Pepper's account of world hypotheses, turned against Ing. Pepper held that each world hypothesis is autonomous, that it should stand on its own evidence, and that eclecticism, the mixing of root metaphors, introduces contradiction. Ing accepts these maxims. Snowden argues that they apply to complexity as much as to anything else. Complexity is not a phase of systems thinking, to be absorbed with its vocabulary (resilience, propensity, entailment, turbulence) while its ontology is left behind. It is a world hypothesis in its own right, or close to one, and it should be treated as autonomous.

The function

Element 1 lets Snowden refuse the taxonomic move. If complexity were a perspective, Ing's map would be fair: the systems tradition has many perspectives, and complexity would be one. Because complexity is a claim, the map makes a category error. It puts a claim about the world in a list of ways of looking at the world. Element 1 also grounds the whole of Cynefin. Cynefin's domains are not lenses one can choose to apply. They are different kinds of situation, each needing different responses, and the dangerous error is to misjudge which kind of situation one is in.

The flag

Element 1 makes anthro-complexity realist about complexity. Some systems really are complex, whatever anyone thinks of them. Chapter 14 will ask what follows when this realism meets the reflexive claim of Element 14. That claim holds that human systems are partly constituted by what their members think of them, and partly changed by what observers say about them. A realism about the complexity of systems that are partly made of descriptions is not impossible, but it is not simple. Ian Hacking's distinction between natural kinds and "interactive kinds", whose members respond to being classified, will be the instrument for the inquiry.

Element 2: Coherence Is Found, Not Presumed

The claim

This is the most philosophically distinctive element of the ontology, and the one the November post develops most carefully. Systems thinking, in the synthetic mode Ing keeps, presumes wholes and infers parts in relation. Ackoff's method makes this explicit: identify the containing whole, explain its behaviour, then explain the part in terms of its role or function within it. Synthesis precedes analysis. The organicist tradition behind Ackoff makes the same presumption more deeply: there is a whole, and the parts are its organs.

Anthro-complexity, Snowden says, presumes neither wholes nor parts. Two corollaries carry the claim. First, decomposing things to understand them is not reductionism. The error of reductionism is not decomposition. It is the assumption that you can know what will emerge when the decomposed parts are recombined. Second, the whole is not there to be identified in advance. It is continually produced by interactions that it, in turn, modulates, and coherence has to be found in practice, at a level that may shift. Hence the method Snowden attributes to the Pachyderm Post: find the level of coherence before decomposing or recombining.

The difference from Ing

Snowden is careful to give Ing credit. Ing's paper explicitly rejects the organicist picture of a system of interest inside a containing whole, in favour of wholes alongside other wholes in a texture over time. Its conclusion accepts that defining a system and its boundary makes little sense in a dispersive world hypothesis. Snowden calls this a significant move away from Ackoff and towards something he would recognise. But the synthetic mode of reasoning remains: a strand, a texture, a family, a clan, a community are wholes taken as given and then woven together. The difference from Ing is therefore narrower than the difference from Ackoff, but it is still, Snowden says, an ontological difference and not just one of emphasis.

Unpacking the claim

The claim has a precise structure and should be stated carefully. There are three positions:

Position Wholes Parts Method
Analytic reductionism Derived from parts Presumed Decompose, then recompose by prediction
Synthetic systems thinking Presumed Inferred from role in whole Identify whole, then explain parts by function
Anthro-complexity Found in practice, at a shifting level Not presumed either Find coherence, then decompose or recombine without predicting emergence

The third position is not a midpoint between the other two. It rejects the shared assumption that one end or the other can be taken as given. What is given is interaction. Wholes and parts are both effects of interaction, and both are provisional. A team is a whole for some purposes and a part for others. It may cohere in a crisis and dissolve in routine. Its boundaries can be found only by watching what it does. And the watching is itself an interaction: the method of finding coherence must be a method of participation, not of survey.

The sources

This element has several sources, and they are not all compatible. One is complexity science's own account of emergence: macro-level patterns arise from micro-level interactions and then constrain those interactions in turn. Juarrero, whose work on constraint runs through the whole anthro-complexity series, gives the most developed account: context-sensitive constraints allow downward causation without any external force, because the pattern of interactions changes the probabilities of future interactions. A second source is Ingold, whom Snowden reads separately in the series. Ingold's meshwork is a tangle of lines of growth and movement, not a network of nodes. In a meshwork there are no presumed wholes and no presumed parts, only the paths along which things become what they are. A third is the actor-network tradition, from which the vocabulary of actants comes (Element 3). A fourth, less acknowledged, is the pragmatist insistence that the meaning of a concept lies in its practical consequences. A whole that cannot be found in practice has no meaning.

The function

Element 2 does three jobs. It separates anthro-complexity from systems thinking at the level of ontology, not merely of method. It grounds the practical move of finding coherence first, which underlies estuarine mapping's concern with the level at which constraints can be shifted and SenseMaker's search for patterns at whatever scale they appear. And it lets anthro-complexity accept decomposition without accepting reductionism: you can take things apart to understand them, provided you do not claim to know what will happen when they are put back together.

The flag

Element 2 refuses to presume wholes and parts. Chapter 19 will argue that when anthro-complexity turns to the human, in Element 12, it presumes parts after all: discrete persons who hold multiple identities, who act with intention, who create abstractions. The three marks of the human are predicates of individuals. If wholes are produced by interaction, so are the individuals who make them up. Mead and Elias, whom Chapters 27 and 28 take in turn, built their whole social theory on this point. A person is not a part presumed before relation. A person is a pattern of relation, found in practice at a level that shifts. Element 2, applied consistently, threatens Element 12.

Element 3: Actants, Not Variables

The claim

Ackoff defines a system's environment as the set of variables that can affect its state. Anthro-complexity works instead with actants, the term Latour uses in actor-network theory for anything, human or non-human, that makes a difference to a course of action. The change of vocabulary is a change of ontology. A variable is a dimension along which a system's state can be measured. An actant is something that acts. A variable is defined relative to a system whose state it describes. An actant is defined relative to a course of action it affects. A system described in variables has a state at each moment, a point in a state space. A field of actants has a direction: a set of tendencies arising from what each actant is doing to the others.

Snowden names two kinds of actant explicitly in the November post.

Constraints, following Juarrero, are one kind of actant. A constraint is not a barrier. It is anything that changes the probability of what follows. Juarrero distinguishes several kinds. Context-independent constraints narrow possibilities without regard to context: a wall, a rule, a deadline. Context-sensitive constraints make the probability of one event depend on another, and so allow patterns to form and persist. Snowden's own vocabulary, worked out with Juarrero, distinguishes governing constraints, which limit and channel, from enabling constraints, which make new things possible: a common language, a shared protocol, a trellis. Constraints have different strengths and different energy costs to change, and estuarine mapping works by locating them.

Constructors, following David Deutsch and Chiara Marletto's constructor theory, are a second kind. A constructor is an entity that brings about a transformation while keeping the capacity to bring it about again. A catalyst is a constructor: it enables a chemical reaction and comes out of the reaction unchanged, ready to do it again. Snowden's example from human affairs is an established ritual. A ritual brings about a transformation (a wedding marries, a graduation graduates, a handover hands over) and remains available to do so again.

The function

The vocabulary of actants does three things. It puts non-human things into the account of human systems: buildings, software, regulations, rituals and documents act, and their action has to be understood. This is the debt to the new materialisms that Snowden acknowledges: matter is not inert, non-human actants make a difference, and humans are entangled with the material world. It replaces the snapshot of a state with the vector of a field, which is the subject of Element 4. And it allows intervention to be directed at constraints and not at outcomes. You cannot change an outcome directly in a complex system, but you can change the constraints within which outcomes arise.

The qualification

Snowden adds a qualification at once, and it matters for everything that follows. To accept that matter acts is not to treat all action as equivalent. He rejects the "flatter readings of Latour", on which every actant counts equally, and Karen Barad's agential realism, on which no distinct relata exist before the relation. The river shapes the city and the planners argue about the river. Both are actants, but only one can change the story it tells about the other. Anthro-complexity's ontology of actants is therefore asymmetric. It is flat in admitting non-humans as actants, and steep in holding that human actants have capacities others lack.

The flag

Two flags. First, the qualification creates a tension Chapter 22 will examine. Latour introduced the term actant precisely to avoid deciding in advance which entities have which capacities. The point of the term is that the analyst does not presume a hierarchy of agents; the hierarchy, if there is one, has to be traced in the network. Taking Latour's term while presuming a hierarchy (humans narrate, rivers do not) keeps the vocabulary and changes the method. That may be justified, but it has to be argued. Second, the choice of constructor theory as a source of actants creates a sharper tension, examined in the same chapter. A constructor is defined by being unchanged by the transformation it brings about. The human actants of anthro-complexity are defined by being changed: by the stories they tell, the identities they move between, the abstractions they make. A ritual that left its participants unchanged would not be a ritual.

Element 4: Dispositional States and Vectors

The claim

With actants come dispositional states and vectors. What matters is the direction and speed of travel, and the modulating structures that make some trajectories more likely than others. Snowden puts it compactly: a state described in variables is a snapshot; a field of actants has a direction.

The idea of a dispositional state needs some unpacking. A disposition, in philosophy, is a property that shows itself only under certain conditions. Glass is fragile even when it is not breaking; salt is soluble even when it is dry. A complex system's dispositional state is its readiness to move in some directions more than others, given its current configuration of constraints and actants. It is not a prediction. A fragile glass may never break. But it is not nothing either. It is real, it can be read, and it bounds what can plausibly happen next.

The vector adds direction and speed. A system is not merely disposed. It is already moving. The question for a practitioner is not "what state is this system in?" but "which way is it going, how fast, and what is shaping that?" The modulating structures (constraints, attractors, the habits and rituals that persist) shape the vector. They are what a practitioner can work on.

The sources

The language of dispositions has a long philosophical history, from Aristotle's dunamis through the twentieth-century analytic debate about dispositional properties. Snowden's more immediate sources are Jullien's propensity, examined in Chapter 3, and the physics of dynamical systems, where a system's trajectory through state space is shaped by attractors and the landscape of possibilities. The move from state to vector is the move from statics to dynamics, from describing where a system is to describing where it is going.

The function

Element 4 is the bridge between the ontology and the practice. Estuarine mapping maps dispositional states: which constraints are holding the current pattern in place, which could be shifted, at what cost and on what timescale. SenseMaker reads vectors: the direction in which narratives are moving across a population, and how fast. The idea of nudging a system towards an adjacent possible depends on it. You cannot move a complex system wherever you like, but you can work with the direction it is already going, and shift the constraints that make one trajectory more likely than another.

The flag

The dispositional state of a human system includes the dispositions of the people in it. Those include their disposition to respond to being read. Chapter 26 will argue that a reading of a human system's vector is itself a force acting on the vector, and that the force is not small. Published patterns change the narratives they summarise. A vector read and fed back to the system is not the same vector afterwards. Jullien's propensity, Chapter 35 will argue, already contains this point: the commander's own disposition is part of the configuration he reads.

Element 5: Far From Equilibrium, With No Normal State

The claim

Complex human systems are far from equilibrium. There is no normal to return to, only a current dispositional state and the plausible directions it can move in. Snowden makes this point against Ing's eurhythmia, the programme's practical aim of a healthy system whose rhythm is neither too fast nor too slow, whose balance has neither too much nor too little of yin or yang, each relative to its contexture. Eurhythmia, Snowden grants, is a welcome replacement for Ackoff's progress towards ideals. But it puts in its place a state of health, a seasonable rhythm, against which disorder is identified. Balance is a useful metaphor for a body. It is a risky one for an organisation, where talk of restoring balance tends to favour whoever defines it.

The sources

The far-from-equilibrium claim comes from Prigogine's thermodynamics of dissipative structures. Systems held far from equilibrium by a flow of energy and matter can produce and maintain order: convection cells, chemical clocks, living organisms. That order is not a state of rest. It is a pattern sustained by continuous flow, and it collapses if the flow stops. An organisation, on this view, is a dissipative structure. Its order is not an equilibrium to which it returns when disturbed. It is a pattern kept up by continuous activity, and it can bifurcate into a different pattern if conditions cross a threshold.

The function

Element 5 does critical work. It lets Snowden reject the medical model of intervention, in which a practitioner diagnoses deviation from a norm and restores the norm, without rejecting intervention altogether. There is no norm to restore, but there is a current state and a range of plausible next states, and some of those are better than others by the lights of the people in the system. Element 5 also carries a political charge, which Snowden makes explicit: any norm of health in a human system is defined by someone, and the definition favours the definer.

The flag

The political charge of Element 5 is the most important thing in this chapter for the argument of the book. Snowden sees, clearly and in so many words, that a norm against which a human system is read favours whoever defines it. Chapter 23 will ask why this insight does not reach anthro-complexity's own instruments. A signifier set is a norm of salience: it defines what is worth noticing. An estuarine map is a norm of energetics: it defines what is costly to change. A Cynefin categorisation is a norm of kind: it defines what sort of situation this is. Each is defined by someone, and each, on Snowden's own argument, will tend to favour its definer. The insight is right, and it applies more widely than he applies it.

Element 6: The Line That Matters Is Between Ordered and Complex

The claim

In the Barcelona lecture Ing captions his surfing slide: mechanisms lead to causality, living systems to propensity. Snowden's first response is that propensity belongs to complex systems whether or not they are alive, and that the line that matters runs between ordered and complex, not between non-living and living. Ing's paper partly answers this. Through Keekok Lee's reading of Chinese thought, wanwu, the ten thousand things, covers the abiotic as well as the biotic: mountains also come into being and wear away. Snowden accepts that the paper's emphasis on living systems is a choice of focus, not a claim that only the living have propensity.

The function

Element 6 is small in the November post but carries real weight in the architecture. It is the basis of Cynefin itself. The fundamental distinction in Cynefin is between ordered systems (clear and complicated), where cause and effect can be known in advance by perception or analysis, and unordered systems (complex and chaotic), where they cannot. The distinction is not biological. A weather system is complex and not alive. A bureaucratic procedure is ordered and carried out by living people. Element 6 keeps Cynefin's distinction from collapsing into a distinction between the inanimate and the animate, which would make complexity a property of life and order a property of mechanism. Complexity science shows that order and complexity cut across that line.

The flag

Element 6 has a consequence that will matter in Book Three. If the line that matters is between ordered and complex, then the line between human and non-human complexity is a second line, drawn inside the complex domain. Anthro-complexity therefore needs two distinctions: ordered/complex, which it shares with complexity science, and human/non-human within the complex, which it claims for itself. The first is grounded in the natural sciences, the second in claims about human capacities. Chapter 13 will ask what grounds the second, and whether the grounds Snowden gives (multiple identities, intention, abstraction) are the right ones. It will suggest that the deeper ground is the one Snowden names as a consequence and not as a cause: human systems are reflexive, and their descriptions become part of them.

The Ontology as a Whole

Gathered together, the six ontological elements form a coherent picture:

Element Claim Principal sources Principal function
1 Complexity is a claim about the world, not a perspective Natural-science origins; Pepper's maxims Refuses the taxonomic move; grounds Cynefin's realism
2 Coherence is found, not presumed; neither wholes nor parts Juarrero; Ingold; emergence; pragmatism Separates from synthetic systems thinking; grounds finding-before-decomposing
3 Actants (constraints, constructors), not variables Latour; Juarrero; Deutsch & Marletto Admits the non-human; directs intervention at constraints
4 Dispositional states and vectors, not snapshots Dispositions; dynamics; Jullien Bridges to practice; grounds mapping and nudging
5 Far from equilibrium; no normal to return to Prigogine Rejects the medical model; exposes the politics of norms
6 The line that matters is ordered/complex Cynefin; complexity science Keeps complexity from being reduced to life

The picture is impressive. It is more rigorous than most of what passes for complexity thinking in organisational life. It answers, in advance, many of the objections that complex responsive process theorists have made to systems thinking. It is realist without being reductionist, dynamic without being formless, and materialist without being flat.

It also contains, already at the level of ontology, three tensions that Book Four will develop into theses:

  • Element 2 against Element 12. If neither wholes nor parts are presumed, then persons, the bearers of the human marks, are not presumed either. Persons are found in practice as patterns of interaction. The human cluster, which presumes them, will have to be re-founded.
  • Element 3 against itself. The vocabulary of actants was built to avoid presuming a hierarchy of agency. Anthro-complexity's qualification restores a hierarchy, and constructor theory supplies a model of action in which the agent is unchanged. Both pull against the human cluster.
  • Element 5 against Elements 4 and 7–10. The insight that a norm favours its definer is applied to eurhythmia. Chapter 23 will argue that it applies equally to the definition of dispositions, vectors and constraints in anthro-complexity's own maps.

None of these tensions is a contradiction in the strict sense. Each is a place where the ontology, followed consistently, leads further than the theory goes. The work of the disputation is to follow it there.

NEITHER WHOLES NOR PARTS PRESUMED —

THEN NEITHER PERSONS —

THEN THE HUMAN MUST BE FOUND, NOT GIVEN.

Addition: Archer's Objection, and the Limits of Relationism

The disputation has argued throughout for a relational account of the human, drawing on Mead, Elias, Giddens and others. Fairness requires that the strongest objection to relationism be heard, and it comes from a tradition the disputation has not yet engaged: the critical realism of Roy Bhaskar and, in social theory, Margaret Archer. This addition sets out Archer's objection and asks what it means for the disputation's argument. Its conclusion is that the objection is serious and partly successful, and that it supports one element of anthro-complexity's position that the disputation has criticised: Snowden's insistence that the people remain distinct.

Archer's objection: central conflation

Margaret Archer, in Realist Social Theory: The Morphogenetic Approach (1995) and Being Human: The Problem of Agency (2000), criticised what she called conflation in social theory. There are three kinds.

  • Downward conflation reduces agents to effects of structure. Agents are bearers of roles, positions and norms, with no independent causal power. Structural functionalism and some structuralisms are guilty of it.
  • Upward conflation reduces structure to effects of agents. Structure is the aggregate or product of individual actions, with no independent causal power. Methodological individualism and rational choice are guilty of it.
  • Central conflation fuses structure and agency so tightly that neither can be distinguished from the other, and so neither can be shown to affect the other. Archer's principal target was Giddens's structuration theory, and specifically the duality of structure. If structure is both the medium and the outcome of action, instantiated only in action, then structure and agency are two sides of one coin and cannot be analysed separately. But then it becomes impossible to ask how structure constrains agency, or how agency changes structure over time, because the two are never apart.

Archer's alternative, the morphogenetic approach, insisted on analytical dualism. Structure and agency are distinct strata of social reality, with different properties and causal powers, and they interact over time. Structure pre-exists any given action, conditions it, is reproduced or transformed by it, and the transformed structure conditions later action. The cycle is structural conditioning, then social interaction, then structural elaboration. It unfolds in time, and structure and agency are distinguishable at each phase because they are temporally separated.

Archer also defended the person against relational dissolution. In Being Human she argued that human beings have powers that are not reducible to their social relations: a sense of self that emerges from embodied practice in the natural and practical orders before and alongside the social; the capacity for internal conversation, reflexive deliberation about one's concerns; and the ability to set priorities among concerns and so define a personal identity. These powers are formed in relation but are not merely relational. Once formed, they are properties of persons, which persons bring to their relations and which can resist and transform those relations.

The objection applied to this disputation

Archer's objection bears on this disputation directly. The disputation's relational account of the human, persons as patterns of relation and distinct as produced (Principle 1), draws on Mead, Elias and Giddens. Archer would ask whether it falls into central conflation. If persons are only patterns of relation, how can they resist the relations that formed them? How can the outsider in Winston Parva refuse the stigma the established impose, if her self is the internalised attitude of the established? How can the narrator in a SenseMaker study tell a story that breaks the signifier set's categories, if her identity is produced by the comparative context the instrument supplies?

The disputation has answers, from its own sources. Mead's I is irreducibly individual, the source of novel response. Elias's open people are intensely real as individuals. Juarrero's persisting pattern is real and distinct. But Archer's objection shows that the answers need to be stated more firmly than the disputation's emphasis on relation has sometimes allowed. Persons are formed in relation, and once formed they have powers of their own: reflexivity, concern and the capacity to resist. These are not dissolved back into relation, and the relational account must not make them so.

What Archer concedes to Snowden

Archer's objection also bears on the disputation's criticism of anthro-complexity. Thesis I criticised the November post's claim that the people remain distinct, even when what they know together does not as an expression of the closed person. Archer would defend the claim. The people do remain distinct, as persons with emergent powers of their own, even when their knowledge is entangled. Snowden's rejection of Barad, of the dissolution of relata into intra-action, is on Archer's account exactly right. Persons are not dissolved into relations. They are real, distinct and causally powerful.

The disputation must therefore qualify Thesis I. The possessive grammar presumes persons as parts before relation, which Archer would also reject, since persons emerge from embodied practice and social relation and are not given at birth. But the possessive grammar's insistence that persons are distinct, with capacities of their own, is defensible on Archer's account, and the disputation's relationism should not deny it. The right position is Archer's. Persons are emergent, formed in relation, and they have irreducible powers once formed. Neither the possessive grammar (persons given before relation) nor a strong relationism (persons nothing but relation) is correct.

The position restated

Position Persons before relation? Persons reducible to relation? Example
Possessive individualism Yes No Macpherson's proprietor; the closed person
Strong relationism No Yes Barad's intra-action; some readings of Gergen
Emergent personhood No: formed in relation No: irreducible powers once formed Archer; Mead's I; Elias's open people; Juarrero's persisting pattern

The disputation's Principle 1, the human is a pattern of relation, distinct as produced, is meant to state the third position. Archer's objection shows that "pattern of relation" can be heard as the second. The principle is therefore amended, under the Covenant.

Principle 1, amended. The human is formed in relation and distinct as formed: an emergent person with irreducible powers (reflexivity, concern, the capacity to resist and transform the relations that formed her) which she brings to every relation and which no relation exhausts. Neither presumed before relation, as in the possessive grammar, nor dissolved into it, as in strong relationism.

What this changes for the instruments

The amendment changes little for the instruments proposed in Book Eight, and it clarifies one thing. The instruments should sense relations: address, joint signification, perspectives and the established and outsiders. They should also honour persons' irreducible powers, especially the capacity to resist the instrument's own framing. The fourth vertex, the narrator's reading of the pattern and the right to contest are ways of honouring that capacity. They let the person's I, in Mead's sense, or her internal conversation, in Archer's, break through the me the instrument offers. An instrument that sensed only relations, treating narrators as nothing but nodes in a figuration, would commit central conflation. One that honoured persons' capacity to resist would not.

FORMED IN RELATION — DISTINCT AS FORMED —

NEITHER PRESUMED BEFORE NOR DISSOLVED AFTER —

THE PEOPLE DO REMAIN DISTINCT, AS EMERGENT PERSONS WHO CAN SAY NO.

Addition: The Six Ontological Elements Taken Slowly

The chapter above states the six ontological elements at the speed of an argument. A reader meeting them for the first time may find that speed too fast. This addition goes back over each one at walking pace. For every element it gives five things in the same order: the claim in plain words, a worked example, the difference from complex adaptive systems theory, the commonest misreading, and a first interrogation with a provisional verdict. The verdicts here are provisional because the full case is made in Book Four. They are given so that the reader knows, while learning what each element means, which parts of it the disputation will later accept and which it will press.

A case to think with

One invented case runs through all four of these slow readings, in this chapter and the three that follow. It is not a real organisation and no real organisation is intended.

Calder Vale Homes is a housing association with about fourteen thousand homes across three towns. For two winters it has had a damp and mould problem it cannot get on top of. Complaints rise each October. The repairs contractor closes jobs quickly, and many reopen within weeks. Surveyors blame tenants' heating habits. Tenants say they cannot afford to heat the rooms and that nobody listens. Housing officers know which blocks are worst but are not asked. The board has a dashboard that shows repair jobs closed within target at ninety-one per cent, and a regulator's letter on the table that shows something else. A new director of housing, Asha, has been told to fix it by spring.

Each element below is asked the same question: what would it mean for Asha?

Element 1 slowly: complexity is a claim about the world

In plain words. Some people use "complex" to mean "I find this hard to understand". On that usage, complexity is a fact about the observer, and it would go away if she were cleverer or had more data. Anthro-complexity denies this. It says that some systems really are of a different kind. In them, the parts change one another as they interact, so there is no stable recipe linking a cause to an effect. That is true of the system, whoever looks at it and however much they know.

The example. Asha's first instinct is that the mould problem is complicated: many factors, poorly measured. On that reading she needs a stock condition survey, a root-cause analysis and a programme. Anthro-complexity asks her to consider a different possibility. Perhaps the pattern is produced by interactions: tenants underheat because of bills, underheating makes condensation, the contractor treats condensation as a wipe-and-paint job because that is what the contract pays for, the job reopens, the tenant stops reporting, the officer stops visiting a flat where she has nothing to offer, and the dashboard improves because unreported mould is not a failed repair. If so, no survey will find the cause, because there is not one. More data will not turn this into a problem with a recipe.

How CAS differs. Here it mostly does not. The Santa Fe tradition also treats complexity as a real property of systems with many interacting adaptive agents. The difference is in what follows. CAS theory expresses the real complexity in a model, usually a simulation with agents and rules. Anthro-complexity holds that for human systems no such model can be relied on, for reasons that Element 12 supplies.

The common misreading. "Everything is complex, so analysis is pointless." Snowden says nearly the opposite. Much of what organisations do is ordered, and there analysis and good practice are exactly right. Whether a given boiler is faulty is an ordered question with an answer. The claim is that complexity is real where it occurs. It does not say that complexity is everywhere.

First interrogation. The claim is ontological: it concerns what the system is. But the judgement that this situation is complex is made by someone. Asha, or her consultant, decides that the mould problem is a complex one. That decision is consequential. It moves authority from surveyors with instruments to facilitators with workshops, and it changes what counts as an acceptable answer to the board. So even if complexity is a fact about the world, the attribution of complexity is an act by a positioned person, and the theory has little to say about that act. Holds, with a gap The ontology is accepted. The gap is the missing account of who classifies.

Element 2 slowly: coherence is found, not presumed

In plain words. Systems thinking tends to begin by naming a whole: "the housing system", "the repairs system". It draws a boundary, lists the parts, and studies how they serve the whole. Anthro-complexity distrusts that first step. It says you may not assume there is a whole with a purpose. You should look for whatever coherence is actually present, at whatever grain it appears, and you should expect it to be patchy.

The example. A systems consultant draws Asha a diagram of "the damp and mould system" with inputs, processes, outputs and feedback loops. It is tidy, and everyone in the room recognises it. The anthro-complexity practitioner asks different questions. Where do things actually hang together? It turns out that one estate has almost no reopened jobs, because a caretaker there calls the same two operatives directly and they bring a hygrometer. That is a small real coherence. It is not on the diagram, because the diagram was drawn from the idea of the system and not from what is there. Meanwhile the diagram's "feedback loop from tenant satisfaction to contract management" exists on paper and nowhere else.

How CAS differs. CAS theory is closer to anthro-complexity here than systems thinking is, since it too builds from agents upward. But a CAS model must specify its agents, its rules and its environment before it runs, and that specification is a presumed whole of another kind. Anthro-complexity claims to presume neither the whole nor the model.

The common misreading. "There are no wholes." The claim is about the order of inquiry. Coherence comes at the end of looking. It is not available as a starting assumption. Some wholes are real. They have to be found.

First interrogation. This is the element from which Thesis I takes its form of words. Anthro-complexity refuses to presume the whole. Does it presume the parts? When it comes to the human, it speaks of persons who hold identities, act with intention and create abstractions. Those persons arrive in the theory ready-made. But the tenants and officers of Calder Vale are who they are partly because of the association. The tenant who "does not report" learned not to. The officer's sense of what is worth raising was formed by years of what happened when she raised things. If coherence must be found and not presumed, the same discipline should apply to the person. Holds, and cuts both ways

Element 3 slowly: actants, not variables

In plain words. A variable is a quantity that can be measured and related to other quantities. An actant is anything that makes a difference to what happens: a person, a rule, a form, a building, a rumour, a spreadsheet column. Anthro-complexity asks you to map actants because in a complex system the things that matter are often not quantities and often not people.

The example. Asha's analysts have variables: average days to complete, reopen rate, satisfaction score. The actant view asks what is doing something at Calder Vale. One answer is the job code. The contractor's system has a code for "mould wash" and no code for "investigate cause", so every visit becomes a mould wash. The code is an actant. So is the ninety-one per cent tile on the board dashboard, which makes directors relax. So is the prepayment meter. So is the story, told on every estate, about the family who complained and got a letter about rent arrears the following week. None of these is a variable. Each makes things happen.

How CAS differs. CAS models have agents, which are usually many instances of a few types, each with a rule set. An actant list is more varied and is not restricted to things that decide. A job code does not adapt, and it is not an agent in the CAS sense, yet it shapes everything.

The common misreading. "So people and job codes are the same kind of thing." Snowden takes the word from Latour and refuses that conclusion. He holds that humans differ from other actants in one decisive way, which is Element 13. Whether one can take Latour's word and leave his flatness behind is discussed in the Latour card and in Chapter 22.

First interrogation. An actant map is drawn by someone, and what gets on to it depends on who is in the room. A workshop of managers at Calder Vale will list the contractor, the tenants' behaviour, the weather and the regulator. They are much less likely to list their own dashboard, or the workshop itself. Yet the workshop is an actant from the moment it is announced. Operatives hear that "they're doing a review" and close jobs more carefully for a month. Holds, incomplete The map should include the mapper. Chapter 21 develops this.

Element 4 slowly: dispositional states and vectors

In plain words. You cannot say what a complex system will do. You can say what it is disposed to do: which way it leans, what it would do easily and what it would do only under great pressure. And you can describe change by direction and speed, a vector, without naming a destination. "More of this, less of that, starting from here" replaces "we will be at point X by March".

The example. A conventional plan for Asha reads: reduce reopened damp jobs to under five per cent by the end of the fourth quarter. A dispositional reading says: at present the system is disposed to close jobs fast and to discourage reporting. It leans that way because the contract, the dashboard and the arrears letters all push the same direction. A vector statement would be: more visits where the operative measures humidity before doing anything, fewer visits coded as mould wash without investigation, and we will watch whether reports from the worst blocks go up, because that would mean people have started to believe reporting is worthwhile.

How CAS differs. CAS theory describes systems by attractors and fitness landscapes, which are cousins of dispositions. But these are properties of a model's state space, computed or simulated. Anthro-complexity's dispositions are read from the field, largely through people's own accounts, and it denies that the landscape can be computed.

The common misreading. "Vectors mean no targets and so no accountability." A vector can be monitored and a team can be held to it. What is given up is the claim to know the end state in advance.

First interrogation. Disposition is the element that carries the estuary image, and Thesis III presses it hard. A river's disposition does not change when a hydrologist measures it. Calder Vale's does. Once staff know that "direction of travel" is what Asha watches, they learn to produce direction. And whose disposition is it? "The system is disposed to discourage reporting" joins together a tenant's fear, an officer's fatigue and a director's incentive. They are different things held by people in very different positions, and they will not move for the same reasons. Holds for the reading, fails for the reader

Element 5 slowly: far from equilibrium, with no normal state

In plain words. Many management ideas assume that an organisation has a healthy normal condition to which it returns after a disturbance. Anthro-complexity says that living and social systems are not like that. They persist only by constant flow and constant change. There is no rest state. What looks like stability is a pattern being continually remade.

The example. The board's language about Calder Vale is medical. The repairs service is "unwell" and needs to be "restored". Asha is asked for a "recovery plan". All of this assumes there was a healthy state to which one can return. Looking back, she finds there never was one. Five years ago mould complaints were lower because reporting was harder, the stock was a little younger and energy was cheaper. The past was not a state of health. It was a different set of pressures. There is nothing to restore. There is only the question of which direction to move from here.

How CAS differs. CAS theory agrees in principle: complex adaptive systems operate far from equilibrium and never settle. In practice many CAS models are analysed by their equilibria and attractors, because that is what the mathematics can characterise. Anthro-complexity is stricter in refusing the language of balance. That is the basis of Snowden's objection to Ing's eurhythmia.

The common misreading. "Nothing is stable, so nothing can be relied on." Patterns can last for decades. The claim is that they last by being reproduced, and that reproduction can fail.

First interrogation. The refusal of a normal state has a political edge that Snowden sees clearly in others: whoever defines the balance benefits from it. The same question can be put to the alternative. If there is no normal state, then "better" must be decided some other way, and anthro-complexity's usual answer is "more stories like these, fewer like those". Someone picks which stories. At Calder Vale that will be Asha and the board, unless something is designed to make it otherwise. Holds The element stands. Its critique of norms should be applied to its own selection of desirable narratives, which is the business of Thesis V.

Element 6 slowly: the line that matters is between ordered and complex

In plain words. The most important sorting question in the whole approach is whether a situation is ordered or complex. In an ordered situation, cause and effect are stable, so you can analyse and then act. In a complex one they are not, so you must act in small ways first and see what happens. Treating a complex situation as ordered produces confident plans that fail. Treating an ordered one as complex produces endless workshops about something an engineer could fix. Cynefin, set out in Chapter 6, is the device for making this distinction and for noticing when a situation is moving across it.

The example. Calder Vale's mould problem is not one situation. Whether a specific wall has rising damp is an ordered question, and a competent surveyor can settle it. How to treat a diagnosed cold bridge is complicated but knowable. Why reporting has collapsed on the Fieldhead estate is complex. And the week after a child is hospitalised and the local paper runs the story is chaotic: act first, stabilise, and make sense afterwards. Asha's most useful early act may be to separate these, so that surveyors are not asked to workshop and tenants are not handed a technical specification.

How CAS differs. CAS theory has no equivalent sorting device for the practitioner. It studies complex adaptive systems, and it does not offer managers a way to decide which of their problems are of that kind. This is one of anthro-complexity's real additions.

The common misreading. "Cynefin is a two-by-two for categorising problems." Snowden insists that it is a sense-making framework, in which the data come before the framework, and not a categorisation framework, in which the framework comes first. The domains are supposed to emerge from how people in the situation actually describe it.

First interrogation. Suppose it is done properly, and the domains emerge from the participants' own material. It is still the case that sorting has effects. To place "why reporting collapsed" in the complex domain is to say that nobody is to blame in the ordinary sense, since outcomes emerged from interactions. That may be true. It is also extremely convenient for anyone who signed the contract and designed the dashboard. A tenant on Fieldhead might sort it differently: as a simple case of a landlord not doing what the law requires. The line between ordered and complex is where knowledge claims and responsibility claims meet, and the theory treats it mainly as a knowledge question. Holds as epistemics, thin as politics Chapters 20 and 23 return to this.

What the six come to

Read slowly, the six ontological elements give Asha a real change of stance. She should stop looking for the cause, stop assuming there is a system with a purpose, list what is actually making a difference, describe leanings and directions instead of end states, give up the idea of recovery to normal, and sort her problems by kind before choosing methods. Each of those is good advice, and none of it is available from the dashboard.

The interrogations have a common pattern too. In each case the element is a sound description of the system as read. In each case the gap concerns the reading: who classifies, who is presumed, who maps, whose disposition, who selects, who sorts. The six ontological elements describe a world well and leave out the describer. The remaining three slow readings ask whether the causal, human and practical elements put her back.

SIX ELEMENTS, ONE STANCE —

LOOK BEFORE YOU PRESUME —

AND ASK WHO IS DOING THE LOOKING

Addition: Cross-Lights From Book Nine

Three later chapters bear on the ontological elements.

On Element 1. The chapter relies on the view that complex systems cannot be compressed into a model. Chapter 56 challenges that view. Its claim is that incompressibility is a notion of the outside observer, and that in human systems compression is done from within and alters what is compressed. Snowden's sentence that all diagnosis is intervention is the hinge.

On Element 2. Snowden's remark against Luhmann, that the boundaries which coupling presupposes are not there to begin with, is the same refusal of presumed wholes, stated more strongly.

On Element 6. The Zen Garden study of Chapter 53 asks what is done when a situation is placed on a figure. Its answer is that sorting routes a problem to someone, and that the sorter belongs on the figure. It also observes that Cynefin has an axis of knowability and no axis of agreement.

Chapter 3

Propensity, Retrospect, and the Turn of the Tide

7,187 words, about 31 minutes of reading

Causality as the Hidden Ontology of Management

Every theory of management carries a theory of causality, and most carry it without noticing. A plan presumes that actions taken now will cause outcomes later. A performance target presumes that effort causes results. A change programme presumes that interventions cause change. A risk register presumes that hazards cause losses with estimable probability. In each case there is a causal arrow from what the manager does to what the organisation becomes, and the arrow points forward. Management is a forward-causal discipline. Its characteristic question is: if I do this, what will happen?

Anthro-complexity's second cluster of elements is a sustained attack on that question. It does not say the question is meaningless. In ordered systems it has an answer, and Cynefin's clear and complicated domains are where that answer can be found. What it says is that in complex systems the forward-causal question cannot be answered. The attempt to answer it distorts both perception and action. Something else has to take its place, and the November post names four things that do: propensity, retrospective entailment, kairos, and practical wisdom. This chapter takes them in turn.

Element 7: Propensity Over Causality

The claim

The paper Snowden discusses uses François Jullien explicitly, along with John Law and Wen-yuan Lin's work on a shi-inflected social science, to put propensity at the centre: a disposition arising from the arrangement of things, not a cause. Snowden endorses it fully. Ing's lecture contrasts water-skiing and surfing. A water-skier's motion is pulled by a boat; that is causality, a force applied from outside along a line. A surfer's motion is read from the wave; that is propensity, a potential arising from the configuration of water, wind and seabed, which the surfer can join but not command. Snowden says he has long made the same point with sailing: you work with the dispositional state of a complex system and move with its propensity; you cannot engineer it causally.

Jullien's 勢

Jullien's A Treatise on Efficacy and The Propensity of Things are among the most influential Western readings of Chinese strategic thought, and their influence on Snowden is long-standing. Jullien's central contrast is between two models of efficacy. In the model he attributes to the Greek and European tradition, the agent forms an ideal end, devises means to reach it, and imposes the means on the situation. Plan, then execute: the model of Clausewitz's friction-ridden war, of the engineer's design, of the manager's strategy. In the model he finds in Sunzi, the Laozi and the Chinese strategists, the agent does not impose a plan. He assesses the potential inherent in the situation, its 勢 (shì), and lets the effect come from that potential, intervening, if at all, upstream, where small moves have large consequences, before the configuration has hardened.

勢 is hard to translate. Jullien renders it as "propensity", "potential" or "disposition". In military writing it is the configuration of a battlefield that makes some outcomes almost inevitable before any blow is struck: a boulder poised at the top of a slope, water dammed behind a sluice. In calligraphy it is the dynamism of a stroke, the force that runs through the brush. In politics it is the positional advantage that makes the ruler's will effective without force. Across all these uses, 勢 is a potential belonging to the arrangement and not to any one element in it, and the wise agent works by joining it, not by opposing or bypassing it.

Law and Lin, in their work on what they call a shi-inflected social science, propose that Western social science could learn from this. They suggest replacing the search for causes with attention to propensities, and replacing representations of social structure with sensitivity to the "dispositions of things" in a given situation. Their work matters because it carries Jullien's reading from strategy into method: it asks what kind of inquiry would be adequate to a world of propensities.

The function

Element 7 is the causal counterpart of Element 4. If complex systems have dispositional states and vectors, the relevant causal concept is not efficient causation, a force producing an effect along a line, but propensity, a potential arising from configuration. This changes the question a practitioner asks. Not "what will happen if I do this?" but "what is this situation disposed to do, and where can I join it?" It changes the kind of intervention a practitioner makes. Not pushing towards a goal, but shifting a constraint so that the system's own propensity carries it somewhere better. It changes the practitioner's posture from author to participant: someone who reads the wave and catches it.

The flag

Here the first flag in this chapter is planted, and it is the deepest. Jullien's 勢 is not a property of the situation viewed from outside. In Sunzi the configuration whose potential the general reads includes the general. His own position, reputation and visible dispositions are part of the configuration. The enemy reads him as he reads the enemy, and his reading of the situation is itself a move in it. The surfer, in Ing's image, reads the wave. But the wave does not read the surfer. In human affairs it does. Chapter 35 will argue that a propensity read from outside is not 勢 at all but a landscape, a terrain to be traversed, and that the distinctive feature of 勢 in its original strategic setting is that the strategist cannot step out of it. The surfing image, and the sailing image Snowden prefers, both keep the reader outside the medium. In that respect they are closer to the Western model Jullien contrasts with 勢 than to 勢 itself.

Element 8: Coherence Only in Retrospect

The claim

Ing's paper gives (con)texturalism-dyadicism a theory of truth based on Robert Rosen's concept of entailment. You can trace an outcome back through what entailed it, but forward-looking causality is not available. The example: a child entails parents, but a marriage does not cause a child. Snowden says this is very close to what Cynefin has always said about the complex domain, that cause and effect are coherent only in retrospect and cannot be used for prediction. The two positions arrive at the same place from different directions.

Rosen's entailment

Robert Rosen was a theoretical biologist whose Life Itself (1991) and Anticipatory Systems (1985) argued that living systems cannot be captured by the mechanistic models of physics. They are "closed to efficient causation": every efficient cause within an organism is produced within the organism. Rosen built a relational biology on the idea of entailment, the logical and causal relations by which one thing implies or produces another. His distinctive point, which Ing takes up, is that entailment is asymmetrical in time. Given an outcome, we can trace the entailments that produced it. Given the antecedents, we cannot generally derive the outcome, because a complex system has too many possible paths, and which one is taken depends on contingencies that cannot be known in advance.

The child-and-marriage example makes the point plainly. Every child has parents. That is an entailment: the existence of a child implies the existence of parents. But a marriage does not cause a child. Many marriages have no children, many children are born outside marriage, and the path from a marriage to a child, when there is one, runs through countless contingencies. Looking back from the child, the parents are certain. Looking forward from the marriage, the child is not even probable in any useful sense.

Retrospective coherence in Cynefin

Cynefin's version of the claim is older and more practical. In the complex domain, Snowden has written for over two decades, cause and effect can be perceived only in retrospect. After the event, a coherent story can be told about why things happened as they did. Before the event, the same story could not have been told, because the path that turned out to matter was one of many, and nothing marked it out. The practical consequence is the decision model for the complex domain: probe, sense, respond. You do not analyse and then act, because analysis cannot reveal what will emerge. You make small, safe-to-fail probes, watch what happens, and amplify what works or dampen what does not.

Retrospective coherence carries a warning. The story told afterwards will be coherent and will seem to explain what happened. That coherence will tempt people to believe the same story could have been told in advance, and that it can be used to predict the next event. This is the error Snowden calls retrospective coherence mistaken for predictability. It is the basis of most case-study teaching, most "best practice" and most consulting. A successful organisation's success is explained in retrospect, and the explanation is sold as a recipe.

The function

Element 8 is the epistemic counterpart of Element 7. If complex systems work by propensity, not efficient causation, then the forward-causal question cannot be answered. The available kind of knowledge is retrospective: knowledge of how things came to be, which can inform but not determine judgement about what to do next. Element 8 grounds the probe-sense-respond model. It grounds anthro-complexity's suspicion of recipes, frameworks applied as templates, and the transfer of practice from one context to another. It also grounds the narrative turn in SenseMaker. Narratives are how humans make retrospective sense, so collecting them gives access to the coherence a population has found in its own past.

The flag

Element 8 says coherence in complex systems is visible only in retrospect. Whose retrospect? A narrative told by a participant about why something happened is a retrospective coherence, and so is an analyst's pattern drawn from a thousand such narratives. They are not the same coherence, and they may conflict. Chapter 32 will argue, through Peirce, that coherence is always coherence for an interpreter. The question of who interprets, the narrator or the analyst, is therefore not a technical detail of SenseMaker's design. It is the question of whose retrospect counts. And because human systems are reflexive, a retrospective coherence that is published becomes a prospective constraint. The story an organisation tells about why it succeeded shapes what it does next. Retrospect, in human systems, does not stay in the past.

Element 9: Propensity Belongs to Complex Systems, Living or Not

The claim

This element was stated in Chapter 2 as Element 6 but has a causal side worth drawing out. The lecture's caption, mechanisms lead to causality, living systems to propensity, aligns two distinctions: mechanism/life and causality/propensity. Snowden separates them. Propensity belongs to complex systems whether or not they are alive. A weather system has a propensity (a storm forming, a front moving) though nothing in it is alive. A river has a propensity to meander, a slope to slide, a market to bubble. Conversely, an ordered system made of living things, such as a well-drilled army on parade or a clinical protocol, is better described by efficient causation than by propensity.

The function

Element 9 protects anthro-complexity against biologism, the view that the special features of complex systems come from life itself. That matters because the human, for anthro-complexity, is distinctive within the complex domain, not because humans are alive. If propensity came with life, humans would share it with every other organism and their distinctiveness would have to be found somewhere else. By making propensity a feature of complexity as such, Element 9 leaves the question of human distinctiveness open, to be answered by the human cluster.

The flag

If propensity belongs to complex systems as such, and humans are distinctive within the complex domain, the human difference must be a difference in how human systems have propensity, not in whether they have it. Chapter 13 will propose that the difference is this: human systems have propensities that include their members' readings of the system's propensity. A river's tendency to meander does not depend on what anyone thinks about it. A market's tendency to bubble depends partly on what its participants think it will do. The propensity of a human system is reflexive. That reflexivity, and not multiple identity, intention or abstraction taken separately, is what makes human propensity different in kind.

Element 10: Kairos, Not Chronos

The claim

Ing's paper prefers kairos over chronos: qualitative, opportune time over clock time. Snowden says this sits well with estuarine mapping and gives one of the richest images in the post:

In paraphrase: kairos is the turn of the tide, the moment the flow reverses, so that what was impossible an hour before becomes easy, or the reverse.

An estuary is governed by that turn. Much of estuarine mapping, Snowden continues, is about knowing which constraints can be shifted now and which must wait for the tide, given the different energies and timescales each requires. Any sailor, Snowden adds, knows better than to fight the tide: one waits until it turns, and only then makes one's move.

Kairos and chronos

The Greek distinction is ancient. Chronos is measured, sequential time: the time of clocks and calendars, in which each moment is like every other and differs only by position. Kairos is qualitative time, the right moment or opportune season, in which moments differ in what they make possible. In rhetoric, kairos was the fitness of a speech to its occasion. In medicine, it was the critical moment when a disease could be turned. In the New Testament, it is the appointed time, the fullness of time. In each use, kairos is not a point on a line. It is a quality of a situation, a ripeness, which may come or go and must be recognised when it comes.

Kairos in Snowden's practice

Snowden's use of kairos ties it to the energetics of constraint. Constraints differ in how much energy it takes to change them and on what timescale they can change. Some can be shifted at once and cheaply. Some need sustained effort over years. Some cannot be changed at all by the people in the system. Estuarine mapping places constraints on these two dimensions, energy and time, and so identifies which shifts are possible now and which must wait. The tide image adds a further dimension. Energy and time are not fixed properties of constraints. They vary with the state of the whole system. A constraint that would take enormous effort to shift at one moment may give way easily at another, because the flow of the whole has turned. The skilled practitioner does not just know what each constraint costs. She knows when the cost will drop.

Lewin, and the time of change

The kairos element connects to the first item of shared ground in the post: Lewin never wrote "refreeze". Snowden has long attacked the three-step model of change management (unfreeze, change, refreeze) on the grounds that it assumes a system can be held stable while you redesign it. He notes that Ing's paper makes the same point from a temporal angle, suggesting the model reflects a twentieth-century pacing. The three-step model is a chronos model of change. Change happens in sequence, between periods of stability, according to a schedule the change manager controls. A kairos model of change has no periods of stability to schedule. It has a continuous flow whose direction and strength vary, and moments when the flow makes a shift easy. The change agent does not schedule the change. She waits for the moment and moves.

The function

Element 10 completes the causal cluster with a theory of timing. If propensity has replaced causality and retrospect has replaced prediction, the practitioner's remaining question is when. Kairos answers it: when the situation is ripe, when the tide turns, when the constraint that was costly becomes cheap. The answer is not a rule. It cannot be specified in advance, because ripeness is a quality of a situation that can be recognised only in it. That leads directly to Element 11.

The flag

The tide turns by the moon. It does not turn because someone was seen waiting for it. In human affairs, the moment when a constraint becomes easy to shift often arrives because someone with standing was seen to wait for it, or to stop waiting. A chief executive who holds back from a reorganisation for six months is not just reading the tide. The holding back is a gesture that everyone in the organisation reads, and their readings change the flow. Mead's insight, developed in Chapter 27, applies: withholding is also a gesture, and its meaning is completed by the responses it calls out. Chapter 21 will argue that the tide image, beautiful as it is, misdescribes the timing of human affairs in exactly this way. It places the practitioner outside the flow she is reading. In human systems the practitioner's waiting is part of the flow.

Element 11: Practical Wisdom, Phronesis and Mētis

The claim

The Barcelona lecture uses Aristotle's distinction between episteme (scientific knowledge of what cannot be otherwise), techne (craft knowledge of how to make things) and phronesis (practical wisdom about how to act well in particular situations), with the weight on phronesis. Snowden says this matches a paragraph on judgement he added to his own definition of a complex system, and adds James C. Scott's mētis.

Phronesis

For Aristotle, phronesis is the virtue of deliberating well about what is good for a human being in a particular situation. It cannot be reduced to rules, because the situations it addresses are particular and variable, and no rule anticipates every case. It is learned through experience and shown in perception: the practically wise person sees what a situation calls for. It is bound up with character. Only a good person can be practically wise, because practical wisdom requires wanting the right things as well as knowing how to get them. In the twentieth century, phronesis was revived as an alternative to technical rationality: by Gadamer, who made it the model for hermeneutic understanding; by MacIntyre, who made it central to virtue ethics; and by Flyvbjerg, who proposed a "phronetic social science" that would study practice in its particularity rather than seeking general laws.

Mētis

Scott's Seeing Like a State (1998) uses the Greek mētis, the cunning intelligence of Odysseus, for the practical, local, experiential knowledge that high-modernist schemes override at their peril. Mētis is the knowledge of a farmer who knows when to plant by the leaves of a particular tree, of a sailor who reads weather in the colour of the sky, of a worker who knows the quirks of a particular machine. It is not systematic. It cannot be written down without loss, and it is lost when practices are standardised. Scott's argument is that states, in trying to make their territories "legible" through cadastral maps, standard measures, planned cities and scientific forestry, destroy the mētis on which the territories' actual working depended, often with disastrous results.

The function

Element 11 supplies the human faculty the causal cluster requires. If propensity must be read, retrospect cannot predict, and kairos must be recognised, then the practitioner needs a capacity that is neither scientific knowledge nor technical skill. She needs judgement: practical, perceptual, experiential, particular. Phronesis and mētis name it. They also connect anthro-complexity to a long tradition of critique of technical rationality, from Aristotle through Gadamer and MacIntyre to Scott. All of them say the same thing in different registers: human affairs cannot be governed by rules alone, and the attempt to govern them so destroys the knowledge they depend on.

The flag

Mētis has a political edge that phronesis lacks, and Scott's book is about it. Mētis is the knowledge that is destroyed when a state or a large organisation makes its domain legible. Legibility is the precondition of administration. The state cannot tax, conscript or plan what it cannot see, and seeing requires simplification: the map, the census, the standard. Chapter 24 will ask what kind of seeing anthro-complexity's instruments do. SenseMaker makes a population's narratives legible to an analyst through a signifier set. Estuarine mapping makes an organisation's constraints legible to a facilitator through a map. Neither is the high-modernist legibility Scott attacks; both are designed to keep context. But both are forms of legibility, both serve an administrator's need to see, and both, on Scott's argument, will destroy some of the mētis they make visible. Adding mētis to the definition of a complex system is a gesture of respect for local knowledge. Chapter 24 will ask whether the instruments show the same respect.

The Causal Cluster as a Whole

Element Claim Principal sources Practical consequence
7 Propensity, not causality Jullien; Law & Lin; Sunzi Read and join the configuration; intervene upstream
8 Coherence only in retrospect Rosen; Cynefin Probe, sense, respond; suspicion of recipes
9 Propensity belongs to complex systems as such Complexity science Human difference must lie in how, not whether
10 Kairos, not chronos Greek rhetoric and medicine; estuarine mapping Know when constraints become cheap to shift
11 Phronesis and mētis Aristotle; Scott Judgement as the irreducible faculty

Taken together, the causal cluster is the most persuasive part of anthro-complexity, and the part that has done the most good in organisational life. It tells managers that the forward-causal question, if I do this, what will happen?, has no answer in the situations that matter most to them, and that pretending otherwise is a choice with costs. It offers them propensity, retrospect, timing and judgement in its place. Each is real, each has a long philosophical pedigree, and each is borne out by the experience of anyone who has tried to change an organisation by plan and watched the plan dissolve on contact with the organisation.

The flags planted in this chapter all point the same way. Propensity in human affairs includes the reader of the propensity (Element 7). Retrospective coherence in human affairs belongs to an interpreter, and published retrospect becomes prospective constraint (Element 8). Human propensity is reflexive (Element 9). The tide in human affairs turns partly because someone was seen to wait for it (Element 10). The practical wisdom the practitioner needs is threatened by the legibility her instruments provide (Element 11). In every case, the element is true, and in every case, its truth in the human domain is more radical than the image through which anthro-complexity states it. The images (the surfer and the wave, the sailor and the tide, the child and its parents) are images of a reader outside a non-reflexive medium. The truth, in human systems, is that the reader is in the medium and the medium reads the reader.

THE SURFER READS THE WAVE —

THE SAILOR WAITS FOR THE TIDE —

BUT IN HUMAN WATERS THE WAVE IS WATCHING TOO.

Addition: Lewin Recovered

The November post's first item of shared ground is that Lewin never wrote "refreeze". Both Ing and Snowden draw on Cummings, Bridgman and Brown's 2016 paper in Human Relations, which showed that the three-step model of change (unfreeze, change, refreeze) is a later reconstruction, with refreezing first appearing in a 1950 conference paper by Festinger, three years after Lewin's death. Snowden adds that Lewin belongs among the abused, not the complicit. This addition asks what the real Lewin offers anthro-complexity, because the real Lewin, recovered from the reconstruction, is closer to anthro-complexity, and to this disputation's argument, than either the three-step model or its critics usually suggest.

Field theory

Kurt Lewin (1890–1947) was a Gestalt psychologist trained in Berlin who emigrated to the United States in 1933 and became one of the founders of social psychology, group dynamics and action research. His central theoretical contribution was field theory. Its core formula is that behaviour is a function of the person and the environment, B = f(P, E), where the person and the environment are understood not as separate entities but as interdependent parts of a single life space. The life space is the totality of facts that determine a person's behaviour at a given moment: her needs, goals and perceptions, and the forces acting on her from her psychological environment. Behaviour is the resultant of forces in the field.

Lewin's field theory is, in the disputation's vocabulary, a theory of dispositional states (Element 4). Behaviour is not caused by a single factor but arises from the configuration of forces in the field at a moment, and changing behaviour means changing the field. It is also a theory of behaviour as the shadow of conditions (Element 18), stated seventy years before Snowden's formulation. And it is relational. The person is not a container with properties facing an environment. She is a region of the life space, defined by her relations to the other regions.

Quasi-stationary equilibrium

What became the three-step model began, in Lewin's own work, as something different. Lewin described group life as a quasi-stationary equilibrium: a state that looks stable but is in fact maintained by a balance of opposing forces, some driving towards change and some restraining it. The level of a group's productivity, for example, is not a fixed property. It is the point at which the forces pushing it up (incentives, pride, pressure) and the forces holding it down (fatigue, norms, resistance to the pace setters) balance. The equilibrium is quasi-stationary because the forces are continually in play, and the level fluctuates around the balance point.

To change the level, Lewin argued, one must change the balance of forces, and it is usually more effective to reduce restraining forces than to increase driving forces, because increasing driving forces tends to increase tension and resistance. That is close to anthro-complexity's emphasis on shifting constraints, on finding which restraints can be loosened, which estuarine mapping does. Lewin also observed that group standards are among the strongest restraining forces, and that changing individuals' behaviour without changing the group's standards tends to fail, because the individual returns to the group's norm. It is better to change the group's standards, through the group's own participation. This is constitutive constraint (Thesis IV), and its change through participation.

Action research

Lewin's third contribution, action research, is the one most relevant to this disputation's argument about the observer. Lewin proposed that social research should be conducted with the people being studied, in cycles of planning, action, observation and reflection, so that research and action are integrated and the researched are participants in the research. His famous dictum is that there is nothing so practical as a good theory, and its corollary is that the best way to understand a social system is to try to change it. Action research rejected the observer outside. The researcher is in the system, acting with its members, and her understanding comes from the action.

Action research is the closest the mid-twentieth century came to the reflexive practice this disputation proposes. It is participatory, it is cyclical (probe, sense, respond, before Cynefin), it integrates research and intervention, and it places the researcher inside.

What the reconstruction lost

The three-step model, as it circulated in change management from the 1950s onward, lost almost everything distinctive in Lewin's thought.

Lewin Three-step model
Field of forces; behaviour as resultant A system to be changed
Quasi-stationary equilibrium, continually in play A stable state to be unfrozen
Reduce restraining forces Drive change through
Change group standards through participation Change behaviour through programmes
Action research: researcher inside, with participants Change agent outside, acting on the organisation
Cyclical inquiry Linear sequence
"Refreeze" absent Refreeze as the end state

Each loss moved Lewin's thought towards the position Snowden and Ing both criticise: the system held stable while it is redesigned, the change agent outside it, and change as a linear programme. Each loss also moved it away from the relational, participatory and inside position that this disputation argues anthro-complexity's own premises require.

The lesson

The lesson of Lewin's fate is twofold. First, it confirms Element 22, resistance to commodification. A careful, relational and participatory body of work was simplified, packaged and sold until what circulated bore little relation to what was written. Second, it shows what was lost in the simplification. The losses were precisely the relational, participatory and reflexive elements, the observer inside and the group's own participation in changing its standards. The commodified version kept the outside position and discarded the inside.

That pattern is the one this disputation has traced in anthro-complexity's own genealogy (Chapter 6). It is not a pattern of bad faith. It is what happens when ideas are carried into organisations by consultants who must sell them and managers who must apply them. What is lost first is what does not fit the client's position, which is the inside. Snowden's defence of Lewin against his reconstructors is a defence of the inside, and the disputation asks that the same defence be extended to anthro-complexity's own instruments, before they suffer Lewin's fate.

LEWIN NEVER WROTE REFREEZE —

HE WROTE FIELD, FORCE, GROUP STANDARD, ACTION RESEARCH —

THE INSIDE WAS LOST FIRST. DEFEND IT IN YOUR OWN INSTRUMENTS.

Addition: The Five Causal and Temporal Elements Taken Slowly

This addition continues the slow reading begun at the end of Chapter 2. The case is the same invented one: Calder Vale Homes, its damp and mould problem, and Asha, the director asked to fix it by spring. The order of treatment is the same: plain words, example, the difference from complex adaptive systems, the common misreading, and a first interrogation with a provisional verdict.

Element 7 slowly: propensity over causality

In plain words. In ordered situations we can say that A causes B. In complex ones we cannot, because the same action produces different results on different days. What we can say is that the situation has a propensity: a built-in tendency that makes some outcomes easy and others hard. A slope does not cause a ball to arrive at a particular spot, but it makes downhill easy. The practical advice is to stop pushing for an outcome and to change the slope.

The idea reaches anthro-complexity from François Jullien's account of the Chinese term 勢 (shì), the potential of a configuration, and from Karl Popper's late view that probabilities are real dispositional properties of situations.

The example. Asha could instruct operatives to investigate causes of mould and not just wash it. This is a push. It will work while she is watching. The propensity of the situation runs the other way: the job code, the payment per visit, the route schedule and the supervisor's targets all make a quick wash the easy thing. A propensity approach changes those. Add a paid job code for "humidity reading and cause noted". Give operatives a meter. Let the caretaker on each estate book a follow-up directly. Then leave it alone and see which way jobs start to flow.

How CAS differs. CAS theory still speaks of causes, at the level of rules. An agent follows a rule, and macro patterns emerge from many agents doing so. Change the rule and you change the pattern. Propensity thinking declines to locate the cause in rules held by agents. It locates the tendency in the whole configuration, including things that follow no rule at all.

The common misreading. "Propensity means you cannot be held responsible for results." It means that the responsible act is to shape conditions and watch, and that claiming to guarantee outcomes is the irresponsible one.

First interrogation. In the Chinese sources 勢 has three senses, and anthro-complexity uses one. It uses the cosmological sense: the tendency of a configuration of things. The strategic sense, in Sunzi, has the general inside the configuration and being read by an opponent. The political sense, in Han Feizi, is the power that attaches to a position regardless of who holds it. Both left-out senses apply at Calder Vale. Asha is being read: the contractor's managers are working out what she will check. And much of the slope she wants to change is positional power. The contract terms lean as they do because the association had a strong hand when it tendered and used it to push risk downward. Holds, in one sense of three Chapter 35 gives the full argument.

Element 8 slowly: coherence only in retrospect

In plain words. After something has happened in a complex system, we can usually tell a story that makes it look as if it had to happen. Before it happened, nobody could have told that story with confidence. Anthro-complexity calls this retrospective coherence. The lesson is that a good explanation of the past is not a recipe for the future, because the next time the small details will differ and the path will go elsewhere.

The example. The estate with almost no reopened jobs looks, in hindsight, like a model. A caretaker built relationships with two operatives, they brought a meter, tenants trusted them, and problems were fixed at source. The obvious move is to write this up as the Fieldhead Model and roll it out across all three towns. Retrospective coherence warns against exactly that. The story makes sense backwards. It depended on that caretaker, those operatives, a supervisor who looked the other way, and a block where neighbours talked. Mandate it elsewhere and you get a procedure without the relationships, which is the wash-and-paint job with an extra form.

How CAS differs. A simulation can be rerun. In a CAS model, one can run history a thousand times and learn the distribution of outcomes, which gives a kind of foresight about tendencies. Anthro-complexity holds that human history cannot be rerun and that the simulated agents are too thin to stand in for the people. So hindsight stays hindsight.

The common misreading. "You cannot learn from experience." You can. What you learn is what conditions made something possible, and you can try to grow those conditions elsewhere. You cannot learn what steps to copy.

First interrogation. Retrospective stories are not only inaccurate as forecasts. They are told by someone, and they do work in the present. Who gets to tell the story of why Fieldhead went well? If Asha tells it, it becomes a story about enabling leadership. If the caretaker tells it, it may be a story about working round a system that obstructs her. If the contractor tells it, it becomes evidence that the contract works when tenants cooperate. Mead's point in The Philosophy of the Present is that every present reconstructs its past for its own purposes. Anthro-complexity sees that the past cannot be projected forward. It gives less attention to the contest over who narrates it. Holds, and can go further

Element 9 slowly: propensity belongs to complex systems, living or not

In plain words. Snowden is careful to say that propensity is not special to living things or to people. A river delta, a weather system and a market all have dispositions. This matters because it keeps the human distinction, Element 12, from being a claim that only humans are complex. Humans are distinctive within the class of complex systems, and for different reasons.

The example. The buildings at Calder Vale have propensities of their own. A 1960s block with solid concrete lintels is disposed to cold bridging whatever anyone does or thinks. Moist air will condense on the coldest surface. That disposition does not listen to stories. It responds to insulation and ventilation. So part of Asha's situation has propensity in the physical sense alone, and it would be a mistake to treat it as a matter of narrative.

How CAS differs. There is little difference here. CAS theory is likewise applied across living and non-living systems. The disagreement is over what is added when the system is human.

The common misreading. "Anthro-complexity is only about people." Its ontology is general. Only its distinctive claims concern the human.

First interrogation. This element sets up the disputation's central question. If propensity is common to estuaries and organisations, and if what distinguishes the human is reflexivity, then the methods for reading human propensity ought to differ from the methods for reading physical propensity precisely where reflexivity bites. The lintel does not care that it is being surveyed. The tenant does care that she is being asked. A practice whose founding image is the estuary has borrowed its picture of reading from the non-reflexive case. Holds The element is accepted, and it is the premise from which Thesis III argues.

Element 10 slowly: kairos, not chronos

In plain words. The Greeks had two words for time. Chronos is clock time: evenly divided, measurable, the time of schedules. Kairos is the right moment: the opening that must be seized because it will close. Complex systems do not change at a steady rate. They sit, and then they tip. So timing matters more than scheduling, and a good practitioner watches for the moment when a small act will carry.

The example. Asha's programme plan is pure chronos: discovery in the first quarter, pilot in the second, rollout in the third. Meanwhile there are moments. In the week after the newspaper story, the board will agree to things it refused for two years. In October, when the first cold snap brings the first complaints and people are angry but not yet resigned, tenants will come to a meeting. When the contract comes up for its break clause, the job codes can be rewritten. Each of those is a window. A plan that schedules "tenant engagement" for February, when everyone has given up for the winter, has the right activity at the wrong time.

How CAS differs. CAS models have tipping points and phase transitions, and they can locate them in a parameter space. They do not have kairos in the Greek sense, because kairos is the moment as it presents itself to someone who must act. It includes the actor's readiness and standing. A simulation has thresholds, and nobody in it faces an occasion.

The common misreading. "Wait for the perfect moment." Kairos is recognised by people who are already active and in contact with the situation. It is not visible from a distance.

First interrogation. A moment is opportune for somebody. The week after the newspaper story is an opening for Asha and a disaster for the tenant whose child was in hospital. The break clause is kairos for the association and a threat to the contractor's staff. And among people, unlike tides, the moment is partly made by how others read your readiness. If the contractor knows Asha is waiting for the break clause, the contractor acts first. Waiting is a move that others interpret, which is the argument of Chapter 21. Holds, but the moment is positional and reflexive

Element 11 slowly: practical wisdom, phronesis and mētis

In plain words. If complex situations cannot be handled by rules, what handles them? Anthro-complexity's answer is a kind of knowing that cannot be written down in full. Aristotle's phronesis is practical wisdom: the trained capacity to see what a particular situation calls for. The Greek mētis, which James C. Scott revived in Seeing Like a State, is cunning local know-how: the pilot's knowledge of this harbour, the farmer's knowledge of this field. Both are learned by long practice in a place, and both are destroyed by schemes that replace local judgement with standard procedure.

The example. The Fieldhead caretaker has mētis. She knows which flats have a tumble dryer venting indoors, which tenants will not open the door to a stranger, and which operative will actually look behind the wardrobe. None of that is in the housing management system. Asha has, or needs, phronesis: the judgement to see that this is the knowledge that matters, and to protect it from her own improvement programme. The contract, with its standard codes and visit times, is what Scott would call a high-modernist scheme. It made the service legible to the board and stripped out the local knowledge that made it work.

How CAS differs. A CAS agent has a rule set, sometimes one that evolves. It does not have judgement. The Santa Fe tradition has little to say about phronesis because its agents are simple by design: the interest lies in what simple agents produce together. Anthro-complexity makes the non-simple agent, with judgement and local knowledge, essential.

The common misreading. "Trust the frontline and ignore expertise." Technical knowledge is right for ordered problems. The claim is that complex situations also need a kind of knowing that expertise does not supply.

First interrogation. This is the element on which anthro-complexity and the disputation are closest, and the question it raises is a sharp one. If mētis is what matters, and if it is destroyed by being made legible to a centre, then what should we think of an instrument that collects thousands of local accounts, has them coded by their tellers into a designed set of signifiers, and displays the patterns to decision-makers? Snowden's reply is that self-signification keeps interpretation with the narrator and so protects the local. Thesis VI replies that the arrangement still moves knowledge from the many to the centre, and that the caretaker does not see the map her story helped to draw. Scott's own test is who a scheme makes the world legible to. Holds, and indicts the method chain

What the five come to

Read slowly, the causal and temporal elements tell Asha to change slopes and not push outcomes, to distrust tidy stories about past success, to remember that some of her problem is plain physics, to watch for openings, and to protect the local knowledge that her own reporting systems tend to erase.

The pattern in the interrogations continues from Chapter 2, and it becomes more specific. Three things recur. First, position: propensity, the right moment and the authorised story all look different depending on where one stands, and power is part of the slope. Second, reflexivity: people read the reader, so that waiting, probing and measuring are moves in a game and not observations of one. Third, direction of flow: knowledge that is local by nature is being gathered toward a centre. These three are, in outline, Theses V, III and VI.

CHANGE THE SLOPE —

WATCH FOR THE MOMENT —

AND REMEMBER THAT THE SLOPE IS WATCHING YOU

Addition: Cross-Lights From Book Nine

Propensity and the knot. Snowden's image of the knot belongs with Element 7. A knot has a structure, and one who understands it can undo it with a few moves. Understanding here is a participant's. It was got by handling knots. Chapter 56 uses the image to argue that how compressible a situation is depends on the knower and is changed by acting.

Retrospective coherence and the alibi. Element 8 says nobody could have foreseen a complex outcome. Snowden distinguishes two uses of that thought. As an alibi it excuses. As a discipline it places responsibility on the design of conditions. Chapter 52 asks that instruments, like probes, carry stated signs of failure.

Mētis and the mature citizen. Element 11 honours local practical knowledge. The foreword to Flinn's book, discussed in Chapter 57, says that the best tool a practitioner can develop is their own practice of mature citizenship. The two thoughts are close, and both raise the question of how such knowing is fostered widely.

Chapter 4

The Human Cluster: Where the Name Comes From

6,324 words, about 27 minutes of reading

The Hinge of the Name

The ontological and causal clusters are shared, in large part, with complexity science as such. A physicist studying dissipative structures, an ecologist studying panarchy and a theoretical biologist studying autocatalysis would recognise most of Elements 1 to 11. They would accept found coherence, actants and constraints, dispositions and vectors, far-from-equilibrium order, propensity, retrospective entailment and timing, perhaps with some changes of vocabulary. What makes anthro-complexity anthro is the third cluster, and Snowden is clear that this is where the name comes from.

The cluster is stated in the November post as the opposite of a move in Ing's programme. That is the best place to begin, because the opposition shows what is at stake.

The Move Anthro-Complexity Opposes

Ing's paper, Snowden says, deliberately downplays what is distinctive about humans. It does so in three steps.

  1. Gibson. It follows Gibson's ecological psychology in treating the human/animal distinction as of limited significance. For Gibson, perception is the pickup of affordances, the possibilities for action an environment offers an organism. Affordances are relational: a surface affords walking to a creature of a certain size and weight. The theory is deliberately continuous across species. A human perceives the world as other animals do, by picking up what it affords for action.
  2. Ingold. It follows Ingold's ecological anthropology in treating humans as one kind of animal. Ingold, trained as an anthropologist of reindeer herders, has spent his career arguing against the separation of the human from the natural. Humans are organisms among organisms, growing and developing within an environment that is itself growing and developing, and the distinction between nature and culture is an artefact of a particular Western intellectual history.
  3. Mayr. It turns to Ernst Mayr's concept of teleonomy, goal-direction arising from a programme, partly because it sidesteps the question of human will. Mayr introduced teleonomy to name the apparent purposiveness of living things without importing teleology in the Aristotelian sense. An organism behaves as if it had goals, because it carries a genetic programme shaped by natural selection, and the programme directs its behaviour towards outcomes that served its ancestors. No conscious purpose is needed.

Ing puts the result in a phrase: attention trumps intention. What matters for systems change is not what people intend but what they attend to: which affordances they pick up, which signals they notice, which rhythms they fall into.

Snowden's response is unambiguous: Anthro-complexity makes exactly the opposite move, and that is where it gets its name.

Element 12: The Three Marks of the Human

The claim

Human systems are complex, but they differ from other complex systems, including other living ones, because humans:

  • hold multiple identities at once;
  • act with intentionality; and
  • create abstractions and shared narratives that change the very constraints they operate within.

Humans are not ants, and models drawn from insect behaviour mislead when applied to organisations.

Each mark needs expanding, and the expansion draws on the wider corpus as well as the November post.

The first mark: multiple identities

Snowden has long argued that a human being is not a single agent with a single identity, as most agent-based models assume. A person is several identities at once, and moves between them according to context, often without noticing. A nurse at work is a professional bound by clinical norms. On the union committee she is an advocate for her colleagues. At home she is a parent. Online she is a member of a community of interest. These identities have different values, different loyalties and different patterns of attention. Which one is active at a given moment depends on the context, and the context can change in an instant: a single phrase in a meeting can shift a person from professional to advocate.

In other writing Snowden distinguishes roles, the formal positions people occupy, from crews, temporary identities aligned to a role and formed around a shared task. A crew is a group that forms, does its work, and dissolves, and its members carry the identity while they are in it. An aircraft cabin crew, a surgical team or a fire crew are examples. The Compass and the Loom uses Snowden's term "crew" for a temporary identity aligned to a role, and notes its importance for anthro-complexity's account of how human collectives coordinate without central control.

The consequence for modelling is severe. An agent-based model gives each agent a set of rules and lets them interact. If a human is several agents, switching between rule sets according to cues the model does not represent, the model's agents are not humans. They are idealisations of one identity at a time. Any emergent pattern the model produces is a pattern in a world of single-identity agents, which is not the world humans live in.

The second mark: intentionality

Humans act with intention. They form representations of possible futures, prefer some to others, and act to bring the preferred ones about. This is not the "as if" purposiveness of Mayr's teleonomy, which needs no representation of the future, only a programme shaped by past selection. It is purposiveness proper: action guided by an image of what is not yet.

The Compass and the Loom glosses this as the ability to act toward imagined futures. The gloss shows why intentionality matters for complexity. A complex system of non-intending agents is shaped by its past: the rules agents follow, the constraints they have co-evolved, the attractors their interactions have formed. A complex system of intending agents is also shaped by its imagined futures, which agents act towards and which, being imagined, can change without any change in the past. A rumour of a merger changes behaviour before any merger occurs, and a vision statement changes behaviour, sometimes, before any vision is realised. In a human system, the future acts on the present through intention.

The third mark: abstraction and narrative

The third mark is the most powerful and the most distinctive. Humans create abstractions and shared narratives that change the very constraints they operate within. A constraint in a non-human complex system (the shape of a riverbed, the chemistry of a cell, the topology of a starling's neighbourhood) is a material fact. It changes only through material processes. A constraint in a human system can be changed by a story. A rule can be reinterpreted, a hierarchy delegitimised, a boundary redrawn on paper and then on the ground. A shared narrative of "how we do things here" is a constraint. A new narrative, if it is taken up, is a new constraint. Humans do not only operate within constraints. They tell stories about the constraints, and the stories change them.

This is what Snowden means by saying that humans are not ants. An ant colony is a complex system of great sophistication. It forages, builds, defends itself and allocates labour without central control, through local interactions mediated by pheromones and contact. But no ant tells a story about the colony, and no story changes the colony's constraints. The colony's constraints are pheromone gradients, nest architecture and the genetic programmes of its castes. They change through material processes and evolution, not narrative. A human organisation is constrained by rules, roles, budgets, buildings and habits, and every one of these can be changed by a story about it. The capacity for abstraction, for making a representation that floats free of any particular material instance and can be applied to many, is what makes this possible. A rule is an abstraction. A budget is an abstraction. A role is an abstraction. And abstractions can be re-described.

The broader frame

The Compass and the Loom sets this element in a broader frame. Human complex adaptive systems, it says, are part of nature but irreducibly different from other natural systems. Agents have intelligence, identity and intention, enacted through language, relationship and practice rather than located in a bounded interior. Humans are homo faber (makers), homo ludens (players) and story-telling beings (storytellers). They theorise their situation and act on the theory. The consequence it draws is the one this book will turn on: any framework applied to a human system becomes an actor within it, so human systems are "reflexively complex", and describing them changes them.

That treatise also develops Snowden's three marks into a pair of triads. The 3Is (intelligence, identity, intention) are what human agents have. The 3Ss (sapience, salience, sentience) are what they do with it. Intelligence shapes and is corrected by sapience; identity shapes salience, what one notices; intention shapes what sentience registers as significant. Neither triad can be addressed alone. This elaboration is useful, and it will be examined critically in Chapter 19, because its grammar (what agents have) shows clearly the possessive form the human cluster takes.

Element 13: The Asymmetry Principle

The claim

The second element of the human cluster is the asymmetry principle, stated in the river-and-planners image:

In paraphrase: the river shapes the city and the city's planners argue about the river. Both are actants, but only one of them can change the story it tells about the other. That asymmetry, Snowden says, is where anthro-complexity starts.

The structure of the principle

The principle has a precise structure. It does not say that humans are actants and rivers are not. Both are. It does not say that humans affect rivers more than rivers affect humans. The river shapes the city at least as much as the planners shape the river. It says that the relation between them is asymmetric in one respect: the capacity to change the story each tells about the other. The river does not tell stories, so it cannot change them. The planners tell stories about the river, and they can change those stories. The river is a threat, a resource, an amenity, a boundary, a sacred presence or a sewer, and the story they tell shapes what they do to the river and what they let it do to them.

The principle is thus a principle of narrative asymmetry within a symmetric ontology of action. Both parties act, but only one narrates. Because narrating changes constraints (Element 12, third mark), the narrating party has a kind of agency the other lacks. It can change the conditions of the relation by changing how it understands the relation.

The function

The asymmetry principle does two jobs. First, it reconciles anthro-complexity's debt to the new materialisms with its insistence on human distinctiveness. Matter acts, and this is the debt; but matter does not narrate, and that is the distinction. Second, it identifies the locus of human distinctiveness in complexity terms. The three marks are capacities. The asymmetry principle says what those capacities do in a complex system: they let one class of actants change the constraints of the system by changing their representations of it. That is the specifically complex consequence of being human.

The flag

The asymmetry principle is the single most important claim in the November post for the argument of this book, and it will be examined in Chapter 20. The flag here is simple. If the capacity to change the story one tells about the other is what makes humans distinctive actants, the principle applies to every human who tells stories about complex systems, including complexity theorists. Cynefin is a story about situations. Estuarine mapping is a story about constraints. SenseMaker produces stories about stories. The November post is a story about how two traditions stand to each other. Each of these changes the constraints of the field it describes. The asymmetry principle, stated without exception, makes anthro-complexity an actant in every system it describes. The question is whether anthro-complexity's practice is designed for that.

Element 14: Entanglement Without a Flat Ontology

The claim

Anthro-complexity is not a retreat from the material. It has some of its origins in the new materialisms. Matter is not inert, non-human actants make a difference, and humans are entangled with the material world rather than standing outside it and acting upon it. But it does not follow, Snowden insists, that the human should be flattened into one more participant in a field of intra-actions, as Karen Barad's agential realism tends to do, or that every actant counts equally, as the flatter readings of Latour would have it.

Barad

Karen Barad's Meeting the Universe Halfway (2007) develops an "agential realism" from Niels Bohr's philosophy of quantum mechanics. Bohr held that the properties of a quantum system are not determinate apart from the experimental arrangement that measures them. Position and momentum are not properties a particle has independently of how it is observed; they are defined only within a particular apparatus. Barad generalises this into an ontology. The basic units of reality are not things but phenomena, entanglements of what was previously called "object" and "observer". Within phenomena, agential cuts enact local distinctions between subject and object, but the distinctions are not given in advance. The key claim, which Snowden rejects, is that relata do not precede relations: there are no distinct things that then enter into relation, only intra-actions from which things emerge.

The tacit–explicit–entangled trialectic

Snowden uses "entangled" in his own work, in the trialectic of tacit, explicit and entangled knowledge. The November post is careful to separate his use from Barad's. When he uses "entangled", he means knowledge that exists only in the interaction of people, artefacts and context, and cannot be extracted from it. He does not mean Barad's claim that no distinct relata exist before the relation. The people remain distinct, even when what they know together does not.

The trialectic is worth stating. Tacit knowledge, in Polanyi's sense, is what we know but cannot tell: how to ride a bicycle, how to recognise a face. Explicit knowledge is what can be codified and transmitted: a procedure, a formula, a map. Entangled knowledge is a third kind, neither in the heads of individuals (tacit) nor in codified artefacts (explicit), but in the interaction of people, artefacts and context. It exists only while the interaction goes on, and it disappears when the interaction is broken up. A surgical team's knowledge of how to work together, a trading floor's sense of the market and a ship's crew's feel for the vessel are entangled knowledge. They cannot be extracted into individual heads or written procedures without loss.

The function

Element 14 positions anthro-complexity between two errors. One is the Cartesian error of a subject standing outside a world of inert objects and acting on them. Anthro-complexity rejects it: humans are entangled with the material world. The other is the flat-ontology error of dissolving the subject into a field of intra-actions in which nothing is distinctive. Anthro-complexity rejects this too: the people remain distinct. The middle position is that knowledge can be entangled while knowers are not. What we know together exists only in our interaction, but we remain distinct persons, and we bring to the interaction the distinctive human capacities of Element 12.

The flag

The middle position is attractive, but Chapter 19 will argue it is unstable. If entangled knowledge exists only in interaction and cannot be extracted, and if identity shapes what one notices and intention shapes what one finds significant (the 3S pairing), then a person's identities and intentions are partly constituted by the entangled knowledge of the interactions they take part in. A nurse's professional identity is not something she carries into the ward and then applies. It is formed and re-formed in the ward, in the entangled knowledge of the team. So "the people remain distinct" is true, but not in the way the November post needs. The people remain distinct as patterns of participation. They are not distinct as bearers of identities and intentions that exist before the participation. Elias's figurational sociology and Mead's social psychology, which Chapters 27 and 28 take in turn, offer the alternative: persons are distinct, but their distinctness is produced in relation. That is neither Barad's dissolution of relata nor Descartes's sovereign subject, and it is the position anthro-complexity needs and does not yet hold.

Element 15: Against the Programme

The claim

The final element of the human cluster is a negative one. Snowden is wary of teleonomy, Mayr's goal-direction arising from a programme, because a programme is a computational metaphor, and he has spent much of the anthro-complexity series arguing against treating the human brain as something that runs one.

The argument against the computational metaphor

The argument against the computational metaphor of mind is old and has many forms: Dreyfus's critique of artificial intelligence, Searle's Chinese room, the enactivist programme of Varela, Thompson and Rosch, and the embodied-cognition movement. Snowden's version, as the series has developed it, draws on several of these and on the neuroscience of predictive processing and embodied cognition. The brain is not a general-purpose computer running programmes over symbolic representations. It is an embodied, situated organ that evolved to guide action in an environment, and its "computations", if the word applies at all, are not separable from the body and world in which they occur. Human cognition is pattern-based, not rule-based. It works by matching the present against a vast store of fragmented past experience, and it is highly susceptible to framing, priming and context.

The rejection of teleonomy follows. If human goal-direction were a programme, it would be in principle specifiable. It could be written down, modelled and predicted. The appeal to intentionality is an appeal to something that cannot be specified: action towards imagined futures, where the imagining is itself open-ended and context-dependent. A programme closes the future. Intention keeps it open.

The function

Element 15 protects the human cluster from being reduced to CAS. Classic complex adaptive systems theory, as Holland developed it, is computational at its core. Agents are rule-following systems; adaptation is the modification of rules by something like a genetic algorithm; emergence is what happens when many rule-following agents interact. If humans ran programmes, they could be modelled as CAS agents, and anthro-complexity would collapse into a special case of CAS. By rejecting the programme, Element 15 keeps the human distinctive.

The flag

Element 15 is an ontological commitment: humans do not run programmes. Chapter 24 will ask whether it is matched by a methodological one. The anthro-complexity instruments, especially as they move towards the human/AI hybrid that The Compass and the Loom examines, translate human narratives into data, data into patterns, and patterns into inputs for computational systems. A theory that says humans are not computers can still build instruments that treat human outputs as computer inputs. Whether that is a contradiction depends on what happens to the human in the translation. Chapter 24 will argue that something important is lost, and that it is the same thing Element 15 was meant to protect.

The Human Cluster as a Whole

Element Claim Against Principal sources
12 Three marks: multiple identities, intention, abstraction/narrative Gibson, Ingold, Mayr as used by Ing; "attention trumps intention" Snowden's corpus; crews and roles; Compass & Loom 3Is/3Ss
13 Asymmetry principle: only humans change the story they tell about the other Flat ontologies The river and the planners
14 Entanglement without flat ontology; people remain distinct Barad; flat Latour Tacit–explicit–entangled trialectic; new materialisms
15 Against the programme Teleonomy; computational theory of mind Enactivism; embodied cognition; anti-AI critique

The human cluster is the heart of anthro-complexity, and it is a strong heart. Its claims are true, or close to true. Humans do hold several identities and move between them according to context. Humans do act towards imagined futures. Humans do create abstractions and narratives that change the constraints they work within. The flat ontologies do obscure something important when they dissolve the human into a field of intra-actions. The computational theory of mind is inadequate to human cognition. And the practical consequence Snowden draws, that models drawn from insect behaviour mislead when applied to organisations, is one of the most important corrections anthro-complexity offers to the management literature, where swarm metaphors and agent-based models are widespread.

The flags in this chapter, however, point to a single pattern. Each element of the human cluster is stated as a property of persons: persons have multiple identities, persons act with intention, persons create abstractions, persons remain distinct. Each element, when its consequences are followed, turns out to be a property of relations between persons. Identities are formed and activated in interaction. Intentions are formed in response to others' gestures. Abstractions change constraints only when they are taken up by others. Distinctness is produced in figuration, not given before it. The asymmetry principle, the most relational of the elements because it is about one actant's capacity to re-describe its relation to another, is the clue. What makes humans distinctive is not what they have but what happens between them when they describe one another and the world they share. That is reflexivity, and it is relational all the way down.

Chapter 14 will develop this into the claim that reflexivity is the deep ground of human distinctiveness, the reason anthro-complexity is right to treat human complexity differently. Chapter 19 will turn it against anthro-complexity, as the possessive fallacy. Between the two lies the work of showing that the difference matters: that a human distinction grounded in properties of persons leads to different instruments than one grounded in reflexive relations, and that the instruments anthro-complexity has built are the instruments of the first.

NOT WHAT HUMANS HAVE —

BUT WHAT HAPPENS BETWEEN THEM —

WHEN THEY DESCRIBE ONE ANOTHER.

Addition: The Four Human Elements Taken Slowly

This is the third of the four slow readings. The case is still Calder Vale Homes and its damp and mould problem. The four elements treated here are the ones that give anthro-complexity its name, and they are the ones the disputation presses hardest. They are therefore taken more slowly than the rest, and the three marks of the human are taken one at a time.

Element 12 slowly: the three marks of the human

In plain words. Anthro-complexity says that human systems are complex systems with something added. People differ from ants, starlings and neurons in three ways that change how the system behaves. They hold several identities and move between them. They act with intention. And they create abstractions, above all stories, that alter the conditions under which they themselves act. Because of these three marks, models built from simple rule-following agents will mislead when applied to people.

The first mark: multiple identities

In plain words. An ant is a forager or a soldier. A person is a parent, a tenant, a nurse on nights, a member of a mosque, a neighbour and a complainant, and may be several of those within the same hour. Which identity is in play changes what the person notices and does, and the switch can be triggered by a small cue.

The example. Mrs Okafor in flat 14 is, to the housing system, a tenant with two logged repairs. When the operative arrives she is a mother with an asthmatic son, and she wants the cause found. When the arrears letter arrives she is a debtor, and she goes quiet. At the residents' meeting she is the woman from the ground floor whom others look to. The operative is also several people: an employee with nine jobs to close today, a tradesman who knows a wash will not hold, and a father. Whether the visit goes well depends on which of these meet. A model that gives each agent one role and one rule cannot represent the visit.

First interrogation. Look at how the mark is worded: humans hold multiple identities. The verb treats identities as items a person has and selects among. But Mrs Okafor did not choose to be a debtor that morning. The letter made her one. Her identity as someone others look to exists only because others look. Mead's account is that each identity is a me, a set of others' attitudes taken toward oneself, called out by a situation. On that account the multiplicity is real, and its source is the multiplicity of relations a person stands in. It is not a set of possessions. The mark is true. Its grammar puts it in the wrong place. This is Thesis I. True, misplaced

The second mark: intentionality

In plain words. People act for reasons and toward ends they can represent to themselves. They plan, promise, deceive, resist and change their minds. An ant following a pheromone trail does none of this. So a human system can do things no rule-following system does: it can anticipate an intervention and act to defeat it.

The example. When Calder Vale introduced its target for jobs closed within twenty days, nobody set out to hide mould. Supervisors intended to hit a number their bonus depended on. Operatives intended to finish their rounds. Tenants intended to avoid trouble. The result was a system that conceals the problem, and nobody intended that. But intention is everywhere in how it came about, and when Asha changes the target, everyone will form new intentions about the new target within a week.

First interrogation. The wording again is that humans act with intentionality, as with an instrument. Two questions follow. First, where do intentions come from? Tomasello's developmental evidence is that the capacity to intend, in the human form, grows out of shared intention: joint attention and joint goals between infant and carer come first. Juarrero's own account, on which anthro-complexity rests, treats an intention as a constraint formed in a context and continually reshaped by feedback from it. On both accounts intention is relational in its origin and in its operation. Second, the example shows what the mark leaves out. The supervisors' intentions were formed by a bonus scheme designed by others. To say that people "act with intention" without saying who shapes the conditions under which intentions form is to leave out power. True, under-described

The third mark: abstraction and narrative

In plain words. People make things that are not physically present: categories, rules, money, plans, reputations, stories. These abstractions then act back on the people who made them. A story about what kind of place this is changes what people will try. This is the mark Snowden weights most, and it supports the asymmetry principle.

The example. Three stories govern Calder Vale. Among staff: "it's lifestyle", meaning that mould is caused by how tenants live. Among tenants: "complain and they come after you". On the board: "we are a ninety-one per cent organisation". None is a physical fact. Each constrains action as firmly as a wall. An operative who believes "it's lifestyle" does not look for a leaking gutter. A tenant who believes the second story does not report. A director who believes the third does not ask. Change a story and the possible actions change.

First interrogation. Here the possessive grammar says that humans create abstractions. But no one at Calder Vale created "it's lifestyle". It formed in thousands of exchanges, was repeated in training, got written into a leaflet about ventilation, and now nobody can say whose it is. Searle's analysis shows a further point. Abstractions that carry obligations exist by collective acceptance, and the power to have one's classification accepted is unequally held. A surveyor's "lifestyle" goes into the file. A tenant's "structural" does not. The mark is right that stories change constraints. It does not say that some people's stories change constraints far more than others'. True, and silent on whose story counts

How CAS differs on all three. The classical CAS agent has one role, a rule set in place of intentions, and no capacity to redescribe the system it is in. Richer agent-based models add some of this back: agents with beliefs, several roles, even models of other agents. Snowden's position is that such additions do not close the gap, because the modeller still fixes in advance what the agents can become, and real people are not so fixed. Chapter 7 discusses how far the richer models go.

The common misreading. "Humans are complex and animals are simple." The claim is narrower. Other species have rich social lives. The three marks pick out what alters the dynamics of the system, and they are meant as a warning against one specific inference: from the success of ant and flock models to their use on organisations.

Element 13 slowly: the asymmetry principle

In plain words. A river shapes a city, and the people of the city tell stories about the river. The stories lead to embankments, bridges and flood plans, which change the river. But the river tells no stories about the city. Influence runs both ways. Narration runs one way. That is the asymmetry, and for Snowden it is the reason humans cannot be treated as just one more kind of actant.

The example. The buildings of Calder Vale act on the tenants: cold walls, wet windows, spoiled clothes. The tenants and staff describe the buildings: "lifestyle", "structural", "hard to heat". Those descriptions decide which blocks are insulated and which are left. The buildings, for their part, describe nobody. Only one side of the relation has a say about the other.

How CAS differs. CAS theory has no such principle. Agents and environment co-evolve symmetrically, each adapting to the other. Nothing in the formalism marks out one class of agents as the describers of the rest.

The common misreading. "Humans are in charge of nature." The principle is about description and its consequences. It makes no claim about control, and a city that describes its river badly is flooded all the same.

First interrogation. This is the principle on which the whole disputation turns, so the argument is set out in steps.

  1. The asymmetry principle says that humans tell stories about other actants, and that these stories change the constraints.
  2. Anthro-complexity is itself a set of stories about human systems: that they are complex, that they have dispositions, that they should be probed and not planned.
  3. Its instruments are stories in a strict sense. A Cynefin map tells an organisation what kinds of situation it has. A set of signifiers tells narrators what dimensions their experience has.
  4. By the principle, these stories change the constraints of the systems they describe.
  5. So anthro-complexity's descriptions are among the forces acting in any system where they are used.
  6. The theory's own account of method does not treat them so. It treats them as ways of reading the system.

The conclusion is that the asymmetry principle, taken at its word, applies to the theory that states it. That is Thesis II. There is a second point. The principle is stated between humans and non-humans. Among humans there is an asymmetry too: some describe and others are described. The surveyor describes the tenant in the file. The tenant's description of the surveyor is not filed anywhere. Between river and city the asymmetry is a fact of nature. Between surveyor and tenant it is a fact of position. The principle has the right shape for a theory of power, and it is used only at the human boundary. Holds It holds so well that it reaches its author. The exemption fails

Element 14 slowly: entanglement without a flat ontology

In plain words. Anthro-complexity agrees that knowledge and action are entangled with things: tools, places, records, other people. Much of what an organisation knows exists only when the right people, artefacts and context come together. It adds a limit. It does not accept the stronger claims made by Karen Barad, for whom there are no separate things before the relation, or by flat readings of Latour, for whom all actants are of equal standing. People are entangled, and they remain distinct people.

The example. How to deal with damp at Fieldhead is not in the caretaker's head, nor in the operatives', nor in the meter. It is in all of them working together on that estate. Move the caretaker and the knowledge does not travel with her in full. That is entanglement. But the caretaker is still a particular woman who can decide to stop making the calls, and the meter cannot. That is the limit.

How CAS differs. CAS models keep agents and environment as separate defined entities that interact. There is interaction without entanglement in this sense. Knowledge, where it appears at all, is a property of an agent.

The common misreading. "Entangled means everything is connected to everything." The claim is specific: certain capacities exist only in certain combinations.

First interrogation. The disputation first read "the people remain distinct" as one more sign of the possessive self, and it has corrected itself. The correction is in the addition to Chapter 2 and draws on Margaret Archer. Persons are formed in relations, and once formed they have real powers of their own, including the power to reflect and to refuse. So Snowden is right against Barad. The remaining question is whether anthro-complexity has an account of how the distinct person is formed, or simply starts with one. A theory can deny that persons dissolve into relations and still owe an account of how relations make persons. Holds against Barad Owes an account of formation

Element 15 slowly: against the programme

In plain words. The mind is not software and the brain is not a computer running it. People are embodied, they act in the world with tools and with one another, and what they know cannot be separated from how they live. So an organisation is not an information-processing machine either, and methods that treat people as processors of inputs will miss what matters.

The example. Calder Vale's customer system treats a damp report as an input to be classified, routed and closed. It is a programme, and tenants are data sources. What it cannot hold is that Mrs Okafor's report was also an act of trust that cost her something to make, and that how it is answered will decide whether she, and the four neighbours who are watching, ever report again.

How CAS differs. This is the sharpest contrast in the whole comparison. CAS theory is computational at root. Its central method is to simulate agents as rule-executing programmes, and John Holland's founding work treats adaptation as a kind of computation. Anthro-complexity denies that this captures human beings.

The common misreading. "So computers have no place." Snowden's own instruments are software. The claim is about what people are. It does not forbid tools.

First interrogation. The denial of the computational picture of the person is accepted in full. The question is whether the method chain honours it. In that chain, a person's account of an experience is turned into positions on a set of designed scales, then aggregated with thousands of others, then displayed as patterns, and increasingly read with machine assistance. Each step is defensible. Taken together they convert human noticing into machine-readable signal for a centre. An ontology that says a person is not an input sits beside a practice in which the person's contribution arrives as one. Hutchins's study of the ship's bridge offers the contrast: there, the people who feed the chart also see it and correct one another. The ontology holds The practice is in tension with it This is Thesis VI.

What the four come to, and why the name is earned

The four human elements together explain why Snowden coined a new word. Complexity science, as it came from Santa Fe, offered organisations a picture drawn from ants, birds and markets of simple traders. It was a real advance on the machine picture. It also carried a hidden premise: that the agents do not redescribe the system they are in. Remove that premise and much of the method must change. Models lose their authority, because the modelled will respond to the model. Prediction gives way to disposition. Design gives way to probing. The human cluster is the reason for all of that, and it deserves a name of its own.

It is also where the disputation's theses gather. The three marks are true and are stated as possessions (Thesis I). The asymmetry principle is true and is not applied to its own author or between humans (Theses II and V). The anti-computational ontology is true and sits beside a computational practice (Thesis VI). In each case the criticism has the same form: the claim is right, and its consequences have not been followed all the way. Calder Vale shows why this matters in practice. A practitioner who believes that people merely have stories will collect the stories. A practitioner who believes that she is now part of the story will ask who sees the collection, who chose its categories, and what the people who told the stories will make of having been asked.

THREE MARKS, ALL TRUE —

AND ALL THREE FORMED BETWEEN PEOPLE —

SOME OF WHOM HOLD THE PEN

Addition: Cross-Lights From Book Nine

The trialectic. The chapter reads the three marks of the human as possessions. Five months before the tree post, Snowden had written that in the complex domain the capacity for change is held by neither practitioner nor system and exists only in their coupling. That denies possession at the level of practice. Chapter 48 credits it, and Chapter 50 asks whether the instruments embody it.

Tied to the teller. The post on Luhmann grounds self-signification in the claim that a narrative stays tied to the person who told it. That is the possessive grammar in its most attractive form, and the right it grounds is accepted. Chapter 52 adds that a story is also tied to the listener it was told to.

The asymmetry among humans. The same post names the people with the power to filter and cites Miranda Fricker on epistemic injustice. The asymmetry between describers and described, which this chapter said was missing, is now acknowledged.

Chapter 5

Method and Practice: Constraints, Sensing, the Pelagian Stance

9,042 words, about 39 minutes of reading

From What Is to What Is Done

The fourth cluster turns from ontology to practice: from what a complex human system is to how one can know it and act in it. The turn is where anthro-complexity separates most clearly from complexity science as an academic discipline. A physicist who studies dissipative structures need not ask how to intervene in them. An anthro-complexity practitioner has to, because the theory was developed in and for practice. Cynefin began as a sense-making framework for knowledge management at IBM. It grew into a consultancy, a training programme, a set of instruments and a professional network. The theory and the practice grew up together and cannot be separated.

This chapter assembles seven practical elements. Two are epistemic: the pragmatism that grounds anthro-complexity's way of knowing (Element 16), and the triadic structure through which it thinks (Element 17). Five concern intervention: behaviour as the shadow of conditions (Element 18), distributed sensing (Element 19), the Pelagian stance (Element 20), context-dependent decision models (Element 21), and resistance to commodification (Element 22).

Element 16: Peircean Pragmatism

The claim

Both Ing and Snowden trace their roots to American pragmatism, but they draw on different strands for different ends. Ing's pragmatism, Snowden says, comes down to two lineages. One runs from William James through Ralph Barton Perry to Pepper's root metaphors. The other runs from Edgar Singer to Churchman and Ackoff, where inquiry is coupled with values and with progress towards ideals. It is a pragmatism of worldviews and of planning. Snowden's is Peircean: a pragmatism of signs, abduction and habit, anchored, as the original Metaphysical Club was, in a commitment to the natural sciences.

The paper lists Peirce with James and Dewey as a contextualist. Peirce, Snowden says, would have resisted that, since he renamed his own position to distance himself from the direction James had taken it. (The historical detail is that Peirce, around 1905, coined the deliberately ugly word "pragmaticism" for his own doctrine, saying it was ugly enough to be safe from kidnappers. Snowden's post says "renamed his own position pragmatism", which reads as a slip for "pragmaticism". The point stands either way.) For Ing, pragmatism is the starting point of a lineage that the programme moves beyond, towards processes as rhythms. For anthro-complexity, Peirce is still live, as the earlier posts in the series on semiotics and the abductive warrant set out.

Three Peircean commitments

Signs. Peirce's semiotics is triadic. A sign is something that stands to somebody for something in some respect. It relates a representamen (the sign-vehicle), an object (what the sign stands for) and an interpretant (the effect of the sign on an interpreter, itself another sign). Meaning is not a two-place relation between a word and a thing. It is a three-place relation that includes interpretation, and interpretation is endless, since each interpretant is a sign that calls out a further interpretant. Semiosis is a process, not a mapping.

Abduction. Peirce distinguished three forms of inference. Deduction derives what must follow from premises. Induction generalises from cases. Abduction, which Peirce also called retroduction or hypothesis, infers the best explanation of a surprising fact: the surprising fact C is observed; but if A were true, C would be a matter of course; hence there is reason to suspect that A is true. Abduction is the only form of inference that introduces new ideas. It is fallible, risky and creative, and Peirce held that it was the beginning of all inquiry.

Habit. For Peirce, a belief is a habit of action, and the meaning of a concept is the sum of the practical consequences we would expect if it were true. Habits are also, in Peirce's later cosmology, the way the universe itself acquires regularity. Laws of nature are habits that the universe has taken, emerging from an original chance. Habit is thus Peirce's term for how patterns stabilise and how they change.

The abductive warrant

The "abductive warrant" Snowden refers to is the earlier part of the series that develops abduction as the logic proper to complex systems. In ordered systems, deduction from known causes or induction from repeated cases can guide action. In complex systems, where cause and effect are coherent only in retrospect and every situation is in some respects novel, the only available inference is abduction: the formation of plausible hypotheses about what is going on, tested by acting on them in small, safe ways. Safe-to-fail probes are abductions put into practice. Each probe embodies a hypothesis, and the response of the system tests it.

The function

Element 16 supplies anthro-complexity's epistemology. It explains why narrative matters: narratives are signs, and their meaning lies in their interpretants, the effects they have on those who hear them. It explains why probes matter: they are abductions. It explains why habit and pattern matter: they are how complex systems stabilise and change. And it ties anthro-complexity to the natural sciences through Peirce's own commitment to scientific method as the model of inquiry, keeping it from becoming one more relativist account of multiple perspectives.

The flag

Peirce's semiotics is triadic, and its third term, the interpretant, is the effect of a sign on an interpreter. Peirce's account of inquiry is social: truth is what the community of inquirers would agree on in the long run, and the individual inquirer is disciplined by the community. Both commitments place interpretation between people, in a community, not inside individuals. Chapter 32 will ask where the interpretant sits in anthro-complexity's instruments. When a narrator signifies a story on a triad, the narrator produces an interpretant. When an analyst reads the pattern across a thousand signified stories, the analyst produces another. Which is the interpretant of the system? Peirce would say both, and that the question of which governs action is a question about the community of inquiry: who belongs to it, who speaks for it, who decides. A Peircean anthro-complexity would have to make that question explicit.

Element 17: Trialectic, Not Dyad

The claim

Ing's programme is dyadic. At its core are yin and yang as inseparable, mutually transforming modes; qi in concentrating and dissipating modes; waxing and waning; with no fixed superiority of one over the other. Snowden credits the insistence that dyadic is not dualistic, that the poles are context-dependent and complementary rather than ranked, as well made and a useful corrective to much Western thinking. But a correction is not a change:

In Snowden's paraphrased argument, people handed a pair will turn it into a dichotomy: a dialectic stiffens into an either/or, sides form, and a winner is expected. He treats this as a habit, not a misunderstanding, so that clearer explanation cannot cure it.

This is an application of Element 18: if you want people to stop dichotomising, change the structure they think with instead of exhorting them to think differently. Hence Snowden's move from two to three. The trialectics series, visualised as the triquetra, came out of that conviction. Three elements in tension cannot be reduced to two sides. Each pair is mediated by the third, so the tension does not resolve into oscillation, balance or synthesis, and the system stays open and generative.

The debt to Peirce

The debt is to Peirce, for whom thirdness, the category of mediation, habit and law, cannot be reduced: it cannot be assembled from relations between pairs, and without it you cannot account for meaning or for how patterns stabilise and change. Peirce's three categories are firstness (quality, possibility, the thing in itself), secondness (reaction, resistance, brute fact, the clash of one thing with another) and thirdness (mediation, law, habit, the general). Peirce's "reduction thesis" holds that all relations can be built from monadic, dyadic and triadic relations, but triadic relations cannot be built from dyadic ones. You cannot get a sign, which needs a sign-vehicle, an object and an interpretant together, out of any combination of two-place relations.

The trigram objection

Ing's paper brings threes in through the trigrams of the Yijing: three lines, each yin or yang, giving eight combinations, which it maps to eight seasons. Snowden's objection is that this is two raised to the power of three, a combinatorics of binaries, not thirdness in Peirce's sense. A dyadic frame, however rich, risks treating complexity as a rhythm between poles when much of what you are dealing with is entanglement and mediation. He adds that he holds this loosely, and would like to know whether the Chinese sources have a way of handling mediation that he is missing.

The function

Element 17 does two jobs. It gives anthro-complexity a structure of thought that resists the dichotomies of management culture: centralise or decentralise, plan or emerge, control or empower. In their place it offers three-way tensions in which no two poles can be set against each other without the third mediating. And it gives SenseMaker its characteristic instrument, the triad: a triangle with a label at each vertex, on which a narrator places a dot to signify how far her story reflects each of the three. Triads, Snowden argues, are harder to game than scales, because there is no obvious "right" answer, and they elicit more nuanced signification than binary or Likert questions.

The flag

Two flags, examined in Chapters 25 and 38. First, the triad as an instrument is a geometric object, a two-dimensional simplex whose three coordinates sum to one, and its geometry has consequences that the triad as a concept does not. A simplex builds in a trade-off: more of one vertex means less of the others. The Compass and the Loom observed that on a barycentric triad, strong-in-all and weak-in-all plot at the same centre. The instrument meant to escape dichotomy may build a different kind of opposition into its geometry. Second, Snowden's question about the Chinese sources is a real question with a real answer. The Chinese tradition does handle mediation, notably through the triad 天地人 (heaven, earth, humanity) and through the concept of 中 (the centre, the mean) as a dynamic and not a static position. Chapter 38 will take up his invitation.

Element 18: Behaviour Is the Shadow of Conditions

The claim

The lecture uses Ackoff's definition of behaviour: a system change which initiates other events. Snowden sets beside it his own recent claim that behaviour is the shadow of conditions. If you want different behaviour, you change the constraints and conditions people act within, not the behaviour directly.

The argument

The claim is an inversion of the behavioural-change tradition, which targets behaviour directly through training, incentives, nudges, exhortation and culture programmes. On the shadow view, behaviour is an effect of conditions, as a shadow is an effect of an object and a light. You cannot change a shadow by working on the shadow. You change the object, the light, or the surface the shadow falls on. Trying to change behaviour directly is like painting the shadow. It may work for a while, but as soon as the paint wears off, the shadow returns to the shape the conditions give it.

The claim has strong support in the social-psychological literature. Lewin's own field theory held that behaviour is a function of person and environment. The situationist tradition, from Milgram and Zimbardo to the replication debates, shows how powerfully context shapes conduct. The "fundamental attribution error", the tendency to explain others' behaviour by their dispositions and not their situations, is one of the most robust findings in social psychology. Snowden's claim takes the situationist lesson to its practical conclusion: work on the situation.

The function

Element 18 unifies anthro-complexity's practice. Estuarine mapping maps constraints, the conditions behaviour is the shadow of. Probes change conditions and watch what behaviour follows. Triads, as Element 17 showed, change the structure people think with instead of exhorting them to think differently. The practitioner never targets behaviour. She targets conditions, and lets behaviour follow.

The flag

The shadow metaphor has an object, a light and a surface, and someone who moves them. It does not say who that someone is. In a human system the conditions people act within include the conditions set by other people: managers, consultants, regulators, designers of instruments. Those people are also acting within conditions, and their behaviour is also a shadow. Chapter 23 will ask about the gradient of power across which conditions are set. If behaviour is the shadow of conditions, the question of who sets the conditions becomes the political question of complexity practice. Chapter 20 will add that the practitioner who changes conditions is also a condition. Her presence, her instruments and her framing are part of the light.

Element 19: Distributed Sensing

The claim

In the eurhythmia section Snowden raises his second concern with Ing's programme: who does the reading. The physician takes the pulse; the body does not. Anthro-complexity tries to distribute sensing through methods such as SenseMaker® and estuarine mapping, so people within the system see the patterns themselves and act on what is already there. Self-healing needs more than a light touch from the physician. It needs sensing capacity within the system.

SenseMaker

SenseMaker is the most developed instrument of anthro-complexity. In outline, a population is invited to share short narratives, micro-narratives, about their experience, prompted by an open question. Each narrator then self-signifies the story: she places it on a set of triads, dyads (sliders between two poles) and other signifiers, indicating what it means to her. The signified narratives are aggregated, and patterns across the population become visible as clusters, distributions and outliers on the signifiers. The patterns can be fed back to the population, which can then explore its own narratives, and the next question becomes, in Snowden's phrase, how do we get more stories like these and fewer like those?

The key design choice is self-signification. In conventional narrative research, an analyst interprets the stories. In SenseMaker, the narrator interprets her own story by placing it on the signifiers, so interpretation stays with the person who had the experience. The analyst sees patterns in self-interpretations, not in raw stories. That is the basis of SenseMaker's claim to disintermediate: to remove the expert who stands between the people and their own meaning.

Estuarine mapping

Estuarine mapping, which began life as affordance mapping, maps the constraints of a system on two dimensions: the energy needed to change a constraint and the time it would take. Participants identify constraints in their situation and place them on the map. The map then shows which constraints are cheap and quick to shift, which are costly and slow, and which are beyond the participants' power. It also shows "counterfactuals", what could be done if certain constraints were shifted. The practice is participatory. People inside the system make the map, and the map shows them where they have room to act.

The function

Element 19 is anthro-complexity's answer to the problem of the outside reader, the problem it raises against Ing's physician. Sensing is distributed among the people in the system. Interpretation stays with narrators. Maps are made by participants. The patterns are fed back. In principle, no expert stands outside the system reading its pulse.

The flag

This element carries more weight in the argument of this book than any other, because it is where anthro-complexity's critique of the outside reader is most fully realised, and where The Compass and the Loom found its limits most clearly. The strength is real: self-signification lets narrators place their own stories, keeping stories linked to patterns. The costs are also real: triads act as search templates and variety reducers (Ashby); the designer's categories determine what is visible; and conversation, later reinterpretation and the narrator's own meaning are lost. Chapter 23 will develop the costs into the claim that distributed sensing distributes the act of sensing but not the frame. The body does take its own pulse, but on a chart the physician printed, and the reading goes back to the physician. Self-healing needs sensing capacity within the system, as Snowden says. Sensing capacity is not interpretive authority.

Element 20: Pelagian, Not Augustinian

The claim

The theological language is Snowden's. Most of what is sold as systems change, he says, carries the Augustinian assumption that the system lacks something an external intermediary must supply. In his reading of eurhythmia he had first worried that the medical language placed the practitioner in the Augustinian position, supplying what the system lacks. He revised that reading when he found that Ing's paper prefers self-healing through internal adjustment to invasive external intervention, aiming to help the body return to its own condition, which is closer to the Pelagian position than he had assumed.

The theological background

The historical controversy, in the early fifth century, was between Augustine of Hippo and the British monk Pelagius over grace and free will. Augustine held that human nature, after the Fall, is so damaged that humans cannot will the good without divine grace. Salvation comes from outside, as a gift that human effort cannot earn. Pelagius held, or was held to have held (the historical Pelagius is hard to recover from his opponents' polemics), that humans retain the natural capacity to choose the good, and that grace assists but does not replace human effort. The Church sided with Augustine. "Pelagian" became a heresy and later a term of abuse.

Snowden's use of the terms is analogical, and it is apt. An Augustinian intervention presumes that the system cannot help itself and must receive something from outside: a consultant's diagnosis, a framework, a transformation programme. A Pelagian intervention presumes that the system has the capacity to heal itself, and that the intervener's role is to help it use that capacity. Anthro-complexity is Pelagian. It works with what is already there, amplifying what is working and dampening what is not, and it distributes sensing so that the system can see itself.

The function

Element 20 is the ethical posture of anthro-complexity's practice. It underlies the critique of the systems-change industry, the preference for working with existing patterns over importing new ones, the emphasis on distributed sensing, and the refusal of the expert position. It also gives the practice a kind of humility. The practitioner does not know better than the system what the system needs.

The flag

The Pelagian posture has a commercial difficulty that Snowden is too honest to deny and that The Compass and the Loom made explicit: a consultancy sells something. If the system has everything it needs, what is the client paying for? The Pelagian answer is that the client pays for help in using what it has: methods, facilitation, instruments. But an instrument is something the system did not have before, supplied from outside. SenseMaker is not a capacity the system already possesses. It is a technology the system licenses. Chapter 17 will ask whether anthro-complexity's instruments are Pelagian in content and Augustinian in form: they help the system see itself, but the seeing needs something only the vendor can supply. The question is not one of hypocrisy. Every practice that is also a business faces it. But it has to be asked, because it bears directly on who does the reading.

Element 21: The Sequence Depends on the System

The claim

In his institutional story about IBM, Snowden recalls that Stephen Haeckel's sense and respond was much vaunted there. An encounter with Haeckel, mediated by Bruce Harold, stimulated the decision models in Cynefin:

  • sense, categorise, respond (clear);
  • sense, analyse, respond (complicated);
  • probe, sense, respond (complex);
  • act, sense, respond (chaotic).

Where sense and respond offered one sequence for an adaptive enterprise, Cynefin made the sequence depend on the nature of the system you were in.

The function

Element 21 is the oldest element of anthro-complexity and in some ways the founding one. It is the claim that there is no single right way to decide, because the right way depends on what kind of situation one is in. It turns the ontological distinction of Element 6 (ordered/complex) into a practical one. It also introduces the move that makes Cynefin distinctive among frameworks: before deciding what to do, decide what kind of situation you are in, and therefore how to decide. Cynefin is a framework for choosing frameworks.

The flag

The prior question, what kind of situation is this?, is itself a decision, and Cynefin's own logic says the right way to make it depends on what kind of situation one is in. Snowden has addressed this regress in many ways: through the aporetic or confused domain at the centre of the framework, through collective sense-making exercises in which groups place their own situations, and through the insistence that the boundaries between domains are fuzzy and dynamic. But the regress has a reflexive side that matters for this book. Categorising a human situation as complex is a gesture that changes the situation. It licenses probes, legitimises experiment, and relieves managers of the demand for a plan. Categorising it as complicated licenses analysis, experts and plans. The categorisation is not a neutral reading of a fact. It is an intervention with consequences for who has authority. Chapter 20 will take this up as an instance of the asymmetry principle applied to itself.

Element 22: Resistance to Commodification

The claim

The last element is less a doctrine than a disposition that runs through the November post. A careful piece of original work is simplified, packaged and sold until what circulates bears little relation to what was written. Lewin is the example. His careful field theory became the three-step model of change management, with a refreezing stage he never wrote. Amy Edmondson's work on psychological safety is a current case: sound in its original form, and now, Snowden says, well on its way to becoming a commodity. Some original authors are abused in this process; others are complicit. Lewin died in 1947, three years before refreezing first appeared, so he at least belongs among the former.

The function

Element 22 is anthro-complexity's immune response. It names the danger every idea that enters organisational life faces: being packaged into a method that can be taught and rolled out, losing its subtlety in the process, and then being sold as the original. The institutional story about IBM is the same response at a larger scale. Systems thinking was more at home in large organisations than complexity science because it offered structure, and complexity has so often been absorbed into systems thinking because institutions find structure easier to host.

The flag

The flag here is the simplest and the most delicate. Element 22 diagnoses a process, the simplification, packaging and sale of ideas, that acts on every idea that enters organisational life, anthro-complexity included. Snowden's distinction between authors who are abused and authors who are complicit is the right one, and it raises the question every author with a consultancy must face: which am I? The Compass and the Loom did not accuse Snowden of complicity. It traced a drift, ontology to epistemology to energetics to decision to product, in which each step moves closer to the client's question. Chapter 17 will take up that drift as the positional reason for treating human complexity differently. It will argue that the drift is not a moral failing but a structural pressure, the same one Element 22 diagnoses in others, and that anthro-complexity is not exempt from it.

The Practical Cluster as a Whole

Element Claim Instrument or practice Flag
16 Peircean pragmatism: signs, abduction, habit Abductive warrant; safe-to-fail probes Where does the interpretant sit?
17 Trialectic, not dyad Triquetra; SenseMaker triads The triad is a simplex
18 Behaviour is the shadow of conditions Change constraints, not behaviour Who sets the conditions?
19 Distributed sensing SenseMaker; estuarine mapping Distributes the act, not the frame
20 Pelagian, not Augustinian Work with what is there Pelagian content, Augustinian form?
21 Sequence depends on system Cynefin decision models Categorising is intervening
22 Resistance to commodification Critique of Lewin, Edmondson, IBM Does the immune response reach home?

The Twenty-Two Elements

With this chapter the reconstruction of anthro-complexity's elements is complete. There are twenty-two in four clusters: six ontological, five causal and temporal, four human, and seven practical. The next chapter traces their genealogy. After that, Book Two places anthro-complexity among its neighbours, and Book Three asks why it treats human complexity differently.

It is worth pausing to say what has been shown. Anthro-complexity is not a loose collection of slogans. It is a body of thought with a coherent ontology, a theory of causality and time, an account of what is distinctive about humans, and a practice that follows from these. Its sources are serious: Peirce, Juarrero, Jullien, Rosen, Prigogine, Scott, Aristotle, Polanyi, Latour. Its claims are, for the most part, true. Its practical consequences are, for the most part, wise. Whatever the following chapters say against it, they say it against something worth arguing with, and that is the condition of a disputation worth having.

TWENTY-TWO ELEMENTS —

FOUR CLUSTERS —

ONE QUESTION LEFT OPEN: WHO DOES THE READING?

Addition: The Seven Practical Elements Taken Slowly

This is the last of the four slow readings. Calder Vale Homes and Asha, its director of housing, are again the case. The seven practical elements say what anthro-complexity asks a practitioner to do. A second addition follows this one and walks through the main methods step by step, because the methods are where most newcomers lose the thread.

Element 16 slowly: Peircean pragmatism

In plain words. Pragmatism holds that the meaning of an idea lies in the practical difference it makes, and that inquiry is how a community settles what to believe. Snowden's pragmatism is that of Charles Sanders Peirce in particular. Three features matter. Reasoning in new situations is abductive: it leaps to the hypothesis that would best explain a surprising fact, and then tests it. Meaning lives in signs that someone interprets. And patterns in the world stabilise as habits, which can be changed.

The example. Asha notices something surprising: complaints from one tower fell sharply last winter while its neighbour's rose. Deduction cannot help, because she has no rule to deduce from. Induction would need many towers and many winters. Abduction says: what, if it were true, would make this unsurprising? Perhaps a new caretaker. Perhaps the residents' association stopped trusting the complaints line and stopped using it. The second hypothesis is the worrying one, and it is cheap to test by knocking on ten doors.

How CAS differs. CAS work is mostly deductive within a model and inductive across simulation runs. Abduction, the forming of a new explanatory idea, happens in the modeller's head before the model is built and is not part of the method.

First interrogation. Peirce held that truth is what a community of inquirers would agree on in the long run, and that a sign's meaning is completed in its interpretant, the effect it has on someone. If one takes Peirce whole, the tenants are inquirers into their own situation and have a claim to see and argue with the hypotheses formed about them. In the usual arrangement they supply observations, and the abduction is done elsewhere. Holds, half-applied Chapter 32 develops this.

Element 17 slowly: trialectic, not dyad

In plain words. Much thinking works in pairs of opposites: order and chaos, structure and agency, yin and yang. Snowden argues that pairs trap thought, since with two terms all you can do is choose, balance or oscillate. Three terms give a field with an interior, in which a position can be a mixture and movement can be in more than one direction. So his instruments prefer triangles to lines, and his concepts come in threes.

The example. Ask a Calder Vale tenant whether the response to her report was "fast or slow" and you learn almost nothing. Ask her to place a mark in a triangle whose corners are "I was listened to", "the problem was fixed" and "I was treated with respect", and she must think about how those three relate in her experience. A mark near "fixed" and far from "listened to" says something a satisfaction score cannot.

How CAS differs. CAS theory has no stake in this. Its variables are whatever the model needs.

The common misreading. "Two is bad and three is good." Snowden's claim concerns what a form of representation makes thinkable.

First interrogation. Two objections. A triangle of this kind is, mathematically, a simplex: the three values must sum to a constant. To move toward one corner is to move away from the others. The tenant who felt listened to and respected and had the problem fixed cannot say so, and neither can the tenant who got none of them. Two independent scales would allow both. So the triad builds a trade-off into the answer, which is a dichotomising move with three ends. Second, Peirce's third term is the interpretant, the one for whom the sign means something. In the instrument, the patterns are interpreted by analysts and sponsors. The narrator gives the sign and the centre supplies the third. The aim holds, the geometry does not deliver it This is Thesis VII, and Chapter 38 answers the charge against yin and yang.

Element 18 slowly: behaviour is the shadow of conditions

In plain words. People mostly do what their circumstances make easy, rewarded or necessary. So if you want different behaviour, do not lecture, train or incentivise individuals. Change the conditions, and behaviour will follow as a shadow follows the thing that casts it.

The example. Calder Vale has twice sent operatives on a course called "Getting to the Root Cause". Nothing changed, because the day after the course each operative still had nine jobs, a code that paid only for washing and a van without a meter. Changing those conditions would do what the course could not.

How CAS differs. A CAS modeller would agree in form: change the environment or the pay-offs and agent behaviour changes. The difference is that anthro-complexity counts stories, identities and rituals among the conditions.

The common misreading. "People have no agency." The element is advice to the person with power over conditions. It does not say individuals cannot choose.

First interrogation. The image has a light source and a wall, and it leaves a question: who arranges the lamp? "Change the conditions" is addressed to someone who can. For the operative, the conditions are simply given. The element is true, and it is a statement about power that does not use the word. There is a second problem. Among people, conditions do not only cause behaviour. People interpret their conditions, and the interpretation is part of the condition. A meter in the van means one thing if operatives asked for it and another if it arrives with a memo about monitoring. Holds, and assumes a position it does not name

Element 19 slowly: distributed sensing

In plain words. No one at the centre can see a complex system. The people spread through it each see a little, in real time. So the way to know the system is to gather many small first-hand accounts continuously, let the people who give them say what they mean, and look at the patterns across them. This is what SenseMaker does, and estuarine mapping does something similar for constraints.

The example. Instead of an annual satisfaction survey, Asha invites every tenant, operative, caretaker and officer to record short accounts of real moments: "tell us about a recent time when something to do with damp or repairs went well or badly". Each teller then places the account on a few triangles and scales. After two months there are nine hundred accounts. Looking at them, Asha sees a cluster she had no category for: stories from operatives about being let in and then being unable to do what was obviously needed.

How CAS differs. CAS research observes agents from outside or simulates them. It does not ask the agents what their behaviour means. Self-interpretation by the agents is a distinctly anthro-complexity move.

The common misreading. "It is a survey with stories." The key difference is who interprets. In a survey the analyst codes the answers. Here the teller does.

First interrogation. The strength is real. The questions are these. Who wrote the triangles? The teller interprets, but within dimensions someone else chose. Who sees the nine hundred accounts and their patterns? Usually Asha and her team. What do tellers learn from having told? Usually nothing, unless a return path is built. And what happens the second time, when people know what the first round led to? An operative whose candid story was followed by a tighter inspection regime tells a different story next time. The sensing is distributed. The seeing is not. Holds as sensing, fails as distributed cognition Theses VI and VIII.

Element 20 slowly: Pelagian, not Augustinian

In plain words. Snowden borrows a fifth-century dispute. Augustine taught that human nature is so damaged that people can do no good without grace given from outside. Pelagius, a British monk, taught that people have a real capacity to act well. Applied to organisations: the Augustinian consultant treats people as the problem and brings salvation in the form of a framework, a culture programme or a transformation. The Pelagian starts from what people are already able and willing to do, and removes obstacles.

The example. The Augustinian reading of Calder Vale is that staff have a "poor culture" and tenants have "poor behaviours", and that both need a change programme. The Pelagian reading is that most operatives would prefer to fix the cause and most tenants would prefer to report, and that the job is to find out what is stopping them.

How CAS differs. CAS theory has no view about human nature. Its agents are neither fallen nor capable. They follow rules.

First interrogation. The stance is admirable, and the disputation shares it. One observation. A Pelagian stance toward the people in the system should include confidence that they can read their own patterns and decide what to do about them. An arrangement in which patterns are read by experts and sponsors, and then turned into interventions for the people, has a trace of grace descending from above. The Pelagian stance, followed through, argues for returning the map to those who supplied it. Holds And it supports the reconstruction in Chapter 45.

Element 21 slowly: the sequence depends on the system

In plain words. There is no single correct order of analysing, deciding and acting. In an ordered situation you gather facts, decide, then act. In a complex one you act first in small ways, see what happens, then respond. In a chaotic one you act at once to stabilise. The mistake is to use one sequence everywhere, and the commonest mistake is to analyse a complex situation at length before doing anything.

The example. For the leaking gutter at number 30: inspect, specify, repair. For the collapse of reporting at Fieldhead: try three or four small things at once and watch. For the week of the newspaper story: get someone senior to the family's door today, and investigate afterwards.

How CAS differs. CAS theory describes systems. It offers no decision sequence to a practitioner.

First interrogation. In the complex domain the sequence begins with a probe. A probe in an estuary is a depth sounding, and the water is indifferent to it. A probe among people is an act that those people interpret. "They're trying something at Fieldhead" is news, and it changes what other estates expect and demand. So the sequence is less like probe, sense, respond, and more like gesture, response, further gesture, which is Mead's account of any conversation. Holds, with the probe redescribed

Element 22 slowly: resistance to commodification

In plain words. Good methods get turned into products: certifications, canvases, maturity models, fixed recipes. Once that happens they are applied without judgement and stop working. Snowden resists this for his own work and criticises it in others. He keeps methods open, changes them, and insists they need understanding and not just procedure.

The example. Asha is offered a "Complexity-Ready Housing Provider" accreditation with a five-level maturity model. It would be easy to buy, and it would produce a certificate for the board. It would also turn everything in this chapter into a checklist.

First interrogation. The resistance is sincere and the disputation credits it. The tension is structural. The instruments are licensed, the practitioners are trained and accredited, and the company must be paid. None of that is discreditable. But a theory which says that abstractions change constraints should say how its own commercial form shapes its methods: which features are there because they scale and sell, and which because they serve the people in the system. Chapter 17 treats this as the positional reason for the human distinction. Holds as intent, unexamined as position

What the seven come to

The practical cluster tells Asha how to proceed. Reason by informed guess and test. Think in fields, not opposites. Change conditions, not people. Gather many first-hand accounts and let tellers interpret them. Assume people are capable. Match the order of acting and thinking to the kind of situation. And refuse to buy a recipe.

Across the four slow readings, the twenty-two interrogations have returned to a small number of questions. Who classifies the situation? Who is presumed, and how were they formed? Who is inside the reading? Whose position sets the slope? Who chose the categories? Who sees the result? What do people learn from being asked, and what do they do differently the next time? Anthro-complexity gives excellent answers to the question what is this system like. Those seven further questions all concern the practitioner and the instrument as parts of the system. A theory whose defining claim is that humans redescribe their world, and thereby change it, is the theory best equipped to answer them. The rest of the disputation asks it to.

TWENTY-TWO ELEMENTS, TAKEN SLOWLY —

SEVEN QUESTIONS LEFT OVER —

ALL SEVEN ABOUT THE ONE WHO ASKS

Addition: The Methods Step by Step

The elements are ideas. The methods are what a practitioner actually does on a Tuesday. Critics and enthusiasts alike often discuss anthro-complexity without being able to say what happens in the room, and that makes both the claims and the objections hard to grasp. This addition therefore walks through the six methods the disputation refers to most often. For each it gives the purpose, the steps as they are usually presented, what the method is good at, and the exact point at which the theses of Book Four take hold. The descriptions follow the general published shape of each method. Practice varies and the methods continue to be revised, which is to their credit.

Calder Vale Homes remains the case.

Method 1: Cynefin, the sorting of situations

Purpose. To help a group see that its problems are of different kinds and need different handling.

The picture. Five domains. On the ordered side are clear, where cause and effect are obvious to any reasonable person and the response is best practice, and complicated, where cause and effect can be found by analysis or expertise and there may be several good answers. On the unordered side are complex, where cause and effect can be seen only afterwards and patterns emerge from interaction, and chaotic, where there is no usable pattern and the need is to act at once to stabilise. In the middle is confused, sometimes called aporetic: the state of not knowing which domain you are in. The boundaries are of unequal kinds. Between clear and chaotic there is a cliff: complacent order collapses suddenly into crisis and is very expensive to recover. Other boundaries are zones one can cross on purpose, called liminal.

The decision sequences. Clear: sense, categorise, respond. Complicated: sense, analyse, respond. Complex: probe, sense, respond. Chaotic: act, sense, respond.

The steps, in the form called contextualisation.

  1. Collect a large number of real items from the people involved: issues, decisions, incidents. At Calder Vale: "the gutter at number 30", "why Fieldhead stopped reporting", "the contract break clause", "the newspaper week".
  2. Without showing the framework, ask the group to place the items relative to one another according to how knowable and predictable each is.
  3. Only then draw the domain boundaries, through the clusters the group has made.
  4. Discuss the items near boundaries and the items people disagreed about. Those disagreements are the most informative part.
  5. Decide, domain by domain, what kind of response is fitting, and consider which items should be moved: for example, a complex issue that has stabilised enough to be handed to experts.

What it is good at. It stops a group from using one method for everything. It legitimises experiment where experiment is needed and expertise where expertise is. It gives people a shared language for the feeling that "this one is different".

Where the theses take hold. The framework is described as sense-making: the data come before the frame. But the items are collected from those invited, placed by those in the room, and the boundaries are drawn by a facilitator. The resulting map then tells the organisation what it may hold people to account for. An item in "complex" is an item for which nobody could have known. Theses II and V ask who was in the room, and what the classification does once it leaves it.

Method 2: constraint mapping

Purpose. To see what is actually shaping behaviour, so that one can change the shaping and stop exhorting the people.

The vocabulary. A constraint is anything that makes some actions more likely and others less. Governing constraints limit: rules, walls, budgets, fixed procedures. They give predictability and are appropriate to order. Enabling constraints connect and make new things possible: a shared rhythm of meetings, a heuristic, a link between two people who did not previously talk. They are what complex systems grow on. Snowden's typology has further distinctions, for instance between constraints that contain from outside like a shell and those that hold together from within like a skeleton, between rigid and flexible, and dark constraints, whose effects can be seen though the constraint itself cannot. Constructors, a term taken from the physics of David Deutsch and Chiara Marletto, are things that bring about a transformation repeatedly without being used up, as a ritual or a standard process does.

The steps.

  1. Ask people throughout the system what shapes what they do. Insist on specifics. "The job code." "The twenty-day target." "Nobody rings back." "You don't want to be the one who escalates."
  2. Sort them. Which only limit? Which connect? Which are suspected but invisible?
  3. Ask of each: who can change it, at what cost, and how fast?
  4. Choose a few to alter, add or remove, and predict in what direction behaviour should shift.

What it is good at. It turns "culture" into a list of concrete things. At Calder Vale it would surface the job code, the meter that is not in the van, and the arrears letter that follows a complaint.

Where the theses take hold. The constraints that most define a human community are neither governing nor enabling in this sense. They are constitutive: they make an action the kind of action it is. "Reporting a repair" at Calder Vale counts, in practice, as "causing trouble", and that meaning is remade every time a tenant stays quiet and every time an officer sighs. Such constraints exist only in being enacted, and mapping them is one more enactment. Thesis IV argues that the typology borrowed from physics has no place for these. There is also the matter of step 3. "Who can change it" is a question about power, and it deserves to be the organising question and not the third.

Method 3: estuarine mapping

Purpose. To decide where effort is worth spending, given that some things cannot be changed, some change by themselves, and some can be shifted.

The image. An estuary, where a river meets the tide. Some features are granite and will be there in a thousand years. Sandbanks shift each season. The water changes twice a day. A sailor who knows which is which can work with the place.

The steps, in the usual presentation.

  1. List the constraints and constructors at work, as in Method 2.
  2. Place each on a grid. One axis is the energy cost of changing it: effort, money, political capital. The other is the time it would take.
  3. Mark off the far corner, where both cost and time are beyond you. This is the counter-factual region: things that are simply not going to change by your agency, such as the law, the climate and the age of the stock. A band beside it holds things at the limit of the possible.
  4. Mark off the near corner, where things change with almost no effort and almost at once. That region is volatile: anything placed there is unstable and may flip without warning.
  5. Work on what is left in between. For each item ask whether to leave it and monitor, to make it cheaper or dearer to change, to break it up, or to remove it.
  6. Turn those choices into a portfolio of small actions, and repeat the mapping regularly, since the map changes as you act.

What it is good at. It is honest about limits. It stops Asha spending a year on what cannot move, and it draws attention to cheap changes she had not noticed. It replaces the picture of an ideal future with the question of what can shift from here.

Where the theses take hold. This is the method that gives the disputation its title. First, cost and time are not properties of a constraint. They are properties of a constraint for someone. Changing the job code costs Asha a difficult meeting, and it is entirely beyond an operative. A single map therefore records one position's view of what is movable, and what is granite to a tenant may be a sandbank to the board. Second, a real estuary does not know it is being charted. At Calder Vale, the contractor will learn within days which contract terms have been marked as changeable, and will act on that. The tide does not come in earlier because the sailor consulted a table. Third, the sailor is outside the water. Asha, her map and her workshop are in it. Thesis III sets out all three, and Chapter 45 proposes mapping by position and mapping the mapper.

Method 4: SenseMaker and self-signified micro-narrative

Purpose. To see the dispositions of a large human system by collecting many short first-hand accounts and having their tellers interpret them.

The steps.

  1. Design a prompting question that draws out a specific experience and not an opinion: "Describe a recent moment when something to do with repairs made you feel hopeful or hopeless."
  2. Design a signification framework. Its main components are triads, which are triangles with a quality at each corner on which the teller places a mark; dyads, which are lines between two poles, often two opposite excesses, so that neither end is the "right" answer; and sometimes canvases on which several items are placed. There are also multiple-choice questions about the teller and the story.
  3. Collect accounts continuously and in large numbers. Each teller records the account and then signifies it. The teller, and not an analyst, says what the story means.
  4. Display the marks. Thousands of dots in a triangle show clusters, empty areas and outliers. Every dot opens the story behind it.
  5. Read the patterns, usually in a workshop with decision-makers and sometimes with tellers. Look especially at outliers, which may be weak signals of something beginning.
  6. Ask the standard question: what could we do to get more stories like these and fewer like those? Act in small ways, keep collecting, and watch whether the dots move.

What it is good at. It joins the scale of a survey to the richness of a story. It removes one layer of expert interpretation. It makes change visible as movement in a pattern, not as arrival at a target. And it takes ordinary people's accounts seriously as evidence, which few instruments do.

Where the theses take hold. At four points, in order.

  • At step 2, someone designs the triangles. The teller interprets freely within a space whose dimensions she did not choose. Bourdieu would ask whether she comes to see her experience through them (Thesis IV).
  • Also at step 2, a triad's values sum to one, so it cannot record "all three" or "none" (Thesis VII).
  • At steps 4 and 5, the pattern is seen by those with access to the display. Tellers supply the dots and usually do not see the picture (Thesis VI).
  • At step 6 and thereafter, tellers learn what telling leads to, and tell accordingly. The second round is collected from people who are now responding to the instrument (Thesis VIII).

Method 5: safe-to-fail probes

Purpose. To learn what a complex situation will support by trying small things, since analysis cannot say.

The steps.

  1. Generate several different small experiments on the same issue. They should include some that contradict one another, some that come at the matter sideways, and some that most people expect to fail.
  2. Check that each is coherent: there is a plausible reason it might work. And check that each is safe to fail: if it goes wrong, the damage is small and can be contained.
  3. Before starting, write down for each how you would recognise that it is working and how that it is not, and what you would do to amplify it or to dampen it.
  4. Run them in parallel, not in sequence.
  5. Watch. Amplify what shows promise and dampen what does not. Expect several to fail and treat the failures as the purchase of information.

The example. At Calder Vale: put meters in four vans. Let two caretakers book follow-ups directly. On one estate, have a tenant volunteer accompany the first visit. On another, suspend arrears letters for six weeks after any repair report. Send a surveyor to do nothing but listen for a fortnight.

What it is good at. It breaks the paralysis of analysis. It spreads risk. It gives failure a respectable place.

Where the theses take hold. "Safe to fail" for whom? The experiment that fails costs the association a little money. It may cost the tenant who took part a winter with mould and the embarrassment of having trusted again. Risk in a portfolio of probes is distributed, and it tends to run downhill toward those with least. Thesis V asks that this be mapped. There is also the point made under Element 21: a probe among people is a public act. The estate that did not get a tenant volunteer draws its own conclusions.

Method 6: the vector theory of change

Purpose. To manage change without pretending to know the destination.

The steps.

  1. Describe where you are, using the dispositions shown by Methods 2 to 4.
  2. Do not define an ideal future state. Define a direction: more of this kind of thing, less of that.
  3. Identify the adjacent possible: the next steps that are actually reachable from here.
  4. Take them, look again, and re-set the direction as the landscape changes.

What it is good at. It prevents the familiar gap between a vision and a reality that never meets it. It keeps attention on the present.

Where the theses take hold. A direction is a choice of what is better, made by someone. And the adjacent possible is not the same for all. What is one step for a director may be ten for a caretaker. A method that begins from "where we are" needs to say who we are.

How the methods fit together

The six form a chain. Cynefin sorts situations and identifies which are complex. For those, constraint mapping and SenseMaker show the dispositions. Estuarine mapping says where effort can count. Safe-to-fail probes act at those points. The vector theory holds the direction and says when to look again. It is a coherent practice, and it is far better suited to a situation like Calder Vale's than a programme plan with a critical path.

The disputation's criticism can now be stated compactly, since the reader has the chain in view. At every link there is a designer, a facilitator or an analyst who chooses items, writes signifiers, draws boundaries, sets costs, selects probes and names the direction. And at every link there are people who know they are being sorted, mapped, signified and probed, and who respond. The methods are presented as ways of reading a system. By the theory's own asymmetry principle they are also ways of writing it. The reconstruction in Book Eight keeps every one of these six methods and adds to each the same three things: a record of position, a return path to those who supplied the material, and a place on the map for the instrument itself.

SIX METHODS, ONE CHAIN —

EVERY LINK HAS A HAND ON IT —

AND THE HAND BELONGS ON THE MAP

Addition: Cross-Lights From Book Nine

Element 20 extended. The Pelagian stance is now one of three. Snowden's trialectic adds the formative encounter and insists that all three modes remain in use. It also says where his own instruments stand: the choice of an area of inquiry may be Augustinian, self-signification is Pelagian, and co-creative designs are the third mode. Chapter 50 examines that paragraph closely.

Element 19 and first attribution. The post on Luhmann states the principle behind distributed sensing as a right: people should give meaning to their own experience before anyone else does. Chapter 52 accepts the right and asks whether the designer of the signifiers has spoken first.

Element 22 and the piper. Snowden says the market rewards the deficit mode. Flinn asks his contributors what they do when they dislike the payer's tune. Both schools acknowledge the pressure.

A mirror image. Chapter 57 sets Flinn's practice beside this method chain: dozens in presence against thousands at a distance.

Chapter 6

Genealogy: From Cynefin to Anthro-Complexity

4,363 words, about 19 minutes of reading

Why a Genealogy

A genealogy, in Nietzsche's sense and Foucault's after him, is not a history of origins. It does not trace a concept back to a pure beginning from which it has declined or progressed. It traces the contingent paths by which a concept came to have the shape it has, and it looks especially for points where the concept was turned to purposes other than the ones it began with. A genealogy of anthro-complexity matters for this disputation because the question of where the human sits in it is partly a historical question. The human cluster did not arrive all at once. It was added to a framework that existed before it, and the framework's earlier shape still governs where the human could be put.

The genealogy offered here is the author's reconstruction from Snowden's published work and from the record of the Cynefin lineage. It is a sketch, not a scholarly history, and dates are given as approximately as the author's knowledge allows. Its purpose is to show a drift, the one The Compass and the Loom identified, from ontology towards product, and to locate the human cluster on that drift.

Stage One: Knowledge Management and Sense-Making (late 1990s to early 2000s)

Cynefin began in knowledge management. In the late 1990s Snowden worked for IBM, which had acquired the consultancy he worked for, and was associated with IBM's Institute for Knowledge Management. The knowledge-management movement of the time was dominated by a codification model. Knowledge was a resource that could be extracted from people's heads, written down, stored in databases and redistributed. The SECI model of Nonaka and Takeuchi, with its spiral of tacit and explicit knowledge, was often read, against its authors' intentions, as a recipe for converting tacit knowledge into explicit.

Snowden's early work was a sustained critique of the codification model. Knowledge, he argued, is not a thing that can be extracted. It is volunteered, not conscripted. People know more than they can tell, and they tell more than they can write. Much of what organisations know exists only in the interactions of their members. Narrative was the vehicle: the stories people tell carry knowledge that no procedure captures. Snowden's early work on anecdote circles, archetypes and narrative patterns was an attempt to work with knowledge in its natural form rather than force it into codified form.

The Welsh word cynefin, which Snowden has glossed as the place of one's multiple belongings, the habitat that shapes us in ways we cannot fully know, first named a model of knowledge flows and cultural belonging. Only later did it name the decision framework now associated with it. The early framework concerned how knowledge moves through communities, what kinds of knowledge are shareable, and what the boundaries between knowledge communities look like.

Where the human sits at this stage. At the centre. The early work is about humans as knowers, people who know more than they can tell, whose knowledge lives in narrative and community, and who resist being treated as data sources. It is, in a sense, already anthro-complexity before complexity. The human is the narrator, and the method is to listen.

Stage Two: The Decision Framework (early to mid 2000s)

The framework took the shape most managers now recognise in the early 2000s. The key text is Kurtz and Snowden's 2003 paper in the IBM Systems Journal, The new dynamics of strategy: sense-making in a complex and complicated world. A wider audience came with Snowden and Boone's 2007 Harvard Business Review article, A Leader's Framework for Decision Making. Here the domains appear as a typology of situations: known (later simple, then obvious, now clear), knowable (complicated), complex and chaotic, with disorder at the centre. Each has a characteristic relation between cause and effect and a characteristic decision model.

The paper's central move was to distinguish ordered from unordered systems and to argue that most management theory assumed order. It also introduced complexity science into the framework in a systematic way. The complex domain is defined by emergent order: cause and effect are coherent only in retrospect, and the appropriate response is probe-sense-respond. Here the scientific sources (Prigogine, Kauffman, Holland, and later Juarrero) begin to carry weight.

Where the human sits at this stage. The human moves from the centre to the decision-maker. The framework asks: what kind of situation is this, and how should a leader decide in it? The leader's perception of the situation becomes the hinge. The HBR article is explicitly addressed to leaders and frames the domains as contexts a leader must diagnose. The narrators of Stage One are still there, but they are now part of the situation the leader reads. The ontological question (what kind of system is this?) has become an epistemological one (how does a leader know what kind of system this is?), and that question has an obvious practical follow-up: what should the leader do?

This is the first step of the drift, from ontology to epistemology. It was not a mistake. A framework that is to be used must say how one recognises which domain one is in. But it introduced a figure, the leader who reads the situation, who stands in a particular relation to the situation. The leader is in it, as Cynefin always insisted, but the leader is also the one who reads it.

Stage Three: Instruments (mid 2000s to 2010s)

Cognitive Edge was founded in the mid 2000s, and with it came the instruments, above all SenseMaker. SenseMaker turned the narrative methods of Stage One into software. Micro-narratives could be collected at scale, self-signified on triads and dyads, and aggregated into patterns visible on dashboards. The key innovation, self-signification, kept interpretation with the narrator and was designed to avoid the analyst's interpretive authority. The scale was new. Where Stage One worked with anecdote circles of a dozen people, SenseMaker could gather tens of thousands of narratives across a population and show patterns no individual could see.

Around the same period the methods multiplied: a growing library of facilitation methods, and later the training and accreditation programmes that would carry Cynefin into organisations worldwide. Practitioners were trained, methods were codified (in the honest sense of being written down for teaching), and a network formed.

Where the human sits at this stage. The human is now a source of signal. That is not a judgement on SenseMaker, which was designed to respect narrators and give them interpretive authority over their own stories. It is a description of what the instrument does. It collects narratives and turns them into patterns. The narrator self-signifies, but the pattern is seen by whoever holds the dashboard, and the dashboard is usually held by the client who commissioned the study. The narrators of Stage One have become respondents. They are honoured respondents, whose self-interpretations count, but respondents all the same, answering a question someone else posed on signifiers someone else designed.

This is the second step of the drift, from epistemology to something like a technology of sensing. It was not a mistake either. Scale matters, and patterns visible only at scale are real and important. But it introduced a second figure beside the leader: the analyst, or the dashboard, which sees what no narrator sees.

Stage Four: Constraints and Energetics (2010s to early 2020s)

From the 2010s Snowden's work drew increasingly on Alicia Juarrero's theory of constraints. Her Dynamics in Action (1999) had argued that intentional action could be understood through context-sensitive constraints rather than efficient causes. Her later Context Changes Everything (2023) developed the account into a general theory of how coherence arises in complex systems. Constraints became the central ontological concept of the Cynefin lineage, displacing the earlier emphasis on domains. The framework came to be seen less as a typology of situations and more as a map of constraint regimes: rigid constraints in the clear domain, governing constraints in the complicated, enabling constraints in the complex, and the absence of effective constraint in the chaotic.

Estuarine mapping grew out of this. It began as affordance mapping and turned into a method for placing constraints on axes of energy and time: what it would cost to change each constraint, and how long it would take. The language of energy (the energy cost of change, the energy gradient, the energy needed to shift a constraint) became more prominent. With it came an interest in the physics of dispositional states, the vectors of Element 4, and the possibility of measuring a system's direction of travel.

Where the human sits at this stage. The human is now part of a field of constraints. The estuarine map does not represent people. It represents constraints, some of which are people's habits, beliefs, roles and relationships, alongside budgets, regulations, technologies and buildings. Mapping is participatory, so the people in the system place the constraints. But what is mapped is not the people. It is the field of forces they are subject to. The language of energy completes the shift. A constraint's energy cost is a property it has, like the height of a hill, and the question for a practitioner is where the terrain gives way.

This is the third step of the drift, to energetics. It was a profound advance in many ways. Juarrero's theory of constraints is one of the most important contributions to complexity thinking in a generation, and estuarine mapping is a powerful practical instrument. But it introduced a third figure: the terrain. Constraints are mapped as a landscape with a topography of energy, and the practitioner, however participatory the mapping, looks at the map.

Stage Five: Anthro-Complexity Named (2010s onward, systematised in the 2026 series)

The term anthro-complexity appears in Snowden's writing over many years as a name for the study of complexity in human systems, distinguished from complexity in general and from computational complexity in particular. The 2026 series is the most sustained attempt to systematise it, setting it against other bodies of thought: semiotics, the new materialisms, Ingold, and now Ing. In the series the human cluster of Chapter 4 is stated explicitly: multiple identities, intentionality, abstraction and narrative, the asymmetry principle, entanglement without flat ontology, and the rejection of the programme. Peirce returns as the foundational philosopher, and the trialectic is developed as anthro-complexity's characteristic structure of thought.

Where the human sits at this stage. In the theory, back at the centre. The human cluster puts what is distinctive about human beings at the heart of what is distinctive about anthro-complexity, and it returns, in a more rigorous form, to the concerns of Stage One: knowledge that cannot be extracted, narrative as the vehicle of meaning, people as knowers and not data sources. The tacit–explicit–entangled trialectic is in a direct line from the early critique of codification.

Stage Six: Products (2020s)

The Cognitive Edge organisation was renamed the Cynefin Company in the early 2020s, and its recent instruments include SwarmCompass and Weaving Risk and Strategy (WRAS, later WRASSE), which The Compass and the Loom examined. They are marketed to organisations as solutions to recognisable problems. SwarmCompass addresses organisations "adrift without direction" and promises to help them find direction through distributed sensing. WRAS weaves risk management and strategy into a single framework with four domains of strategic posture (directed, opportunist, hypothesis-based, distributed) and axes of natural versus stimulated emergence and fine versus coarse granularity.

Where the human sits at this stage. The Compass and the Loom found that the human is now sensor and substrate. In SwarmCompass, people are sensors whose noticing is gathered to steer the organisation; the brochure's deliverables, on that treatise's reading, turn human noticing into training data for automation. In WRAS, people are part of a "substrate", a stratigraphy of technological, cultural, procedural and network layers on which strategy is built. In both, the root assumption is what the treatise called read, then act from above: a manager who stands outside the system, reads it, and then chooses whether to let emergence happen or to stimulate it.

This is the final step of the drift, to product. The Compass and the Loom put the whole path in a sentence: the lineage drifts from ontology (what kind of system is this?) to epistemology, energetics, decision, and finally product, and each step moves closer to the client's question.

The Drift Displayed

Stage Approx. period Central concept Where the human sits Characteristic figure
1. Knowledge & narrative late 1990s Knowledge cannot be conscripted At the centre: knower, narrator The listener
2. Decision framework early–mid 2000s Domains; ordered vs unordered Decision-maker; leader reads situation The leader
3. Instruments mid 2000s–2010s Narrative at scale; self-signification Source of signal; honoured respondent The analyst / dashboard
4. Constraints & energetics 2010s–early 2020s Juarrero's constraints; energy cost Part of a field of constraints The terrain
5. Anthro-complexity named 2010s–2026 Human cluster; Peirce; trialectic In theory, back at the centre The narrator restored
6. Products 2020s SwarmCompass; WRAS/WRASSE Sensor and substrate The steersman

What the Genealogy Shows

The table shows something the elements of Chapters 2 to 5 could not show on their own. The human cluster, the heart of anthro-complexity, was articulated at Stage Five. But the instruments through which anthro-complexity is practised were built at Stages Three, Four and Six, when the human sat elsewhere: as respondent, as part of a constraint field, as sensor and substrate. The theory caught up with the human after the instruments had been designed around different figures.

That is not unusual. Theories often catch up with practices; Hegel's owl of Minerva flies at dusk. But it means the instruments do not necessarily embody the theory. They embody the figures of the stages at which they were made. SenseMaker embodies the analyst who sees what no narrator sees. Estuarine mapping embodies the terrain whose energetics can be read. SwarmCompass and WRAS embody the steersman who reads and then acts from above. When Stage Five restored the human to the centre of the theory, it did not redesign the instruments around the restored figure. It reinterpreted them. SenseMaker is now described as distributed sensing that keeps narrators' interpretive authority, and estuarine mapping as participatory. Those descriptions are true. But the instruments' structure (who designs the signifiers, who holds the dashboard, who reads the map, who receives the pattern) is that of the earlier stages.

There is a second thing the genealogy shows, and it is subtler. The drift is not only a matter of commercial pressure, though commercial pressure is part of it. It also follows an internal logic. Each step answers a question the previous step raised:

  • Stage One says knowledge lives in narrative and community. It raises the question: how does a leader make use of it? Stage Two answers with the decision framework.
  • Stage Two says the leader must recognise which domain she is in. It raises the question: how does she see the situation at scale? Stage Three answers with SenseMaker.
  • Stage Three says patterns are visible at scale. It raises the question: what shapes the patterns? Stage Four answers with constraints.
  • Stage Four says constraints have energy costs. It raises the question: what can a client do with that? Stage Six answers with products.

Each answer is reasonable, and each moves the theory one step closer to the client's question. Each step also moves the human one step further from the centre, because the client's question is always some version of what should I do?, and that question is asked from the position of someone who acts on the system, not someone who is acted on by it. A theory that answers the client's question is a theory that adopts the client's position. The client's position is outside, or above, the people whose narratives are gathered and whose constraints are mapped.

Stage Five, the naming of anthro-complexity, can be read as Snowden's recognition of the drift and an attempt to correct it. The return to Peirce, the renewed emphasis on narrative and entanglement, the insistence that humans are not ants, and the asymmetry principle all restate the concerns of Stage One with the rigour of Stage Four. The November post's critique of Ing, who does the reading?, is a Stage One question asked against a Stage Two answer. Snowden sees the problem. The question this book asks is whether he has seen it in his own instruments or only in Ing's.

The Genealogical Thesis

The genealogy can be summed up in a thesis that the rest of the book will test:

The genealogical thesis. Anthro-complexity's theory of the human (Stage Five) and its instruments (Stages Three, Four and Six) belong to different moments of the lineage and embody different figures of the human. The theory places the human at the centre as narrator, knower and reflexive actant. The instruments place the human as respondent, as part of a constraint field, and as sensor and substrate. Reinterpreting the instruments in the language of the theory has not redesigned them. The tensions identified in Chapters 2 to 5 (Elements 2 against 12, 3 against itself, 5 against 4 and 7–10, and the flags on Elements 16 to 22) are the traces of that mismatch.

The adversary has a strong reply. The instruments were never designed around a figure of the human as "respondent" or "substrate". From the start they were designed to protect the human from those positions. SenseMaker's self-signification exists precisely so the analyst cannot impose meaning on narratives. Estuarine mapping is done by the people in the system, not by consultants. The disintermediation Snowden has always championed aims to remove the expert who stands between the people and their own meaning. To say the instruments embody the analyst, the terrain and the steersman is to describe what they are designed against.

This reply wins half. The intent of the instruments is as the adversary describes, and the intent is admirable. The genealogical thesis concerns not intent but structure: who designs the signifiers, who holds the dashboard, who receives the patterns, who decides what the map is for. Intent can sit inside structure without changing it. A well-meant survey is still a survey. The test is not what the instruments were designed to avoid but what they do when they are used: by a client who commissioned them, on a population that did not, for purposes the population may not share. Chapters 19 to 26 will apply that test thesis by thesis. partial

The Shape of What Follows

Book One has laid out the claim. It has stated twenty-two elements in four clusters, read the November post's architecture and image, and traced a genealogy in which theory and instruments belong to different moments. Book Two now turns to the neighbours: complex adaptive systems, from which anthro-complexity must be distinguished; systems thinking, from which Snowden has just distinguished it; complex responsive processes, which he does not name but which shadows the whole argument; and critical systems thinking, which he names in his postscript as engaging with complexity far more directly. Only when the neighbours are mapped can the question of Book Three be asked properly: why does anthro-complexity treat human complexity differently, and is it right to?

THE LISTENER — THE LEADER — THE DASHBOARD —

THE TERRAIN — THE NARRATOR RESTORED — THE STEERSMAN:

SIX FIGURES, ONE LINEAGE, AND THE THEORY ARRIVING LAST.

Addition: Stage One as a Resource

The genealogy traced six stages and found that the human was at the centre at Stage One, the period of knowledge management and narrative, and moved outward thereafter, returning to the centre only in theory at Stage Five. This addition asks what Stage One offers the reconstruction of Book Eight. Its methods were developed before the instruments of the later stages, and several of them embody the relational account of the human more fully than the instruments that replaced them.

The methods of Stage One

The methods associated with Snowden's early narrative work, as far as the published record allows them to be described, include the following.

Anecdote circles. Small groups gathered to share stories about their experience, prompted by questions designed to elicit anecdotes rather than opinions. Stories were told to one another, in conversation, and one story would prompt another. The facilitator's role was to keep the stories flowing and to avoid interpreting them.

Archetype extraction. From a body of collected stories, groups identified the characters who recurred (the hero, the villain, the fool, the trickster) and built archetypes, composite characters that represented recurring patterns in the organisation's self-understanding. Archetypes were created by the group, from its own stories, and they often revealed what the group thought about its leaders, its customers or itself in ways that direct questioning would not.

Emergent themes. Groups sorted stories into themes that emerged from the stories themselves, not into categories supplied in advance.

The use of narrative to surface tacit knowledge. Narrative was valued because stories carry knowledge that their tellers cannot state. A story about how a crisis was handled carries knowledge of how to handle crises that no procedure captures.

What these methods embody

Read with the disputation's vocabulary, the Stage One methods embody several of the principles that Book Eight proposes.

Stage One method Disputation principle it embodies
Stories told to one another, in conversation Address (Bakhtin); participatory sense-making; the conversation of gestures (Mead)
One story prompting another Correspondence (Ingold); meaning completed by response
Facilitator does not interpret The interpretant with the community (Peirce)
Archetypes built by the group Representation by the represented, not by an analyst (Hobbes reversed)
Emergent themes, not supplied categories Frames, not seers; no imposed signifier set
Narrative as carrier of tacit knowledge 理, the grain known by the knife; mētis
Small groups Overlapping horizons (Hutchins) at their natural scale

The Stage One methods are, in this reading, the most relational of anthro-complexity's methods. They gather stories in relation, interpret them in community, and let the categories emerge from the stories. They are the listener's methods, and the listener is the figure the reconstruction proposes to restore (Chapter 44).

Why they were superseded

The Stage One methods were not abandoned. They remain in the Cynefin repertoire, and practitioners use them. But they were superseded as the primary instruments by SenseMaker and the later methods, for reasons the genealogy identified. They did not scale. An anecdote circle works with a dozen people, and an organisation has thousands. They were slow, because each circle took hours and synthesis across circles took longer. They were dependent on the facilitator's skill, so a poor facilitator produced poor results, and skill was hard to transfer. And they did not produce the kind of output that clients could act on: dashboards, patterns and numbers.

SenseMaker solved each problem. It scaled to tens of thousands, it was fast, it reduced dependence on facilitator skill by building the methodology into software, and it produced patterns that clients could see. Each solution was real. Each also lost something the Stage One methods had: conversation, mutual response, emergent categories, interpretation in community and the small group's overlapping horizons.

Recovering Stage One at scale

The reconstruction does not propose returning to Stage One, which would be to accept the adversary's Article 7 charge that the alternative cannot scale. It proposes recovering what Stage One embodied within the scaled instruments. Several of Chapter 45's protocols are recoveries of this kind.

  • The address (SenseMaker step 6) recovers something of the conversation. It asks to whom the story would be told, restoring the story's relational context.
  • The narrators' reading (step 12) recovers interpretation in community. Narrators read the patterns their stories produced.
  • The fourth vertex (step 8) recovers emergent categories. Narrators propose what the signifiers miss.
  • Nested bridges (addition to Chapter 31) recover the small group's overlapping horizons, at every level of a large organisation.
  • The knife-holders' grain (estuarine step 7) recovers narrative as a carrier of tacit knowledge.

And one Stage One method could be recovered directly, as a complement to SenseMaker at scale: the anecdote circle as the grating at the knot. Where a SenseMaker pattern shows a tangled place, a cluster of contested stories or a conflict between groups, the pattern is not resolved statistically. It is taken to anecdote circles of those whose stories make up the cluster, who tell their stories to one another, in conversation, and make sense of the knot together. That is Cook Ding slowing at the tangled place (Chapter 36), with Stage One's method as the slow knife.

The genealogy reconsidered

The genealogy traced a drift from the listener towards the steersman. Stage One is not a lost golden age. Its methods had real limitations, and the later stages addressed them. But Stage One contains, in its methods, a relational account of the human that the later stages' instruments did not carry forward, and that Stage Five's theory restated without recovering the methods that embodied it. The reconstruction is, in one sense, a return to Stage One with the later stages' resources. The listener comes back, carrying SenseMaker, estuarine mapping and Juarrero's theory of constraint, and uses them as the listener would.

STORIES TOLD TO ONE ANOTHER — ARCHETYPES BUILT BY THE GROUP — THEMES THAT EMERGE —

THE LISTENER'S METHODS DID NOT SCALE; THE LISTENER'S PRINCIPLES CAN —

TAKE THE ANECDOTE CIRCLE TO THE KNOT.

Addition: Cross-Lights From Book Nine

A parallel genealogy. The Zen Garden study traces what happened to Ralph Stacey's diagram between 1993 and the present. It was copied, simplified into four boxes, relabelled so that its political axis disappeared, and attached to purposes its author rejected. He disowned it. The study's term for what a framework should be is a threshold object: valid where it points beyond itself, and to be put down there.

Cynefin's history is different in one respect and alike in another. Its author has kept revising it and has resisted its packaging. It is nonetheless an easily drawn figure, and easily drawn figures travel without their authors. Chapter 53 applies the lesson.

Snowden on the Stacey matrix. He places it near systems thinking, on the ground that it defines complexity by what observers agree and are certain about. The Garden study's reply is that the agreement axis recorded who wants what, which is a fact about the situation.

Chapter 7

Complex Adaptive Systems: The Santa Fe Inheritance

4,930 words, about 21 minutes of reading

The Neighbour That Lends the Name

Anthro-complexity must be distinguished from complex adaptive systems theory if its name is to mean anything. The difficulty is that the two share a great deal. Both hold that order can emerge from local interaction without central control. Both treat systems as far from equilibrium, nonlinear and path-dependent. Both reject the forward-causal certainties of mechanistic management. Both draw on the same founding figures. And much of what is written about organisations as complex systems uses the vocabulary of CAS, of agents, adaptation, emergence and fitness landscapes, without distinguishing it from anything else. When a manager says her organisation is "a complex adaptive system", she is usually reaching for the same insight Cynefin offers. So the distinction has to be drawn with care, and the first step is to state what CAS theory actually is, in its own terms and at its best.

The Santa Fe Moment

The Santa Fe Institute was founded in 1984 by a group of scientists, many associated with Los Alamos, who believed that a new science of complexity was emerging across disciplines and needed an institutional home outside the departmental structure of universities. Its founding figures included the chemist George Cowan, the physicist Murray Gell-Mann, and later the economist W. Brian Arthur, the computer scientist John Holland and the theoretical biologist Stuart Kauffman. Its ambition was unification: to find the common principles underlying phenomena as different as the immune system, the economy, ecosystems, the brain and the origin of life. M. Mitchell Waldrop's Complexity (1992) told the story for a general audience, and the phrase "complex adaptive system" became the banner of the programme.

The programme's characteristic method was computational. Its scientists built models (cellular automata, genetic algorithms, agent-based simulations, Boolean networks) and ran them to see what patterns emerged. The method was not incidental. It was how the programme claimed to study phenomena that resisted both traditional analysis, which needs decomposability, and traditional statistics, which needs large numbers of independent events. A complex adaptive system is neither decomposable nor a collection of independent events, but it can be simulated. The simulation became the laboratory.

Holland: Agents, Rules and Adaptation

John Holland's Hidden Order (1995) is the most systematic statement of the CAS programme. Holland invented the genetic algorithm in the 1960s and 1970s, a method of optimisation modelled on natural selection, and his account of complex adaptive systems carries the genetic algorithm's logic into every domain it touches.

For Holland, a complex adaptive system is a system of agents, each described in terms of rules. The agents interact, and in interacting they adapt: they change their rules in response to experience, so that rules which do well become more frequent and rules which do badly decline. Holland identified seven basic features common to all such systems: four properties and three mechanisms.

Aggregation. Agents form aggregates, and the aggregates behave as agents at a higher level. Cells form organs and organs form organisms. Firms form industries and industries form economies. Aggregation is how hierarchy arises without design.

Tagging. Agents carry tags, identifying markers that let other agents recognise them and interact selectively. A flag, a trademark, a pheromone or an antigen is a tag. Tags make aggregation possible, because they let agents distinguish partners from strangers.

Nonlinearity. Interactions are nonlinear. The behaviour of the whole is not the sum of the behaviours of the parts, so it cannot be found by adding up the parts.

Flows. There are flows of resources (goods, signals, energy) through networks of agents. Flows have multiplier effects and recycling effects, which make the system's behaviour sensitive to its network structure.

Diversity. Agents are diverse, and the diversity is dynamic. When an agent is removed, the system typically adapts so that another agent fills its niche. Diversity is a product of adaptation and also a condition of it.

Internal models. Agents have internal models, structures that let them anticipate. A bacterium swimming up a chemical gradient has an implicit model: the gradient predicts food. A human trader has an explicit model. Internal models are how agents act on the future, and they are themselves subject to adaptation.

Building blocks. Internal models are built from reusable components, building blocks that can be recombined to form new models. Holland's example is the human face, recognisable through a finite set of features recombined in endless variety. Building blocks are how adaptation achieves novelty without starting from scratch.

The architecture is elegant and has been influential far beyond Santa Fe. Every one of the seven features can be found in human organisations. Teams aggregate into departments. Uniforms, titles and brands tag. Interactions are nonlinear. Money, information and attention flow. People are diverse. Each person has internal models of the organisation, colleagues and markets. And those models are built from reusable components: categories, scripts, stories.

Gell-Mann: Schemata and Compression

Murray Gell-Mann, a Nobel laureate in physics for his work on the classification of elementary particles, gave the CAS programme a philosophical statement in The Quark and the Jaguar (1994). His central concept was the schema. A complex adaptive system, on his account, gathers information about its environment and its own interaction with it, identifies regularities in that information, and compresses the regularities into a schema, a model or theory, which it uses to act. The schema is tested against experience, and schemata that predict well survive while those that predict badly are modified or discarded.

For Gell-Mann the same logic runs through biological evolution (the genome is a schema), the immune system (antibodies are schemata for recognising antigens), individual learning (beliefs are schemata), scientific inquiry (theories are schemata), and the evolution of human societies (cultures are schemata). The key operation is compression: identifying what is regular in the information and discarding what is random. Gell-Mann defined a measure he called effective complexity, the length of the shortest description of a system's regularities, to capture how much structure a system has.

Gell-Mann's account is important for this disputation because it places at the centre of CAS theory something close to what anthro-complexity places at the centre of the human: the capacity to form representations of the world and act on them. A schema is an abstraction, and a CAS agent with a schema acts on its abstraction of the environment, not on the environment directly. If anthro-complexity's third mark of the human is the capacity to create abstractions that change constraints, Gell-Mann's CAS agents already have something like it.

Kauffman: Landscapes and the Adjacent Possible

Stuart Kauffman's The Origins of Order (1993) and At Home in the Universe (1995) brought two concepts into the CAS programme that have since spread widely. The first is the fitness landscape, specifically the NK model. Kauffman represented the space of possible configurations of a system as a landscape whose height at each point is the configuration's fitness. Adaptation is movement across the landscape towards higher ground. The NK model tunes the ruggedness of the landscape by varying the number of interactions among the system's components (K) relative to their number (N). When K is low, the landscape is smooth and adaptation is easy. When K is high, it is rugged, with many local peaks, and adaptation tends to get stuck. Between the extremes lies a regime where adaptation works best, and Kauffman argued that evolution tends to push systems towards it: towards "the edge of chaos".

The second concept, developed in Kauffman's later work, is the adjacent possible: the set of configurations that can be reached in one step from the present. The adjacent possible is not fixed. Every time a system moves into it, new possibilities open that were not adjacent before, so the space of possibilities expands as the system explores it. Kauffman argued that the biosphere, the economy and human culture all expand into their adjacent possibles, and that the expansion cannot be predicted, because the possibilities it opens could not have been listed before they became adjacent. The Compass and the Loom relies on this claim against SwarmCompass's promise to map "unknown unknowns": Kauffman's adjacent possible cannot be listed in advance.

Kauffman's work also included autocatalytic sets: collections of molecules each of whose formation is catalysed by another member of the set, so that the set as a whole sustains itself. Kauffman proposed them as a model for the origin of life. They are a striking example of order arising from interaction without design, and they bear on anthro-complexity's interest in constraint closure, a set of constraints each maintained by others in the set, which The Compass and the Loom sets against WRAS's substrate.

Arthur: Increasing Returns and the El Farol Problem

W. Brian Arthur's work on increasing returns and path dependence in economics made the CAS programme matter to the social sciences. Classical economics assumed diminishing returns. As a firm grows, each additional unit of output costs more, so markets tend towards equilibrium with many firms sharing them. Arthur showed that in many markets, especially technology markets, returns increase. The more a product is adopted, the more valuable it becomes (through network effects, learning effects, standards), so markets tend towards lock-in, where one product dominates for reasons that may have more to do with early accidents than with quality. The QWERTY keyboard and the VHS videotape became canonical examples.

Arthur's El Farol problem (1994) is more important for this disputation. El Farol was a bar in Santa Fe with live music on Thursday nights. Arthur imagined a hundred people deciding independently each week whether to go. If fewer than sixty go, the evening is enjoyable; if more than sixty, it is too crowded. Each person must predict attendance, and the problem is that any prediction method, if widely shared, defeats itself. If everyone predicts low attendance, everyone goes and it is crowded. If everyone predicts high attendance, no one goes and it is empty. There is no deductively rational solution. Arthur modelled agents who each hold a set of predictors (simple rules for forecasting attendance from past weeks), use whichever has worked best recently, and update as the results come in. The result was that attendance fluctuated around sixty. The agents' predictions formed an "ecology" of forecasting rules, none dominant, each kept in check by the others.

The El Farol problem is a model of reflexivity. Each agent's prediction affects the outcome it predicts, and the agents' beliefs about the system are part of the system. This bears directly on anthro-complexity's human distinction. Arthur's agents are CAS agents with rules and internal models, and they are reflexive. Their models of the system change the system. If reflexivity is the deep ground of human distinctiveness, as Chapter 4's flags suggested, then CAS theory has already incorporated some of it.

Reflexivity in CAS: how far does it go? Arthur's El Farol agents, and the related "inductive reasoning" models of the Santa Fe artificial stock market, are reflexive in a precise and limited sense. Agents' predictions affect outcomes, and outcomes feed back into the selection of prediction rules. This is first-order reflexivity: beliefs about the system shape the system. What the models do not contain is second-order reflexivity, in which agents change their beliefs about the rules of the game: what counts as a good evening, whether sixty is the right threshold, whether the bar should be expanded or another bar opened. Nor do they contain agents who form coalitions, tell stories about the bar, or reorganise themselves into crews. Their reflexivity is reflexivity within fixed constraints. Anthro-complexity's third mark of the human, abstractions that change the very constraints agents work within, is reflexivity about constraints. The difference between the two is the difference between a market in which traders' expectations move prices and a market in which traders lobby to change the rules of trading. CAS models the first. Human systems do both.

Agent-Based Modelling and Its Images

Agent-based modelling is the practical arm of the CAS programme, and its canonical models have shaped how complexity is imagined in organisational life.

Schelling's segregation model (1971, before Santa Fe but adopted by it) showed that strong residential segregation can arise from mild individual preferences. Agents on a grid who are content unless fewer than a third of their neighbours are like them will, through repeated moves, produce a highly segregated pattern. No one wants segregation, but it emerges. The model became a parable of unintended macro-consequences of micro-motives.

Reynolds's boids (1987) showed that realistic flocking can arise from three local rules: separation (avoid crowding neighbours), alignment (steer towards the average heading of neighbours) and cohesion (steer towards the average position of neighbours). There is no leader and no plan, yet the flock moves as one. Boids became the canonical image of self-organisation, and through it the starling murmuration became the emblem of complexity in a thousand presentations.

Epstein and Axtell's Sugarscape (1996) was a "growing artificial societies" programme. Agents on a landscape of sugar (and later spice) foraged, traded, reproduced, fought and transmitted culture according to simple rules, and the simulation reproduced recognisable social phenomena: wealth inequality, trade networks, migration and conflict. It made a strong methodological claim: if you didn't grow it, you didn't explain it. A social phenomenon is explained when a population of agents with simple rules can be shown to generate it.

Ant-colony models, from Deneubourg and Goss's work on trail formation to Dorigo's ant colony optimisation, showed how pheromone-mediated foraging produces efficient routing without any ant knowing the route. Like the boids, they became images of organisational possibility: decentralised, adaptive, intelligent without an intelligence.

These models are brilliant, and they have taught real lessons. Schelling's model has been a valuable warning to policymakers. Boids explained flocking. Sugarscape showed that inequality can emerge from equal starting conditions. But they share an image of the human, or of the agent standing in for the human, that is the target of anthro-complexity's critique. The agent has a fixed identity (a colour, a type, a caste). It follows simple rules. Its rules may adapt, but only by selection among a predefined space. It does not narrate. It does not move between identities. It does not reflect on the rules of its world or change them by re-describing them. It is, in Snowden's phrase, an ant.

CAS Comes to Organisations

The CAS programme entered organisational thinking through several routes in the 1990s and 2000s. Ralph Stacey's Complexity and Creativity in Organizations (1996) drew heavily on CAS theory and agent-based modelling, before Stacey turned away from it towards complex responsive processes (Chapter 11). Plsek and Greenhalgh's influential 2001 BMJ series on complexity in health care described health systems as complex adaptive systems and drew lessons for clinical practice and management. Margaret Wheatley's Leadership and the New Science (1992) popularised a looser version. The business-strategy literature took up fitness landscapes, the edge of chaos and increasing returns. By the 2000s "complex adaptive system" had become a standard description of organisations in management education.

The organisational CAS literature varies greatly in quality. At its best (Plsek and Greenhalgh, the more careful strategy work) it uses CAS concepts to illuminate real features of organisations: the limits of control, the importance of local interaction, the danger of linear plans. At its worst it uses them as metaphors with no discipline: the organisation is an ant colony, the leader is a gardener, change is the edge of chaos. Snowden's long-standing hostility to "complexity metaphors" in management is aimed at the worst, and much of it is deserved.

Cilliers: Complexity Without Agents

Paul Cilliers's Complexity and Postmodernism (1998) is an important counterpoint within the complexity literature. It is not part of the Santa Fe programme and is more philosophical than computational. Cilliers argued that complex systems are incompressible: there is no description of a complex system simpler than the system itself that captures everything about it. Any model of a complex system leaves something out, and what it leaves out may be what matters. This does not make models useless. It makes them partial, and the choice of what to leave out is a choice with consequences, for which the modeller is responsible.

Cilliers's argument bears on CAS theory and on anthro-complexity alike. Against CAS, it says the agent-based model is a compression, and its agents' simplicity is a modelling choice, not a fact about the system. Against anthro-complexity, or any framework that summarises a complex human system in a map or a pattern, it says the same thing: the map is a compression, and the compression has ethical weight. Cilliers wrote that because our models of complex systems are necessarily limited, we have to take responsibility for the limits, and that this is an ethical as well as an epistemological matter. The Compass and the Loom concluded its discussion of prestatability with a version of this: every published compression of a human system is itself an intervention.

The CAS Image of the Human

Gathered together, the CAS programme offers a definite image of the human, or rather of the agent through which the human is modelled:

Feature CAS agent (Holland, Gell-Mann, ABM) What it leaves out (anthro-complexity's charge)
Identity Fixed type, tag or role Multiple identities, switching by context
Cognition Rules; internal models; schemata Pattern-based, embodied, susceptible to framing
Purpose Fitness; utility; implicit in selection Intention towards imagined futures
Adaptation Selection among rules Re-description of the situation
Representation Schema compressing regularities Narrative; abstraction that changes constraints
Reflexivity First-order (El Farol): beliefs shape outcomes Second-order: beliefs reshape rules of the game
Relation to constraints Operates within them Changes them by telling stories about them
Method of study Simulation; "grow it to explain it" Narrative; sense-making; probe

The table makes the anthro-complexity critique look decisive, and in many respects it is. Holland's agents and Sugarscape's foragers are not humans. Organisations are not ant colonies. The lessons of agent-based modelling, real as they are, are lessons about populations of rule-following agents, and organisations are not such populations.

What CAS Already Knows

Fairness requires a qualification, and it is important for the argument of this book. The CAS programme at its best already knows much of what anthro-complexity claims as its distinctive insight.

Gell-Mann knew about abstraction. His CAS agents form schemata, compressed abstractions of their environment, and act on them. Human science, in his account, is the supreme example of schema formation, and he was explicit that human schemata include theories about society that change society.

Arthur knew about reflexivity. The El Farol problem is a model of beliefs shaping the outcomes they predict, and Arthur drew from it a general lesson: in economic systems, expectations are part of the system and cannot be treated as external forecasts. His later complexity economics develops this into an account of economies as systems "in formation", constantly constructing themselves through agents' responses to the patterns their own responses create.

Cilliers knew about the limits of models. His incompressibility argument is the strongest available case against treating any model of a complex system as complete, and he drew ethical conclusions from it.

Holland knew about internal models and anticipation. His agents anticipate, and anticipation changes behaviour now in light of imagined futures, which is intentionality in a thin but real sense.

The ABM community knows its agents are idealisations. Serious modellers do not claim their agents are humans. They claim the models isolate mechanisms that may operate in human systems, and test what those mechanisms can and cannot produce. Schelling did not say people are grid-dwelling coloured squares. He said mild preferences can produce strong segregation, and that was worth knowing.

The qualification does not dissolve the anthro-complexity critique, but it does change its shape. The difference between anthro-complexity and CAS is not that CAS ignores abstraction, reflexivity, anticipation and the limits of models. It is that CAS models them as properties of agents within a simulated system whose rules the modeller has specified, and anthro-complexity insists that in human systems the rules are themselves subject to the agents' abstractions, the modeller is one of the agents, and the system cannot be simulated because its rules are not fixed.

The CAS theorist's strongest reply is that anthro-complexity's human distinction is a difference of degree, not of kind. Humans form more complex schemata than bacteria, are more reflexive than El Farol agents, and switch between more identities than Holland's tagged agents. But the same logic of agents, models, adaptation and emergence applies. On this reply, anthro-complexity is CAS with richer agents, and its refusal to model is a refusal to do science.

The reply wins half. It is right that the features anthro-complexity calls distinctively human appear, in thinner form, in CAS models, and that the distinction between human and non-human complexity is not as sharp as the slogan "humans are not ants" suggests. It is wrong that the difference is only of degree. When the rules of a system are themselves revised by its agents' descriptions of the system, and the modeller's description is one of those descriptions, the system cannot be simulated in Holland's sense. The simulation would have to include itself. That is a difference of kind, and it is the ground on which anthro-complexity stands. But note where the difference lies. It does not lie in the agents' properties (identity, intention, abstraction), which differ in degree. It lies in the reflexive relation between description and system. That is the thesis of this book. partial

The Distinction Drawn

The chapter can end with a provisional statement of the distinction between anthro-complexity and CAS, to be refined in the next chapter's twelve-dimension comparison and tested in Book Four:

The distinction, provisionally. CAS theory models human systems as populations of adaptive agents whose rules, however rich, are specified by a modeller standing outside the model. Anthro-complexity holds that human systems cannot be so modelled, because their agents' descriptions of the system, including the modeller's, change the rules of the system. Anthro-complexity states this difference as a difference in the agents: humans have multiple identities, intentions and abstractions that CAS agents lack. This book argues that the difference lies in the relation between description and system, and that stating it as a difference in agents concedes to CAS the very framework (agents with properties) that the difference should overturn.

CAS GIVES THE AGENT A MODEL OF THE WORLD —

ANTHRO-COMPLEXITY GIVES THE AGENT A STORY —

THE HARDER TRUTH: THE STORY IS ALSO TOLD ABOUT THE STORYTELLER.

Addition: Complexity Economics and the Economy in Formation

The chapter noted that the CAS programme at its best already knows much of what anthro-complexity claims as its distinctive insight, and it cited W. Brian Arthur's El Farol problem as a model of reflexivity. Arthur's later work goes further, and it deserves a fuller account, because it represents the CAS programme's own movement towards the position this disputation argues for. It also sharpens where the remaining difference lies.

The economy as a system in formation

Arthur's Complexity and the Economy (2014) collected his papers and set out complexity economics as an alternative to equilibrium economics. Its central claim is that the economy is not a system in equilibrium, which economic agents' choices bring into balance, but a system in formation: constantly constructing itself through agents' responses to the patterns their own responses create. Agents form expectations, strategies and technologies. These create an outcome, a pattern of prices, behaviours or technologies in use. Agents respond to the outcome by revising their expectations, strategies and technologies, and the revisions create a new outcome. The economy is always becoming, never in balance.

Arthur emphasised that this makes the economy reflexive in a deep sense. The outcomes agents respond to are the aggregates of their own previous actions. Agents are reacting to a world they have made, and their reactions remake it. Equilibrium economics treats outcomes as given to agents by an external structure. Complexity economics treats them as produced by agents and fed back to them.

Technology and the adjacent possible

Arthur's The Nature of Technology (2009) developed a related account of technological change. Technologies are combinations of earlier technologies, and each new technology becomes a component available for further combination. The space of possible technologies therefore expands as technologies are created, which is Kauffman's adjacent possible applied to the economy. Arthur argued that the economy is an expression of its technologies. As technologies change, the economy restructures itself around them, and new industries, institutions and ways of life emerge.

This account bears on anthro-complexity's instruments. Narrative instruments, dashboards and AI hybrids are technologies. On Arthur's account they become components of the economy, and of organisations, that restructure them. The organisation adopts SenseMaker and restructures itself around it: new roles (analysts), new practices (rounds of sensing), new expectations (that patterns will be returned and acted on), and new vocabularies. The instrument does not sit outside the organisation, sensing it. It is a technology that the organisation becomes partly an expression of.

Arthur and the human distinction

Arthur's complexity economics moves the CAS programme towards anthro-complexity's human distinction in several respects.

Feature Classic CAS (Holland) Complexity economics (Arthur) Anthro-complexity
Equilibrium Adaptation towards fitness No equilibrium; formation No normal; dispositional states
Reflexivity Rules adapt to environment Agents respond to outcomes they created Abstractions change constraints
Agents' models Rules; schemata Expectations, hypotheses, revised continually Narrative, sense-making
Novelty Recombination of building blocks Technologies as combinations; adjacent possible Emergence; exaptation
Method Simulation Simulation and history Narrative; probe; map

The table shows that Arthur's version of CAS shares with anthro-complexity the rejection of equilibrium, the emphasis on reflexivity and the importance of novelty. The remaining differences are in the agents' models and the method. Arthur's agents form expectations and hypotheses, formally modelled as competing predictors whose performance determines their use. Anthro-complexity's agents form narratives and make sense, and are not formally modelled. Arthur's method includes simulation. Anthro-complexity's rejects it for human systems.

Where the difference remains

The disputation's argument predicts where the remaining difference lies. Chapter 8 argued that the differences of kind between anthro-complexity and CAS are relation-located: in second-order reflexivity, the place of the interpretant, contested evaluation and the reflexive relation between description and system. Arthur's complexity economics has first-order reflexivity in abundance, since agents respond to outcomes they created. Does it have second-order reflexivity, agents re-describing the rules of the game?

In part. Arthur's account of technology includes agents creating new technologies that restructure the economy, which is a change in the game's rules, of a kind. But the agents in his models do not re-describe the economy's constitutive constraints (property, money, contract) through narrative, form coalitions to change them, or contest what counts as a good outcome. Those are the forms of second-order reflexivity that political economy and social theory study, and that complexity economics, for all its sophistication, leaves to other disciplines.

There is a further difference, and it is the one this disputation has pressed hardest. Arthur's models are built by a modeller who stands outside them. The modeller specifies the agents' predictor sets, the outcome function and the updating rules, runs the simulation and observes what emerges. The model does not include the modeller, and the modeller's description of the economy is not, in the model, one of the descriptions agents respond to. Yet in the real economy it is. Economists' models are published, read by agents and acted on (MacKenzie, Chapter 34). Complexity economics, like classic CAS, has a theory of reflexivity among its agents and no theory of its own reflexivity as an actor in the system it models.

That is the same gap this disputation finds in anthro-complexity. Both schools have seen that human systems are reflexive, and both have built theories of that reflexivity. Neither has fully applied the theory to its own descriptions. The difference between them is that anthro-complexity has a principle, the asymmetry principle and any framework applied to a human system becomes an actor within it, that, applied to itself, would close the gap. Complexity economics has no such principle, and would need to borrow one from the sociology of performativity.

The lesson

The CAS programme, in Arthur's hands, has moved a long way towards anthro-complexity's human distinction. The gap between them is narrower than the slogan humans are not ants suggests, and the CAS theorist's reply in Chapter 7, that the difference is one of degree, is stronger against Arthur's agents than against Holland's. But the remaining difference, the reflexive relation between description and system including the describer, is a difference of kind. It separates both schools from a fully reflexive practice. Anthro-complexity is closer to closing it, because its own asymmetry principle points the way.

THE ECONOMY IN FORMATION RESPONDS TO OUTCOMES IT CREATED —

THE MODELLER STILL STANDS OUTSIDE THE MODEL —

BOTH SCHOOLS SEE REFLEXIVITY; NEITHER YET SEES ITSELF IN IT.

Addition: Cross-Lights From Book Nine

Incompressibility as a paradox of the modeller. Chapter 56 argues that the idea of an incompressible system has its home in the modelling of complex adaptive systems. It assumes a fixed object, a describer outside it, and success as reproduction of the original. Those are the conditions of a simulation. They are not the conditions of a practitioner in an organisation. The author's term for what happens there is altered compressibility.

The reduction of complexity. Snowden's post on Luhmann sides with Bateson against Ashby's emphasis on reduction. Chapter 56 uses Ashby differently. His law of requisite variety can be met by cutting the world down to a model or by bringing enough variety to meet it. The second is knowing by taking part.

Chapter 8

Twelve Dimensions of Difference

3,671 words, about 16 minutes of reading

Method of the Comparison

The previous chapter set out the CAS programme in its own terms and drew a provisional distinction. This chapter refines it along twelve dimensions. For each dimension it states the CAS position, states the anthro-complexity position, and then assesses the difference. Is it a difference of kind or of degree? Is it well grounded? And where does it locate the human distinction: in properties of agents, or in the relation between description and system?

The assessment uses a consistent scale. A difference is of kind when the two positions cannot be placed on a single continuum, so that moving along the CAS dimension, however far, would never reach the anthro-complexity position. A difference is of degree when they can. A difference is agent-located when anthro-complexity grounds it in what human agents have or do as individuals. It is relation-located when the ground is the reflexive relation between descriptions and the systems described. The argument of this book predicts that the differences anthro-complexity states as agent-located will turn out to be of degree, and the differences that really are of kind will turn out to be relation-located.

Dimension 1: The Basic Unit

CAS. The basic unit is the agent, an entity with rules, an internal model and the capacity to adapt. Everything else (aggregates, flows, networks, landscapes) is built from agents and their interactions. Even the environment is often modelled as agents or as a field that agents modify.

Anthro-complexity. The basic unit is the actant: anything, human or non-human, that makes a difference to a course of action. Agents are one kind of actant. Constraints, constructors, rituals, artefacts, regulations, buildings and software are others. The analysis concerns a field of actants, not a population of agents.

Assessment. This is a difference of kind, and a well-grounded one. A field of actants that includes non-agentive constraints cannot be reached by enriching a population of agents. The constraints are not agents with very simple rules. They are a different category of thing, which shapes the probabilities of action without acting in the agentive sense. Juarrero's account gives the difference its rigour: a context-sensitive constraint is a relation among events, not an entity with rules.

Location. Neither agent- nor relation-located in the sense at issue. It concerns the ontology of complex systems in general, not the human distinction specifically. A starling murmuration, analysed in anthro-complexity's terms, would also be a field of actants: birds, the topological rule of seven neighbours, light, predators.

Verdict on the difference: holds as a difference of kind, at the level of general complexity ontology.

Dimension 2: What Drives Emergence

CAS. Emergence is driven by local interactions among agents following rules. The canonical demonstrations are boids, Schelling, Sugarscape and ant trails: simple rules, many agents, complex outcomes. The rules are the generative mechanism. Change the rules and you change what emerges.

Anthro-complexity. Emergence is shaped by constraints and modulating structures acting on dispositional states. Agents are only part of what is going on. The same agents in a different constraint regime produce different patterns, and the practitioner works on the constraints, not the agents.

Assessment. The difference is partly of kind and partly of emphasis. CAS models do include constraints. Schelling's grid, the boids' neighbourhood radius and Sugarscape's landscape are constraints, and varying them changes outcomes. The difference of emphasis is that CAS foregrounds rules and treats constraints as parameters, while anthro-complexity foregrounds constraints and treats agents' rules as one constraint among others. The difference of kind is Juarrero's: context-sensitive constraints allow downward causation, in which patterns at a higher level change the probabilities of events at a lower level without any external force. Most CAS models have only upward causation (rules generate patterns) and treat any apparent downward causation as a further effect of rules.

Location. General complexity ontology, as in Dimension 1.

Verdict: partial. Of kind where Juarrero's downward causation is at stake, of emphasis otherwise.

Dimension 3: The Status of the Human

CAS. The human is one more agent type. Insect colonies, markets, immune systems and organisations run on the same logic of agents, rules and adaptation. Humans may be richer agents, with more complex internal models, but they are agents of the same kind.

Anthro-complexity. Human systems are a distinct kind of complex system. Humans are not ants, and models drawn from insect behaviour mislead when applied to organisations.

Assessment. This is the central dimension, and the following dimensions break it into its parts. In summary, the difference is of kind if it rests on the reflexive relation between human descriptions and the systems described, and of degree if it rests on properties human agents have more of than other agents. The November post states it in terms of properties (three marks), with the asymmetry principle pointing towards the relational ground.

Location. Stated as agent-located, with a relational hint.

Verdict: to be decided by Dimensions 4 to 7.

Dimension 4: Identity

CAS. An agent typically has one type or role at a time, identified by a tag. Tags can change as an immune cell matures or a firm rebrands, but at any moment the agent is one thing. Some models allow multiple tags, but rarely the switching of whole rule sets by context.

Anthro-complexity. Humans hold multiple identities at once and move between them by context, often without noticing. A person is not one agent but several, and which one is active depends on cues the person may not consciously register. Crews and roles are the mechanisms.

Assessment. As stated, this is a difference of degree. A CAS agent could be given several rule sets and a context-sensitive switching rule, and some models do this: agents switch strategies when the environment changes, or carry several "personas" activated by different neighbours. Such an agent would not be a human, but the gap between it and the anthro-complexity picture would be a gap in richness, not in kind. Moving along the CAS dimension (more rule sets, subtler switching) approaches the anthro-complexity position.

The difference becomes one of kind only when one asks where identities come from. In CAS, the modeller specifies the identities and the switching rule. In human systems, identities are formed and re-formed in interaction. A person becomes a "union advocate" partly because others treat her as one, and an identity can be invented on the spot when a new situation calls for it. The repertoire of identities is not fixed in advance. It is produced in the system, by the system's own descriptions of who is who. That is a difference of kind, and it is relation-located: it rests on the reflexive production of identity in interaction, not on the number of identities an individual holds.

Location. Stated as agent-located (humans hold multiple identities). Of kind only when relation-located (identities are produced in interaction).

Verdict: partial. Agent-located, it is of degree. Relation-located, it is of kind.

Dimension 5: Cognition

CAS. Agents have rules and internal models, and adapt by modifying them. In Holland's architecture, cognition is rule-based. In Gell-Mann's, it is schema formation by compression. In either case it is, at least in principle, specifiable: one could write down the agent's rules and schemata.

Anthro-complexity. Human cognition is pattern-based, embodied and susceptible to framing, priming and context. The brain does not run a programme. Cognition cannot be specified as rules, because it is not rule-following. It is the matching of present situations against a vast, fragmented store of past experience, in ways shaped by the body, the environment and the social context.

Assessment. This is a difference of kind if the anti-computational argument is right, and the weight of contemporary cognitive science (enactivism, embodied cognition, predictive processing) supports it. A rule-following agent cannot be made into a pattern-matching, embodied one by adding rules. However, the difference is not specifically human. Animal cognition is also embodied, pattern-based and context-sensitive. An ant's foraging, a starling's flocking and a dog's recognition of its owner are not rule-following in the strict sense either. The rules of ABM are modellers' abstractions of animal behaviour as much as of human behaviour. The anti-computational argument therefore separates real cognition from CAS models of cognition. It does not separate human from non-human cognition.

Location. Agent-located, but not human-specific.

Verdict: partial. Of kind against CAS models, not a ground for human distinctiveness.

Dimension 6: Purpose and Intention

CAS. Purpose is implicit in selection. Agents behave as if they had goals because rules that serve fitness survive. Anticipation through internal models gives a thin intentionality: agents act now in light of predicted futures. But the goal is fitness or utility, specified by the model.

Anthro-complexity. Humans act with intention, towards imagined futures that are open-ended, revisable and not reducible to fitness or utility. Teleonomy, Mayr's goal-direction from a programme, sidesteps human will and so misses what is distinctive.

Assessment. Stated as a property of individuals, the difference is again one of degree in one respect and of kind in another. Of degree, because CAS agents with internal models do anticipate, and richer models anticipate more richly. Of kind, because human intentions are not specified by any fitness function. They are formed, revised and abandoned in ways no model can specify in advance, since they respond to imagined futures that are themselves products of imagination. But here too the difference of kind is relational, as Chapter 27 will argue through Mead. Human intentions are formed in the conversation of gestures. An intention is not a private state that precedes action and then expresses itself. It takes shape in the response others make to one's gestures, and its meaning is completed by that response. Intentions are social before they are individual. Their open-endedness is the open-endedness of a conversation, not of a private imagination.

Location. Stated as agent-located. Of kind when relation-located.

Verdict: partial.

Dimension 7: Reflexivity

CAS. First-order reflexivity is modelled: agents' predictions affect the outcomes they predict (El Farol, artificial stock markets). Second-order reflexivity, in which agents change the rules of the game by re-describing it, is rarely modelled, because modelling it would require the model to include its own rules as variables the agents can change.

Anthro-complexity. Humans create abstractions and shared narratives that change the very constraints they operate within. Only humans can change the story they tell about the other. Any framework applied to a human system becomes an actor within it.

Assessment. This is the dimension on which the difference is most clearly of kind, and it is relation-located. Second-order reflexivity cannot be reached by enriching first-order reflexivity. An El Farol agent with better prediction rules is still predicting attendance at a bar whose threshold is fixed. A human who proposes moving the music to Fridays, or opening a second bar, has changed the problem. No refinement of prediction rules gets you there, because the move is not a prediction. It is a re-description. And re-descriptions change constraints only when they are taken up, which is a relational fact about whether others accept the story.

This is also the dimension on which the human distinction is strongest and on which anthro-complexity is most exposed. If second-order reflexivity is what makes human complexity different in kind, then every description of a human system, the practitioner's and the theorist's included, is a potential re-description that changes the system's constraints. Anthro-complexity's frameworks are not exempt. A theory grounded in second-order reflexivity must be reflexive about itself.

Location. Relation-located. This is the ground.

Verdict: holds as a difference of kind, with consequences for anthro-complexity's own position.

Dimension 8: The Ontology of Relations

CAS. Interactions are typically pairwise or local, and system behaviour is the aggregate of many local interactions. Network models add structure but keep the dyadic edge as the unit of relation.

Anthro-complexity. Relations involve entanglement and mediation. Knowledge can exist only in the interaction of people, artefacts and context (the entangled third of the trialectic). Peirce's thirdness, the mediation that cannot be built from pairs, is needed to account for meaning and for how patterns stabilise and change.

Assessment. A difference of kind, if Peirce's reduction thesis is right that triadic relations cannot be composed from dyadic ones. That is a contested claim in logic, but defensible. It is not specifically human, though. Peirce held thirdness to be a universal category, present wherever there is law, habit or regularity, including in the physical universe. The specifically human form of thirdness is semiosis, the sign relation with its interpretant. Semiosis is relational by definition, and it includes the interpreter. A human sign relation is triadic because it includes an interpretant, the effect on an interpreter. So Peirce's thirdness, when it grounds the human distinction, grounds it relationally: in the place of the interpreter within the relation, not in a property the interpreter carries outside it.

Location. Relation-located.

Verdict: holds as a difference of kind, with the human form relationally grounded.

Dimension 9: Adaptation

CAS. Adaptation is movement towards higher fitness on a landscape, or selection of better-performing rules. There is an implied optimum, or at least an implied gradient: better and worse configurations exist, and adaptation moves towards the better.

Anthro-complexity. There is no normal and no optimum. There is a current dispositional state and the plausible directions it can move in, and what counts as better is defined by people in the system, who may disagree and whose definitions favour whoever defines them.

Assessment. A difference of kind, and well grounded. Fitness in a CAS model is specified by the modeller. In a human system, what counts as fitness is contested by the agents, and the contest is part of the system's dynamics. An NK landscape has a fixed topography. A human "landscape" of possibilities is reshaped by the agents' arguments about which peaks are worth climbing. This difference is also relation-located. It concerns the reflexive relation between agents' evaluations and the evaluative landscape. And it carries the political charge Snowden identifies against eurhythmia: a fitness function, like a norm of health, favours whoever defines it.

Location. Relation-located.

Verdict: holds.

Dimension 10: Prediction and Simulation

CAS. Simulation is the method. Agent-based models are used to explore what patterns can emerge from specified mechanisms, to test policies in silico, and sometimes to forecast. "If you didn't grow it, you didn't explain it."

Anthro-complexity. Coherence only in retrospect. Simulation of human systems misleads, because the agents are not like human beings and the rules are not fixed. The method is to map dispositions, probe with safe-to-fail experiments, and sense the responses.

Assessment. A difference of kind, following from Dimension 7. If agents' descriptions of the system change its rules, and the model is one such description, then the model cannot simulate the system: the act of simulating, if published or acted on, changes what is being simulated. This is not true of non-human complex systems. A simulation of a starling flock does not change the flock. A published simulation of a human system can. That is performativity, which Chapter 34 examines through MacKenzie's study of the Black–Scholes model, which shaped the market it modelled.

But the same reasoning applies to anthro-complexity's alternatives. A map of dispositions, once published, is a description that changes the dispositions. A SenseMaker pattern, fed back to a population, changes the narratives it summarised. Mapping and sensing, as well as simulating, are performative in human systems. The difference between anthro-complexity and CAS on this dimension is real, but it is not the difference between a method that changes its object and one that does not. Both change their objects. The difference is that CAS has no theory of its own performativity, and anthro-complexity, which has the beginnings of one in Dimension 7, has not yet applied it to its own instruments.

Location. Relation-located.

Verdict: holds against CAS, and the same reasoning applies to anthro-complexity's instruments.

Dimension 11: The Position of the Knower

CAS. The modeller stands outside the model. She specifies the agents, rules and environment, runs the simulation, and observes what emerges. The model does not include her.

Anthro-complexity. Sensing is distributed. People in the system see the patterns themselves. Narrators self-signify. Maps are made by participants. In principle, no expert stands outside the system reading it.

Assessment. Here the difference is real in intent and incomplete in execution, and this is the dimension on which the argument of the whole book turns. The intent is a difference of kind: anthro-complexity aims to put the knower inside the system, where CAS keeps her outside. The execution keeps significant outside positions. Someone designs the signifiers, someone holds the dashboard, someone convenes the mapping, someone decides what the patterns are for. The Compass and the Loom named the root assumption of the two instruments it examined: read, then act from above. The reading may be distributed, but the acting-from-above has not gone away. The knower has been moved partly inside, without being fully inside.

Location. Relation-located (the position of the knower in relation to the known).

Verdict: partial. A difference of kind in intent, of degree in execution.

Dimension 12: Intervention and Its Ethics

CAS. Intervention is technical: tune rules, set incentives, find leverage points, change parameters. Its ethics, when considered, are those of policy: will the intervention produce good outcomes?

Anthro-complexity. Intervention targets constraints and conditions, not behaviour. It proceeds by safe-to-fail probes and amplifying what works. It is Pelagian, working with what the system has. And it is both possible and dangerous, because intentionality makes human systems responsive to intervention in ways that cannot be predicted.

Assessment. A difference of kind, and the most practically important one. An intervention that changes constraints and watches what emerges differs in kind from one that sets rules and predicts outcomes, and it embodies an ethic of humility the CAS approach lacks. The "possible and dangerous" clause is relation-located. Intervention in human systems is dangerous because people respond to the intervention as an intervention. They read it, interpret it, resist it, game it, and tell stories about it. A constraint changed in a human system is a gesture that calls out a response, and the response may defeat the purpose.

The ethical side of the difference has a consequence that anthro-complexity has stated (against Ing) and not yet fully applied (to itself). If intervention in human systems is dangerous because people read it, then the instruments of intervention, the signifier set, the map, the framework, are also read, and their reading is part of the intervention. The ethics of anthro-complexity's practice must include the ethics of its instruments: who designs them, who benefits from what they reveal, who consents to being sensed. The Compass and the Loom found these questions absent from the brochure vocabulary of the instruments it examined.

Location. Relation-located.

Verdict: holds against CAS. The ethics of the instruments is an open question.

The Twelve Dimensions Displayed

# Dimension CAS Anthro-complexity Kind or degree? Located in Verdict
1 Basic unit Agents Actants Kind General ontology holds
2 Driver of emergence Rules Constraints on dispositions Partly kind General ontology partial
3 The human One agent type Distinct kind (see 4–7) Stated agent, hinted relation —
4 Identity One tag at a time Multiple, switching Degree (agent); kind (relation) Stated agent partial
5 Cognition Rules, schemata Pattern, embodied Kind vs models; not human-specific Agent partial
6 Intention Fitness; thin anticipation Open-ended intention Degree (agent); kind (relation) Stated agent partial
7 Reflexivity First-order Second-order: re-describes constraints Kind Relation holds
8 Relations Pairwise, local Entanglement, thirdness Kind Relation (interpretant) holds
9 Adaptation Towards fitness No norm; contested good Kind Relation holds
10 Prediction Simulation Retrospect; probe Kind; applies to both Relation holds
11 Knower's position Outside Distributed (intent) Kind in intent; degree in execution Relation partial
12 Intervention Technical Constraint-shifting; dangerous Kind Relation holds

What the Comparison Shows

The table confirms the prediction this chapter started with, and it does so more sharply than expected.

Every dimension on which the difference between anthro-complexity and CAS is clearly of kind and specific to the human (7, 8, 9, 10 and 12) is relation-located. The difference rests on the reflexive relation between descriptions and the systems described, on the place of the interpreter within the sign relation, or on the contested and reflexive character of evaluation and intervention.

Every dimension on which anthro-complexity states the difference as agent-located (4 and 6, and in a different way 5) turns out to be of degree, or not specific to the human, unless it is reformulated relationally. Humans holding more identities than CAS agents is a difference of degree. Humans having identities that are produced in interaction is a difference of kind. Humans having richer intentions than CAS agents is a difference of degree. Humans forming intentions in the conversation of gestures is a difference of kind.

Dimension 11, the position of the knower, is where anthro-complexity's intent and its execution part company. It is the bridge to Book Four.

The chapter can therefore end with a sharper statement than the last one:

The comparison thesis. The differences between anthro-complexity and complex adaptive systems theory that are both of kind and specific to the human are all grounded in reflexive relations, not in properties of individual agents. Anthro-complexity's statement of the human distinction in terms of agent properties (the three marks) makes it look like a difference of degree, which the CAS theorist can absorb by enriching the agents. Its real strength, the asymmetry principle and second-order reflexivity, is relational, and it applies to anthro-complexity's own descriptions as much as to anyone's.

WHERE THE DIFFERENCE IS IN THE AGENT, IT IS ONE OF DEGREE —

WHERE THE DIFFERENCE IS IN THE RELATION, IT IS ONE OF KIND —

AND THE RELATION INCLUDES THE ONE WHO DESCRIBES IT.

Addition: The Twelve Differences as Cards

Each of the twelve dimensions of this chapter has a card. The front states the difference in a phrase. The back gives the difference, the disputation's reading of it, the verdict, and its connections to Book Nine.

Chapter 9

Ants, Starlings, Bees: The Non-Human Exemplars

4,832 words, about 21 minutes of reading

Good to Think With

Lévi-Strauss observed that animals are chosen as totems not because they are good to eat but because they are good to think with. A totemic system uses differences between animal species to think about differences between human groups. The eagle clan and the bear clan are distinguished as eagles and bears are distinguished, and the natural taxonomy lends the social one an air of givenness. The Compass and the Loom applied this to the animal-named role systems of the Cynefin Company's recent instruments, observing that role systems built on animal names look like natural taxonomies. The same applies, with more force, to the animal exemplars of complexity thinking: the ants, starlings, bees and fish that appear on slides, in brochures and in books as images of how organisations could work.

Animal exemplars are good to think with, and they are dangerous for the same reason. They lend a social proposal the authority of nature. If starlings coordinate without leaders, perhaps organisations could too. If bees choose nest sites by distributed sensing, perhaps teams could choose strategies the same way. If ants find the shortest path without any ant knowing it, perhaps markets, or algorithms, could find the best allocation without any planner. Each analogy may be illuminating. Each also imports, silently, everything that distinguishes the animal case from the human one, and presents the import as a lesson.

Anthro-complexity's slogan, humans are not ants, is a refusal of the import. This chapter examines the exemplars to see what the refusal rests on. It finds three things. First, the exemplars, read carefully in their own scientific literature, already contain much of what the slogan claims: the biology is more interesting and more cautionary than its management uses suggest. Second, management uses of the exemplars often reverse what the biology shows. Third, the most informative exemplar is not an animal at all but a human crowd that resembles one. Comparing the human crowd with the animal flock shows where the human difference really lies.

Ants: The Colony Without a Plan

What the biology shows

The ant colony is the oldest exemplar of complexity and the one Snowden names. Deborah Gordon's long-term studies of harvester ants in the Arizona desert, summarised in Ant Encounters (2010), are the most careful modern account of how colonies work. Gordon's central finding is that colony behaviour is regulated by rates of interaction. An ant does not receive instructions. It encounters other ants, briefly touching antennae, and the rate at which it meets ants of different task groups (foragers returning with food, patrollers returning from scouting) changes the probability that it will switch tasks or go out to forage. No ant knows how much food the colony needs or how many foragers are out. The colony's allocation of labour emerges from the statistics of encounters.

Gordon's work has two features that matter here. First, colonies differ. Some are more cautious, foraging less in dry conditions, and the differences are heritable and affect reproductive success. Colonies have something like dispositions, which arise from the interaction network and not from any individual ant. Second, Gordon has been explicit and persistent in warning against using ant colonies as models for human organisations. Ants, she has argued, have no individual identities in the relevant sense, no representations of the colony, and no capacity to change the rules of their interaction by discussing them. A human team that coordinated only by encounter rates would not be a team. The lesson of the ant colony, for Gordon, is not "organise like ants". It is that there are ways of regulating collective behaviour without central control that may suggest questions about human systems, provided one remembers how different the two are.

How the exemplar is used

In management literature the ant colony appears as an image of decentralised intelligence: the colony as a model for self-organising teams, ant colony optimisation as a model for logistics and routing, the "wisdom" of the colony as a model for crowdsourced decision-making. The trail-laying ant, whose pheromone trails strengthen with use so that the shortest path to food becomes the most travelled, is an emblem of how good solutions can emerge without design.

What the use reverses

The use reverses the biology in a precise way. The biology shows that colony behaviour emerges from interaction without representation. No ant represents the colony, its needs or its paths. The management use asks humans, who cannot help representing their organisation, to behave as if they did not: to follow local cues and trust the emergent outcome. That is not decentralisation. It is a demand that people suppress the very capacity, representing the whole and talking about it, that distinguishes them from ants. Self-organising teams, when they work, work because people do represent the whole and talk about it: they discuss priorities, negotiate tasks, argue about direction. The ant exemplar, used as a model, obscures this.

Snowden's slogan is the right response. Ants coordinate without representation. Humans cannot coordinate without it, because they represent whether they mean to or not. A model of human coordination built on ants leaves out the main mechanism.

Starlings: Coordination Without an Agent

What the biology shows

The starling murmuration is the most beautiful exemplar, and since the work of the STARFLAG project in Rome the most precisely measured. Ballerini and colleagues (2008), using stereo photography to reconstruct the three-dimensional positions of thousands of starlings in flight, found that each bird interacts with a fixed number of nearest neighbours, about six or seven, regardless of their distance. The interaction is topological, not metric. A bird responds to its six or seven nearest neighbours whether they are half a metre or five metres away. That makes the flock robust: when it stretches or compresses under attack by a predator, each bird keeps the same number of interaction partners, so cohesion is preserved.

Cavagna and colleagues (2010) went further and found that correlations in the birds' velocity fluctuations are scale-free. The correlation length grows with the size of the flock, so a change of direction by one bird is effectively transmitted across the whole flock, however large. This is the signature of a system near a critical point. The flock is poised so that information propagates without attenuation, which is why it can respond as a whole, almost instantly, to a predator on one side.

The Compass and the Loom draws the conclusion that matters: murmurations are a measured physical system in which coordination has no agent. There is no leader, plan or representation of the flock, and no bird decides where the flock will go. There is a topological coupling rule and a near-critical state, and from these the murmuration emerges.

How the exemplar is used

The murmuration is the most common image of complexity in organisational presentations. It stands for agility, collective intelligence, alignment without hierarchy and the ability to turn as one. Recent products, The Compass and the Loom notes, have gone further and suggested that human swarms, coordinated by sensing technologies, could achieve the murmuration's responsiveness.

What the use reverses

The use reverses the biology at the point of agency. In the murmuration, coordination has no agent, and that is what makes it a murmuration. A human group that turned as one would do so either because someone led and others followed (an agent), or because everyone shared a representation of where to go (a plan), or because each person read the others and adjusted (which, in humans, always includes representing what the others are doing and why). None of these is a murmuration. The Compass and the Loom draws the contrast in a phrase: human rooms, unlike flocks, are places where entrainment has an agent. When a room of people "falls in" behind a direction, someone set the direction, often the first confident voice. The swarm image hides that person.

The use also reverses the biology at the point of criticality. The flock's responsiveness depends on its being near a critical point, poised between order and disorder. In human organisations, criticality is the condition of panic, rumour and the cascade: a small signal spreading without attenuation through a population poised to respond. That is not agility. It is fragility. An organisation that propagated every fluctuation across the whole without attenuation would be at the mercy of every rumour. Human institutions are, to a large extent, devices for attenuating signals: committees, procedures, deliberation and delay. Some of them are wise for that reason.

Bees: The Democracy That Is Not One

What the biology shows

Thomas Seeley's work on house-hunting in honeybees, summarised in Honeybee Democracy (2010), is the most sophisticated account of collective decision-making in any non-human species. When a colony outgrows its hive, the queen and about half the workers leave and form a swarm that clusters on a branch. A few hundred scout bees then search for a new nest site. Scouts that find a promising site return and perform a waggle dance advertising it, with an intensity and duration proportional to its quality. Other scouts, recruited by the dance, visit the site and, if they agree, dance for it too. Sites compete for recruits.

The decision is not made by consensus. It is made by quorum. When the number of scouts at a candidate site reaches a threshold (about fifteen to thirty), those scouts switch from dancing to "piping", a signal that warms up the swarm for flight, and the swarm leaves for the chosen site. The decision is reached before all the scouts agree. Some may still be dancing for other sites when the swarm departs.

Seeley and colleagues (2012) also showed that scouts deliver stop signals, short head-butts accompanied by a vibration, to scouts dancing for rival sites. The stop signals inhibit competing dances, speed the decision and prevent deadlock between sites of similar quality. Seeley drew an analogy with the brain, where populations of neurons representing alternatives inhibit one another until one crosses a threshold. The swarm is something like a brain made of bees.

Once the decision is made, the swarm commits wholly to one site. There is no splitting and no minority faction setting off elsewhere.

How the exemplar is used

The bee swarm appears in management literature as an image of distributed decision-making, collective intelligence and "democracy". Seeley himself drew lessons for human groups in Honeybee Democracy: compose the decision-making group of individuals with shared interests and mutual respect; minimise the leader's influence; seek diverse solutions; aggregate knowledge through open debate; use quorum responses for cohesion, accuracy and speed.

What the use reverses

The Compass and the Loom found that the instrument it examined misreported the bee biology and reversed most of it: swarms choose by quorum, use stop signals, commit wholly to one site and decide before consensus. That brochure presented the swarm as an image of alignment through consensus, the opposite of what Seeley found. The swarm does not align. It inhibits alternatives until one reaches threshold, and then commits.

The reversal matters beyond a single brochure. The bee swarm, read correctly, is an image of decision under conflict. Alternatives compete, competitors are actively suppressed, and the process is designed to end debate, not prolong it. It is less like a democracy than like a well-run tournament with a sudden-death rule. Seeley's lessons for human groups are mostly about structuring the competition (diverse search, independent evaluation, a threshold rule) and not about achieving harmony. The Compass and the Loom reformulated the lesson as one of the Tidal Loom's principles: quorum with inhibition. Decide by threshold, institutionalise stop signals, then commit wholly, and seek a decision people can live with rather than consensus.

But the deeper point is that the bee lesson is useful for humans precisely because humans are not bees. A swarm's scouts do not form coalitions, tell stories about the other scouts' motives, resent the stop signals, or remember next year that their site was overruled. Humans do all of these. A quorum rule in a human group is a political institution. It distributes power, creates winners and losers, and will be gamed, contested and renegotiated. The bees give us a mechanism. Only humans can turn it into an institution, and once it is an institution it is subject to everything that makes human complexity different.

Cleaner Wrasse: The Fish That Manages Its Reputation

What the biology shows

The cleaner wrasse, Labroides dimidiatus, is a small reef fish that removes parasites from larger "client" fish at cleaning stations. The relationship is a mutualism with a temptation. Cleaners prefer the client's protective mucus to its parasites, but biting mucus makes the client jolt and may drive it away. Redouan Bshary and colleagues have shown in a long series of studies that cleaners manage this temptation in sophisticated ways. Bshary and Grutter (2006) found that cleaners behave more cooperatively, biting mucus less, when they are being watched by potential clients waiting nearby. This is an "image-scoring" effect: clients choose cleaners partly by watching them work, and cleaners appear to adjust their behaviour to the audience. Cleaners also treat clients differently according to their options. Clients with access to several cleaning stations get better service than resident clients with only one. And cleaners working in pairs punish partners who cheat.

More recently, Kohda and colleagues (2019) reported that cleaner wrasse appear to pass a version of the mirror test. Marked with a coloured spot visible only in a mirror, they try to scrape it off, a behaviour previously associated with self-recognition in great apes, elephants and some birds. The interpretation is contested, but the finding has made the wrasse a prominent case in debates about animal cognition.

How the exemplar is used

The Compass and the Loom examined the use of the cleaner wrasse as an exemplar in the Weaving Risk and Strategy materials (the name WRASSE itself appears to allude to it) and treated it, like the bees, as a borrowed animal model that needs scrutiny as a method metaphor. The wrasse stands for symbiotic relations between organisations and their environments, for the management of mutual dependence, and for adaptive cooperation.

What the wrasse teaches about the human distinction

The wrasse is the most interesting of the exemplars for this disputation, because it complicates the slogan humans are not ants in a way the other exemplars do not. The wrasse does something that looks very like reputation management. It behaves differently when watched, adjusting its conduct to the inferences an audience will draw. That is a form of reflexivity: the wrasse's behaviour is shaped by others' representations of its behaviour. If reflexivity is the human distinction, the wrasse seems to have some of it.

The adversary of anthro-complexity, the CAS theorist or the comparative biologist, has a strong case here. Audience effects, tactical deception, partner choice, punishment and possibly self-recognition are found in fish. Reputation, the shaping of behaviour by others' representations of it, is not a human monopoly. If the human distinction rests on reflexivity, it is a difference of degree, and the wrasse sits somewhere on the continuum between the ant and the human.

The adversary wins this half. First-order reflexivity, behaviour shaped by others' likely inferences about it, is not uniquely human, and anthro-complexity's slogan overstates the gap. What the wrasse does not do, and what no non-human animal is known to do, is change the rules of the cleaning market by re-describing it. No wrasse proposes a code of conduct for cleaning stations, forms a guild, tells clients a story about why mucus-biting is acceptable, or redefines what counts as "cleaning". The client fish cannot complain to a regulator. Second-order reflexivity, re-describing the constraints and so changing them, is the difference of kind, and the wrasse sharpens rather than blurs it. The wrasse shows that reputation is not the human distinction. The capacity to argue about what reputation should be for is. partial

The Gorilla: An Exemplar of Blindness

One more exemplar belongs here, though it is human, because it is used in the same way as the animal ones. The "invisible gorilla" of Simons and Chabris (1999) showed that observers asked to count basketball passes often failed to notice a person in a gorilla suit walking through the scene. Drew, Võ and Wolfe (2013) extended the paradigm to expert radiologists, inserting a small gorilla image into CT scans of lungs. Most radiologists searching for nodules failed to see it, though eye-tracking showed that many had looked directly at it. Only about one in six reported it.

The Compass and the Loom examined how this study was used in SwarmCompass, where it became a warrant for finding "the 17%" who see what others miss. That treatise argued that the gorilla anecdote adds a conformity episode the 2013 study does not contain, and that the study locates blindness in the task frame, not in individuals. Radiologists missed the gorilla because they were searching for nodules: the task set what they could see. The lesson is not that some people are seers and others blind. It is that the frame determines what is seen. The Tidal Loom's principle draws the consequence: frames, not seers. Rotate the search template, so that different people see different gorillas.

The gorilla study matters here because it is the human exemplar that comes closest to the non-human ones. It shows humans behaving in a way governed by a frame they do not choose and cannot see. In that respect they are like the ant following the pheromone trail or the starling responding to its six neighbours. But the frame is a human artefact. Someone set the task, chose the search target and designed the instrument. And the frame can be changed by changing the task, which an ant's trail or a starling's neighbourhood rule cannot. The gorilla study shows humans at their most ant-like and, in the same moment, shows the difference: the constraint is a description, and descriptions can be redescribed.

The Exemplar That Matters: Flash Mob and Murmuration

The Compass and the Loom sets beside the murmuration a human phenomenon that looks like one: the flash mob. The first flash mobs, organised by Bill Wasik in New York in 2003, gathered strangers by email at a specified place and time to perform a brief, pointless action. In one, a crowd gathered around a rug in Macy's and told a salesperson they were a commune looking for a "love rug". Then they dispersed. Wasik later described the project, in an essay of 2006, as an experiment on conformity and on the appetite for joining something for its own sake.

The flash mob and the murmuration look alike: a crowd that forms suddenly, moves together and disperses, with no visible leader. The comparison shows the difference. The murmuration has distributed, unreflexive agency. The flash mob has reflexive agency, because the participants know they are a mob. Each participant knows what a flash mob is, that she is taking part in one, that others know it too, and that the whole thing is a performance of spontaneity that was in fact organised. The flash mob is not like a murmuration. It is a representation of a murmuration, enacted by people who share the representation. Its coordination comes not from a topological coupling rule but from a shared script, "score, not algorithm" in The Compass and the Loom's phrase. A script is interpreted by bodies, while a murmuration has coupling, not code.

The flash mob also shows what the murmuration cannot: the seed of an institution. A flash mob has an organiser, the person who sent the email. The organiser's role is invisible in the performance but decisive in its existence. If the organiser speaks for the mob, explaining what it meant, calling the next one, or turning it into a brand, the mob begins to become something else: a movement, an organisation, a constituency. The Compass and the Loom, following Canetti, calls the organiser the crowd crystal, the small, rigid, constant group that precipitates crowds. It traces the thresholds (flash, refrain, crystal, representation, capture, Leviathan) by which a momentary crowd can become a state. Only the human collective can become a Leviathan. A flash mob carries the seed of one, once someone speaks for it. Chapter 42 will develop this fully.

What the Exemplars Teach

Exemplar What the biology shows How it is used What the use reverses What it teaches about the human distinction
Ant colony (Gordon) Interaction-rate regulation; colony dispositions; no representation Decentralised intelligence; self-organising teams Asks humans to coordinate without representing, which they cannot do Humans coordinate through representation
Starling flock (Ballerini, Cavagna) Topological coupling; scale-free correlation; no agent Agility; alignment without hierarchy Hides the agent of human entrainment; mistakes criticality for agility Human entrainment has an agent, the first confident voice
Bee swarm (Seeley) Quorum; stop signals; commitment before consensus Distributed decision; "democracy" Presents inhibition and threshold as alignment and consensus Human quorum rules are political institutions
Cleaner wrasse (Bshary) Audience effects; partner choice; punishment Symbiosis; adaptive cooperation Treats a mutualism as a model of strategy First-order reflexivity is not unique; second-order is
Gorilla (Drew and others) Task frame governs perception "Find the 17%" seers Locates blindness in persons, not frames Human constraints are descriptions, and can be redescribed
Flash mob (Wasik) Coordination by shared script (Not usually an exemplar) — Reflexive agency; the seed of an institution

The Lesson for Anthro-Complexity

The exemplars bear out anthro-complexity's slogan, humans are not ants, but they show that the slogan is right for a reason it does not state. Humans are not ants because humans coordinate through representations of their collective, and those representations can be redescribed. The capacities the slogan names (multiple identities, intention, abstraction) are part of this, but the decisive feature is that the representations are shared and contested. They exist between people, not inside them. And the redescription of a shared representation changes the constraints of the collective.

The exemplars also show something the slogan's users should notice. Every exemplar is read by someone. Gordon reads the ants, Seeley the bees, Bshary the wrasse, Ballerini and Cavagna the starlings. Each reading is careful, and each reader is outside the system being read. That is right for non-human systems: the ants do not read Gordon. The exemplars are good to think with because they can be read from outside without the reading changing what is read. The danger of carrying them into human systems is that the posture comes with them. The practitioner who reads a human organisation as Gordon reads a colony has adopted a position the human organisation does not allow, because the organisation reads back.

That is the lesson anthro-complexity's slogan points towards and does not quite draw. The problem with the ant exemplar is not only that humans are richer agents than ants. It is that the ant exemplar comes with an observer's posture that human systems do not permit. One can refuse the ant and keep the posture. Chapter 21 will argue that this is what anthro-complexity has done with its preferred exemplars, which are not animals but tides, estuaries and rivers.

THE ANTS DO NOT READ GORDON —

THE BEES DO NOT READ SEELEY —

THE ORGANISATION READS THE CONSULTANT.

Addition: The Alpha Wolf and the Captive Observer

One more exemplar belongs in this chapter, because it shows, more starkly than any other, how an observer's position can produce the very pattern it reports, and how the pattern, once reported, can travel into human organisations as a lesson from nature.

The alpha wolf

For much of the twentieth century, wolf packs were described as dominance hierarchies. At the top was an "alpha" male and female, who had fought their way to the top and maintained their position by aggression, with "beta" and lower-ranking wolves submitting to them. The model came largely from studies of captive wolves. Rudolph Schenkel's 1947 study of wolves in a Swiss zoo was influential, and L. David Mech's 1970 book The Wolf popularised the alpha terminology. In captivity, unrelated adult wolves were placed together in enclosures, and they fought for rank. The fighting was observed, and the dominance hierarchy was described.

Mech's later field research on wild wolves, especially on Ellesmere Island in the Canadian Arctic, led him to a striking conclusion, published in 1999 as "Alpha status, dominance, and division of labor in wolf packs". Wild wolf packs are not dominance hierarchies of unrelated adults competing for rank. They are families: a breeding pair and their offspring of the past one to three years. The so-called alphas are simply the parents. Their "dominance" is the authority of parents over their young, not the result of fights for rank. The aggressive struggles observed in captivity were an artefact of captivity itself, of placing unrelated adult wolves together in an enclosure from which they could not disperse, as young wild wolves naturally do when they mature. Mech spent years asking his publisher to stop printing The Wolf, because it perpetuated the alpha model he now believed was wrong.

The lesson for exemplars

The alpha wolf is an exemplar of the observer producing the pattern. The researchers did not invent the fights. They really happened. But they happened because of the conditions the researchers' institutions had created: unrelated adults, confined together, unable to disperse. The pattern was real in the enclosure and false of wolves. And the pattern, reported as a fact about wolves, travelled, into popular culture, into dog training (where "dominance" methods were justified by appeal to wolf packs), and into management and self-help literature, where "alpha" became a term for a type of person and a style of leadership.

The alpha wolf thus illustrates three of this book's theses at once.

Thesis II (the asymmetry principle applies to itself). The observers changed the system they observed. Not by observing it, since the wolves did not read the papers, but by the conditions under which they observed it. The description of the system included, unacknowledged, the effects of the describers' institutions.

Thesis III (humans are not estuaries). The wolves were not reflexive about being studied, but their captivity was a constraint the observers imposed, and the observed pattern was a response to it. The observer was inside the system, as the designer of its enclosure, while describing it as if from outside.

Thesis VIII (degradation and performativity). The pattern, once reported, was performative in human systems. Dog trainers who believed in dominance trained dogs by dominance, and produced dogs whose behaviour seemed to confirm the model. Managers who believed in alphas promoted and behaved as alphas, and produced organisations in which alpha behaviour seemed natural. The exemplar did not merely describe. It was an engine (Chapter 34).

Enclosures in organisations

The alpha wolf raises a question that every narrative instrument should ask: what enclosure are we observing in? An organisation is not a wild pack. It is a constructed environment, with its own enclosures: reporting lines, performance systems, open-plan offices, restructurings that place unrelated groups together, and the absence of a way to disperse when the conditions become intolerable. The patterns a narrative instrument observes in an organisation are patterns in that enclosure. Stories of conflict between merged teams, of political struggles for position, of toxic managers and fearful staff may be real, and they may also be artefacts of an enclosure the organisation's designers built and its instrument's commissioners maintain.

A reflexive practice would treat every pattern as potentially an enclosure effect, and would ask, before reading the pattern as a fact about people, what conditions produced it, and who set those conditions. Snowden's principle that behaviour is the shadow of conditions (Element 18) is the right principle here. Mech's correction shows how far it reaches. Even the behaviour that seems most like the nature of the creature may be the shadow of its enclosure, and the enclosure was built by the observer's own institution.

The exemplars revisited

Exemplar Observer's position What the position produced
Ant colony (Gordon) Long-term field observation Interaction-rate regulation; Gordon's own warning against human analogies
Starling flock (STARFLAG) Stereo cameras on a roof Measurement without disturbance; topological coupling
Bee swarm (Seeley) Field experiments with marked scouts Quorum and stop signals; reversed in management uses
Cleaner wrasse (Bshary) Reef observation and experiment Audience effects; first-order reflexivity in fish
Gorilla (Drew and others) Laboratory task with radiologists Frame governs perception; "find the 17%" misread
Alpha wolf (Schenkel; Mech) Captive enclosure of unrelated adults A dominance hierarchy that wild packs do not have

The alpha wolf completes the set. The starlings showed that some observation leaves its object unchanged. The alpha wolf shows that some observation produces its object, through the conditions under which it is made. Human organisations, which are reflexive, enclosed and observed by instruments their own leadership commissions, are much closer to the captive wolves than to the wild starlings.

THE FIGHTS WERE REAL — THE ENCLOSURE MADE THEM —

AND THE PATTERN, REPORTED AS NATURE, TRAINED THE DOGS AND PROMOTED THE MANAGERS —

ASK WHAT ENCLOSURE YOU ARE OBSERVING IN.

Chapter 10

Systems Thinking and the Taxonomic Move

4,677 words, about 20 minutes of reading

The Neighbour Snowden Names

The November post is about systems thinking, and its conclusion, complexity is not a kind of systems thinking, is the most quotable line in it. This chapter examines that conclusion. It sets out the systems lineage as Ing reconstructs it, the taxonomic move Snowden contests, and Snowden's three arguments against the move. It finds the arguments sound, and gives its verdict on Thesis 2: holds. It then asks a question the post does not ask. Is the taxonomic move something only systems thinkers make? Or does anthro-complexity make it too, in its own idiom, when it places other bodies of thought on its own map? And do Pepper's maxims, which Snowden turns against Ing, also cut against the eclecticism of anthro-complexity's own sources?

The Systems Lineage

Pepper's world hypotheses

Stephen Pepper's World Hypotheses (1942) proposed that metaphysical systems grow from root metaphors, basic analogies drawn from common-sense experience and then extended to cover everything. Pepper identified four that he judged relatively adequate:

  • Formism, whose root metaphor is similarity. Things are instances of forms or types. Plato and Aristotle.
  • Mechanism, whose root metaphor is the machine. Things are configurations of parts pushing on parts. Democritus, Descartes, Hobbes.
  • Contextualism, whose root metaphor is the historic event, the act in its context. Things are events whose quality and texture depend on the situation in which they occur. James, Dewey, Mead.
  • Organicism, whose root metaphor is the organism, or the integration of fragments into a coherent whole. Hegel, Bradley, Royce.

Formism and mechanism are analytic: they begin with elements and build wholes. Contextualism and organicism are synthetic: they begin with complexes or wholes. Formism and contextualism are dispersive: they do not insist that the world forms a single system. Mechanism and organicism are integrative. Pepper later added a fifth hypothesis, selectivism, built on the root metaphor of the purposive act, in Concept and Quality (1966).

Pepper's maxims, which both Ing and Snowden accept, follow from the theory. Each world hypothesis is autonomous and should be judged by its own evidence, with its scope and precision. Eclecticism, the mixing of root metaphors, is confusing, because metaphors from different roots carry incompatible commitments. And the best available attitude is a "reasonable" multiplicity: hold several hypotheses at once, each in its own domain, without fusing them.

Contextualism and the causal texture

Ing's paper places Pepper's contextualism behind Emery and Trist's work on the causal texture of organisational environments. In their 1965 paper, Fred Emery and Eric Trist proposed that organisational environments differ not only in content but in texture: the way their elements are causally connected. They distinguished four ideal types:

  1. Placid, randomised. Goods and bads are randomly distributed and relatively unchanging. Simple tactics suffice.
  2. Placid, clustered. Goods and bads cluster in predictable ways. Strategy, as distinct from tactics, becomes possible.
  3. Disturbed-reactive. Several organisations of the same kind compete in the same environment, so each must take account of the others' moves. Operations and campaigns become necessary.
  4. Turbulent fields. The dynamics arise from the field itself, not just from the interaction of identifiable organisations. The ground is in motion. Individual organisations cannot adapt successfully by their own actions alone, and new forms of collaboration and shared values become necessary.

Snowden calls the turbulent field one of the more important contributions from the systems tradition, and he is right. It anticipates much of what complexity thinking would later say about environments that cannot be mapped as arrangements of separate actors. Ing's tracing of it to Pepper's contextualism is, Snowden says, a useful piece of intellectual history.

Singer, Churchman and Ackoff

The second pragmatist lineage runs from Edgar Singer, a student of William James who taught at Pennsylvania, to C. West Churchman and Russell Ackoff, who studied with him. Singer's "experimentalism" coupled inquiry with ethics. The pursuit of knowledge was part of the pursuit of ideals, and the progress of inquiry was progress towards a better world. Churchman's The Design of Inquiring Systems (1971) and his later work on the "systems approach" developed Singer's vision into a philosophy of design in which the systems thinker had to "sweep in" ever more of the context, the boundary judgements and the ethical commitments of an inquiry. Ackoff's interactive planning, his "idealised design" and his definitions of system, purpose and behaviour turned it into a management practice. In Snowden's summary, this is a pragmatism of worldviews and of planning.

The three eras

Ing's paper organises the history of systems thinking into eras. Mechanism dominates the 1950s and 1960s: cybernetics, operations research, systems engineering. Social systems dominate the 1970s and 1980s: soft systems methodology, critical systems thinking, the sociotechnical tradition. Ecology dominates the 1990s and 2000s: resilience thinking, panarchy, social-ecological systems. In the ecological era Holling's adaptive cycle, Ostrom's institutional analysis of the commons and the social-ecological systems literature appear as the latest phase of systems thinking. Rosen's relational biology is brought in to supply a theory of truth. Beyond the ecological era Ing proposes his own hypothesis, (con)texturalism-dyadicism, with texture as root metaphor and dyadic logic drawn from Chinese medicine.

The Taxonomic Move

In the lecture

In the Barcelona lecture, Ing takes the traditions of systems thinking from Ramage and Shipp's Systems Thinkers: early cybernetics, general systems theory, system dynamics, soft and critical systems, later cybernetics, learning systems, and complexity theory (Prigogine, Kauffman, Lovelock). He favours three: general systems theory, soft and critical systems, and learning systems. Complexity is not among them. Snowden adds that Ramage and Shipp are, in his opinion, referenced too often, and that their framing does not carry the authority people give it. In the postscript he notes that the map is Ramage and Shipp's, not Ing's, and that his objection is to its adoption and the consequences.

In the paper

The paper makes the same move by narrative. Complexity science as such, with its origins in physics, chemistry and theoretical biology and its specific claims about emergence, constraints and dispositionality, does not appear. Its vocabulary appears in other forms: resilience in Holling, propensity in Jullien, entailment in Rosen, turbulence in Emery and Trist. Each term is present, and the science from which it came is absent.

Two forms of subsumption

Snowden distinguishes two forms. Complexity is "placed on the map and then set aside", which is the lecture's move. And it is "absorbed piecemeal into a systems narrative", which is the paper's move. In the first, complexity is acknowledged as a tradition and declined. In the second, its concepts are adopted without its ontology. Snowden judges that neither holds up.

Snowden's Three Arguments

First: origins

Complexity science comes from the natural sciences. Its key figures were not working within a systems-thinking tradition aimed at management or social intervention. Making Prigogine a systems thinker in the sense that Ackoff or Checkland were is a choice made by the map, not something the work itself claims. Systems thinking has a different lineage, which the paper sets out carefully: operational research, Tavistock social psychiatry and American pragmatism. Snowden adds that systems thinking was hugely valuable in its time, and that he spent part of his own career with Frank Land and others trying to make it work. The issue is that the science has moved on, not that systems thinking was wrong.

Assessment. The argument from origins is sound as far as it goes, and it is the right kind of argument against a genealogical narrative. If the paper says the systems lineage runs through mechanism, social systems and ecology, it is fair to point out that the ecological era's sources (Holling's adaptive cycle, built on the mathematics of nonlinear dynamics, and Rosen's relational biology) came from a different lineage and were absorbed into the systems narrative. Prigogine did not think of himself as a systems thinker in Ackoff's sense. Kauffman did not either. Their work was physics and biology. It addressed management, if at all, only by analogy. The genealogical claim that they belong to the systems tradition is a reconstruction, not a discovery.

There is a limit, though. Origins do not settle identity. A tradition can absorb concepts from elsewhere and make them its own, as anthro-complexity has absorbed Peirce, Latour and Jullien. The argument from origins shows that complexity science did not come from systems thinking. It does not show that systems thinking cannot legitimately take up complexity's concepts. That needs the other two arguments.

Second: scope

If systems thinking is a perspective on parts, wholes and their relations, as the lecture defines it, almost any body of thought that concerns itself with relationships can be included. A definition that broad gains coverage at the cost of the ability to make distinctions.

Assessment. This argument is decisive. A definition that includes everything distinguishes nothing. If systems thinking is any concern with parts, wholes and relations, then Aristotle, Hegel, Marx, Durkheim, Darwin and the Buddhist doctrine of dependent origination are all systems thinkers. The category has stopped picking anything out. The lecture's definition is generous, and generosity is admirable in a pedagogue. But it cannot support the claim that complexity is a phase of systems thinking, because on that definition everything is.

Third: Pepper's maxims

Each world hypothesis is autonomous, should stand on its own evidence, and should not be mixed with others. Snowden accepts all three maxims and argues that they apply to complexity as much as to anything else. Complexity is not a perspective one adopts. It is a claim about the nature of certain systems, with its own commitments, and those commitments conflict with key assumptions of the traditions it has been folded into. Taking complexity seriously means treating it as autonomous, not absorbing its vocabulary while leaving its ontology behind.

Assessment. This is the most elegant of the three, because it turns the interlocutor's own resources against him. If Ing accepts Pepper's maxims, he must accept that absorbing complexity's vocabulary into a systems narrative is a form of eclecticism, and Pepper says eclecticism is confusing. The "commitments that conflict" are stated in the three areas of conflict that follow: wholes presumed versus found, the human set aside versus placed at the centre, dyad versus triad. Each is a point where complexity's ontology and the systems traditions' ontology diverge. Absorbing one into the other hides the divergence.

The institutional story

Snowden adds an institutional story. Systems thinkers were always more at home in large organisations than complexity scientists, other than those safely confined to the laboratory. Systems thinking offers structure: models, boundaries, purposes, and a method that can be taught and rolled out. Corporations like that. Snowden lasted seven years at IBM before he became, in his word, an irritant. That is not a judgement, he says, but it may say something about which body of thought an institution finds easy to host, and why complexity has so often been absorbed into systems thinking rather than met on its own terms.

Assessment. The institutional story is the most interesting part of the argument for this disputation, because it is a complexity argument about the history of ideas. It says that the absorption of complexity into systems thinking is not only an intellectual error but an emergent pattern, shaped by the constraints of institutional life. Institutions select for ideas that offer structure, and complexity, offering less, is reshaped to fit. That is behaviour as the shadow of conditions, applied to intellectual history. It is persuasive. Chapter 17 will apply it to anthro-complexity's own institutional history.

Verdict on Thesis 2

Thesis 2. Snowden is right that complexity is not a kind of systems thinking. The taxonomic move that folds Prigogine, Kauffman and Holling into a systems lineage buys coverage at the cost of distinction. holds

The argument from origins shows that the genealogy is a reconstruction. The argument from scope shows that the definition supporting the reconstruction distinguishes nothing. The argument from Pepper's maxims shows that absorbing complexity's vocabulary without its ontology is the eclecticism Ing's own authority warns against. The institutional story explains why the move is so common: institutions find structure easier to host than complexity.

The Mirror: Does Anthro-Complexity Make the Taxonomic Move?

The verdict is firm, and that makes the next question necessary. A disputation that judges another's move must ask whether its own side makes the same move.

Cynefin as a map of traditions

Cynefin is a framework for choosing frameworks. Its domains (clear, complicated, complex, chaotic) are kinds of situation, and each has an appropriate kind of response. Snowden has often placed other bodies of thought within the framework. Systems thinking, in many of his writings, is appropriate to the complicated domain, where analysis by experts can establish cause and effect. Six Sigma and process engineering belong in the clear domain. Scenario planning, when it pretends to predict, belongs in the complicated domain and fails in the complex. In this use Cynefin is a map of traditions. It places each body of thought in the domain where it is valid, and it reserves the complex domain, the domain of emergence and probe-sense-respond, for complexity thinking.

This is a taxonomic move. It places systems thinking on a map and then, for the situations that matter most, sets it aside. Systems thinking is valid in the complicated domain. In the complex domain it misleads. The structure is the same as Ing's lecture, which places complexity on Ramage and Shipp's map and then favours other traditions. Each places the other in a bounded region of its own map, and each reserves the central region for itself.

Is the mirror fair?

The mirror is not entirely fair, and the unfairness should be noted. Cynefin's placement of systems thinking in the complicated domain is not a claim about genealogy. It does not say systems thinking is a phase of complexity thinking or descends from it. It says systems thinking's methods are valid for a certain kind of situation. That is closer to Pepper's reasonable multiplicity, each hypothesis in its own domain, than to Ing's narrative absorption. Snowden can reasonably say that Cynefin respects the autonomy of systems thinking by giving it a domain of validity, where Ing's narrative absorbs complexity without giving it one.

But the mirror is fair in one respect, and the respect matters. Cynefin's map is drawn from inside complexity thinking. The domains are defined by complexity's own concepts: the relation of cause and effect, the presence or absence of constraints, the possibility of prediction. Placing systems thinking in the complicated domain is therefore a judgement made by complexity about systems thinking, in complexity's terms. A systems thinker might well reject it. Checkland's soft systems methodology, for instance, was explicitly designed for "messy" situations of conflicting worldviews, which Cynefin would place in the complex domain, not the complicated. Assigning soft systems methodology to the complicated domain is a taxonomic move that the tradition so placed has not agreed to. It is a map drawn by one neighbour of another neighbour's territory.

That is exactly the move Snowden objects to in Ing. Ing's map is drawn from inside systems thinking, in systems thinking's terms, and it places complexity where complexity has not agreed to be placed. The two moves are mirror images.

Proposition granted: A map of traditions drawn from inside one tradition, in its own terms, misplaces the others.

Then it must be granted that Ing's map, drawn from inside systems thinking, misplaces complexity. — Granted. This is Snowden's argument.

Then it must be granted that Cynefin's map, drawn from inside complexity, misplaces systems thinking, insofar as it assigns systems thinking to a domain systems thinkers would not accept. — Granted, with the qualification that Cynefin's map gives systems thinking a domain of validity and does not absorb it.

Then it must be granted that the relation between complexity and systems thinking cannot be stated from inside either. — Granted. That is what Snowden's tree, rock and water picture: three actants, none containing the others.

Then it must be granted that the picture is drawn from outside both. — Doubted. The picture is drawn by Snowden, in the November post, as a move in the argument between complexity and systems thinking. It is drawn from inside complexity, as the complexity theorist's account of the relation. Systems thinking has not agreed to it either. Ing might say textures, not neighbours.

Conclusion. There is no outside position from which the relation between two traditions can be drawn. Every map of the relation is a move in it. This does not make the maps arbitrary. Some are better grounded than others, and Snowden's is better grounded than Ramage and Shipp's. It does make them gestures, whose meaning will be completed by the response they call out.

Pepper's Maxims Turned Again

Pepper's maxims cut in a second direction, and here the mirror is sharper. Snowden uses them to argue that complexity must be treated as autonomous, not absorbed into a systems eclecticism. But anthro-complexity's own sources are diverse:

Source Root metaphor or tradition What anthro-complexity takes
Peirce Contextualism/selectivism; semiotic realism Signs, abduction, habit, thirdness
Latour Actor-network theory; flat ontology Actants
Juarrero Complexity; dynamical systems; philosophy of action Constraints
Deutsch & Marletto Physics; constructor theory Constructors
Jullien Comparative philosophy; Chinese strategic thought Propensity, 勢
Rosen Relational biology Retrospective entailment
Prigogine Thermodynamics Far-from-equilibrium order
Gibson Ecological psychology Affordances
Ingold Ecological anthropology Meshwork (with reservations)
Polanyi Philosophy of tacit knowing Tacit knowledge
Barad (rejected) Agential realism "Entangled" (in a different sense)
Aristotle; Scott Virtue ethics; anarchist political anthropology Phronesis; mētis

Is this eclecticism in Pepper's sense? Several of these sources have incompatible root metaphors. Latour's actants were defined to avoid presuming a hierarchy of agency, and anthro-complexity restores one. Constructor theory's constructors are defined by being unchanged, and anthro-complexity's human actants are defined by being changed. Gibson's affordances are continuous across species, and anthro-complexity insists on human distinctiveness. Ingold's meshwork denies presumed parts, and anthro-complexity's human cluster presumes persons. In each case anthro-complexity takes a term and modifies its commitments. Snowden accuses the systems narrative of doing exactly this to complexity: absorbing its vocabulary while leaving its ontology behind.

The adversary's reply is strong, and it deserves its box.

Anthro-complexity is not eclectic in Pepper's sense, because it has a single ontology, complexity, into which its sources are integrated. It takes from each source only what the complexity ontology can use, and it modifies what it takes so that the result is consistent. Latour's actant is taken without his flatness, because complexity's ontology needs non-human agency and does not need flatness. Constructor theory is taken as a model of persistent transformative capacity, not as a theory of unchanged agents. Gibson's affordances are taken as the relational structure of possibility, not as a denial of human distinctiveness. This is not eclecticism. It is what every serious body of thought does with its sources. Pepper himself acknowledged that the world hypotheses borrow from one another.

The adversary wins this half too. Anthro-complexity is integrated by its complexity ontology in a way Ing's narrative of eras is not, and the borrowing it does is the kind any living tradition does. But the reply concedes the crucial point: anthro-complexity modifies what it takes. The modifications are where the human cluster is lodged: Latour without flatness, constructors applied to rituals, Gibson with the human distinction restored. Book Four will argue that the modifications are not neutral. Each is a point where anthro-complexity restores a figure of the human (the narrator, the intender, the abstractor) that the source was designed to dissolve. Pepper's warning applies at exactly those points. The root metaphor of the source and the root metaphor of the modification conflict, and the conflict is hidden in the vocabulary. partial

The Root Metaphor of Anthro-Complexity

The chapter ends with a question Pepper would ask. If anthro-complexity is a world hypothesis, or close to one, what is its root metaphor?

Not the machine, which it rejects, nor the organism, whose wholes it refuses to presume. Not similarity or form, which it finds too static. It is close to contextualism, the historic event in its context, and Snowden's Peircean pragmatism places it there, though he resists Ing's listing of Peirce as a contextualist. But anthro-complexity has a more specific image, which recurs throughout the November post and the wider corpus, and which functions as a root metaphor in Pepper's sense: an image from common experience extended to cover everything. The image is the estuary: the meeting of river and sea, governed by the turn of the tide, whose constraints have different energies and timescales, in which a sailor waits for the flow to turn and then moves. Around it cluster the tide, the river, the wave, the surfer, the sailor, the waterfall and the rock.

If that is anthro-complexity's root metaphor, Pepper's theory makes a prediction. The metaphor will carry commitments into every domain it is extended to, including the human. Those commitments will be the estuary's, not the human's: a medium that does not read the reader, and a reader who stands on the bank or the deck and reads the flow. Chapter 21 will test the prediction.

EVERY MAP OF NEIGHBOURS IS DRAWN BY A NEIGHBOUR —

EVERY ROOT METAPHOR CARRIES ITS SOIL —

AND THE ESTUARY'S SOIL IS SILT, NOT SPEECH.

Addition: Ing's Programme Given Its Due

A disputation that criticises one party for misplacing another owes the third party fairness too. David Ing's systems-changes programme appears in this book mainly as the occasion of Snowden's post and as the target of the taxonomic move. This addition gives it its due. It states what the programme offers that anthro-complexity, and this disputation, might learn from. It also notes where the programme already anticipates some of the disputation's own criticisms of anthro-complexity.

What Ing's programme gets right

Texture over structure. Ing's root metaphor is texture, the weave of strands over time, and not structure or organism. Texture is a relational image. It foregrounds how things are woven together, not what they are made of or what whole they compose. Snowden credits Ing's paper with rejecting the containing whole in favour of wholes alongside wholes in a texture over time. In this respect Ing's programme is closer to Ingold's meshwork (Chapter 30) than anthro-complexity's actants are. The disputation's reflexive meshwork, and The Compass and the Loom's Tidal Loom, are textural images. They are Ing's root metaphor, turned to anthro-complexity's purposes.

Rhythm and time. Ing's emphasis on processes as rhythms and on eurhythmia, healthy rhythm relative to context, brings time into systems thinking in a way that most of the systems tradition has not. Snowden welcomes the preference for kairos over chronos. The disputation would add that rhythm, unlike the tide, is something participants make together. A rhythm is kept by those who keep it, as dancers keep time, and it is not imposed by a moon. Eurhythmia, read that way, is a relational concept of timing that the estuary lacks (Chapter 21).

The medical model as self-healing. Snowden revised his first reading on this point: the paper prefers self-healing through internal adjustment to invasive external intervention. Classical Chinese medicine, as Ing draws on it, does not treat diseases as entities to be removed. It reads the body's configuration and helps it return to its own balance. Law and Lin's reading of Chinese medicine (Chapter 35) stresses that the practitioner is inside the configuration and that her intervention changes it. That is a reflexive model of practice, not the Augustinian one Snowden first feared.

The dyad, carefully. Ing's insistence that dyadic is not dualistic, that yin and yang are context-dependent, complementary and mutually transforming, not ranked opposites, is, as Snowden says, well made and a useful corrective. Chapter 38 argued that the Chinese sources contain mediating thirds that Ing's dyadicism may miss. But the dyad, read as Ing reads it, is not the binary Snowden fears. Yin and yang are not two sides to be taken. They are two modes of one process, each containing the seed of the other.

Attention. Ing's phrase attention trumps intention is the one Snowden contests most sharply, and the disputation has sided with Snowden that intention matters. But the phrase carries a truth. What people attend to shapes what they can intend. Snowden himself says that much of what he does is about changing what people attend to. The salience of the 3Ss in The Compass and the Loom is attention. Ing's emphasis on attention is, read generously, an emphasis on the conditions of intention, on the frames within which intentions form. That is close to the disputation's account of intention as shared and formed in context (Chapters 13, 29).

Where Ing anticipates the disputation

Several of the disputation's criticisms of anthro-complexity are anticipated, in different terms, by Ing's programme.

Disputation's criticism of anthro-complexity Ing's anticipation
Possessive grammar of the human (Thesis I) Ing's Ingoldian organism-person, grown in an environment, not a bearer of properties
Estuary places reader outside (Thesis III) Ing's textures have no outside; the weaver is in the weave
Constructors unchanged by their work (Thesis IV) Ing's processes as rhythms: everything changes as it continues
Instruments as steersman (Ch 6) Ing's preference for self-healing over intervention

The anticipations are partial. Ing does not frame them as criticisms of anthro-complexity, and his programme has the weaknesses Snowden identifies: synthetic reasoning, a norm of health, the physician's reading and the downplaying of human distinctiveness. But they show that the disputation's argument is not unique to its author or to CRP. A careful systems thinker, drawing on Ingold, Gibson and Chinese medicine, has arrived at several of the same places.

Where the disputation sides with Snowden against Ing

The disputation's verdicts on the November post's three conflicts can be stated plainly, now that Ing has been given his due.

Wholes, presumed or found. Snowden is right that synthetic reasoning presumes wholes, and that anthro-complexity's refusal to presume wholes or parts is the stronger position. The disputation has argued only that anthro-complexity should extend the refusal to persons (Thesis I). holds for Snowden.

Humans, animals and intention. Snowden is right that a theory which sets intention aside cannot account for human systems' difference from ecological ones. The disputation has argued only that the difference lies in relation, not in possession. holds for Snowden, with the disputation's qualification.

Dyads and triads. Partial for both. Ing's dyadicism misses the Chinese sources' mediating third, and Snowden's reading of the trigrams misses their positional structure (Chapter 38). partial.

Eurhythmia and the norm. Snowden is right that a norm of health favours whoever defines it. The disputation adds that Ing's eurhythmia, read as a rhythm participants keep together, is less vulnerable to this than a norm imposed by a physician. partial.

Neighbours

The November post ends by calling complexity and systems thinking neighbours who borrow from each other. This addition has tried to show what anthro-complexity might borrow from Ing in particular: texture as a relational root metaphor, rhythm as something kept together, and a model of practice from Chinese medicine in which the practitioner is inside the configuration. Each borrowing would help anthro-complexity with exactly the gaps the disputation has identified. That is what neighbours are for.

TEXTURE, NOT STRUCTURE — RHYTHM KEPT TOGETHER, NOT A TIDE THAT TURNS ALONE —

THE PHYSICIAN INSIDE THE PULSE —

NEIGHBOURS BORROW; HERE IS WHAT TO BORROW.

Addition: Cross-Lights From Book Nine

The series continues. The day after the tree post, Snowden published a comparison with another neighbour, the sociology of Niklas Luhmann. Its method is the one this chapter describes: common ground first, then what he calls the different geology beneath. Chapter 52 examines it.

A rule for borrowing. That post says that a concept moved from one field to another needs its own justification in the new one, and calls one pairing of thinkers misappropriation. The rule bears on this chapter's subject, since the dispute with Ing is partly over who may take whose concepts.

Boundary critique re-based. This chapter credits the systems tradition with its questioning of boundaries. Chapter 56 argues that nothing is bounded in itself and that the questions apply to boundaries someone has drawn. They keep their use.

Chapter 11

Complex Responsive Processes: The Neighbour Not Named

4,340 words, about 19 minutes of reading

The Absent Interlocutor

The November post names its interlocutor, Ing, and its neighbours: systems thinking, complexity science, critical systems thinking in the postscript. It does not name the body of thought that has done more than any other in the English-speaking world to work out what complexity means for human organisations as human. That is the complex responsive processes perspective, developed by Ralph Stacey, Douglas Griffin and Patricia Shaw at the University of Hertfordshire from the late 1990s, and carried on by Chris Mowles, Kevin Flinn and the graduates of the Hertfordshire Doctorate of Management.

The omission is not a fault. A post has a subject, and this one's is Ing. But it is significant. Complex responsive processes (CRP) made the move Snowden makes against Ing, setting human complexity apart from systems thinking and from CAS, twenty-five years ago, and grounded it differently. Where anthro-complexity grounds the human distinction in properties of persons (identities, intentions, abstractions), CRP grounds it in relations between persons: the conversation of gestures, the figuration of power, the paradox of the individual and the social. Where anthro-complexity builds instruments, CRP has largely refused to. Where anthro-complexity keeps a practitioner who designs interventions, CRP insists there is no position outside the conversation from which interventions could be designed.

This chapter sets out CRP as the neighbour not named, because the argument of Book Four draws heavily on its resources, and because the disagreement between CRP and anthro-complexity, which has a history, is the clearest available test of where the human distinction lies.

From CAS to Complex Responsive Processes

Stacey's turn

Ralph Stacey's Complexity and Creativity in Organizations (1996) was one of the first serious attempts to bring the Santa Fe programme into organisational theory. It drew on CAS theory, agent-based models and the edge of chaos, and argued that creativity in organisations arises at the boundary between stability and instability. The book was influential, and Stacey soon turned against its approach.

The turn is documented in Complexity and Management: Fad or Radical Challenge to Systems Thinking? (2000), written with Griffin and Shaw, and in Stacey's Complex Responsive Processes in Organizations (2001). Its core is a critique of the idea that organisations are complex adaptive systems. A CAS is a model, a population of agents in a computer whose rules are specified by a modeller. To say an organisation is a CAS is to say it can be described as such a population, and that the people in it are agents whose rules a modeller could specify. Stacey came to think this was false and harmful. It was false because people are not rule-following agents. It was harmful because it smuggled back the manager-as-modeller, a figure outside the system who designs its rules, whom complexity was supposed to have shown to be impossible.

The five teleologies

The most powerful instrument of the critique is the analysis of teleologies in Complexity and Management. A teleology, in Stacey, Griffin and Shaw's sense, is an assumption about what movement into the future is for: what drives change, and towards what. They identify five.

  1. Secular natural law teleology. Movement follows timeless laws, and the future is a repetition or unfolding of the past according to those laws. Newtonian mechanics.
  2. Rationalist teleology. Movement is towards goals chosen by autonomous, reasoning individuals. The future is what rational agents design and then bring about. Kant's moral agent, and the manager who sets strategy.
  3. Formative teleology. Movement unfolds what is already enfolded. The future is the realisation of a mature form implicit in the beginning, as the oak is implicit in the acorn. Kant's account of organisms, and systems thinking's account of organisational development.
  4. Adaptionist teleology. Movement is adaptation to an environment through selection. The future is whatever survives. Neo-Darwinism.
  5. Transformative teleology. Movement is the continual iteration of identity and difference, producing the genuinely new, which was not enfolded in the past and is not chosen by any individual. The future is under perpetual construction in local interaction, and no one stands outside it. Hegel's dialectic read through Mead and Elias.

The analysis carries a diagnosis, and The Compass and the Loom applies it to the instruments it examines. Most management theories combine two teleologies without noticing: the organisation is formative, unfolding what is enfolded, while the manager is rationalist, choosing and steering. The text calls this the signature of systems thinking. The organisation develops according to its nature, and the manager, outside it, designs the conditions for its development. Stacey and his colleagues called it dual or split causality. Different causal principles are applied to the system and to the person who acts on it, and the person is exempted from the causality that governs the system.

CRP's alternative is transformative teleology applied without exemption. The organisation is not a system unfolding a form. It is a pattern of interaction among people, continually reproduced and transformed in local interaction. The manager is not outside it choosing. She is one of the people interacting, and her choices are gestures in the conversation, whose meaning is completed by others' responses. Every chooser is chosen too, as The Compass and the Loom puts it.

Mead and Elias

CRP's sources are not the natural sciences but social psychology and sociology: George Herbert Mead and Norbert Elias, read together with the complexity sciences as an analogy.

From Mead, CRP takes the conversation of gestures. A gesture is an act that calls out a response in another, and the meaning of the gesture is not in the gesturer's intention but in the social act of gesture and response together. When humans become able to take the attitude of the other towards their own gestures, to anticipate how the other will respond and respond to that anticipation, gestures become significant symbols, and with them come mind and self. Mind, for Mead, is not a thing inside the skull. It is a social process internalised: the conversation of gestures carried on silently with oneself. The self is not a substance. It is the relation between the "I" (the spontaneous response) and the "me" (the attitudes of others taken towards oneself), and above all the "generalised other", the organised attitudes of the whole community.

From Elias, CRP takes the figuration: a web of interdependent people, bound together by mutual needs and dependencies, whose patterns of interaction no one designs and no one controls. Power, for Elias, is not a possession but a characteristic of every relationship, a balance of dependencies in which each party needs the other more or less. In long chains of interdependence, as The Compass and the Loom summarises, no player controls the game, though powerful players feel they stand above it. Elias's critique of the closed person, the image of the individual as a sealed container whose inner self faces an outer world, is the critique of the self-image that makes the feeling of standing above the game possible.

From the complexity sciences, CRP takes an analogy. Patterns of human interaction are like the patterns that emerge in complex systems: produced by local interaction, not designed, at once stable and unstable, and capable of novelty. But the analogy is only an analogy. CRP does not say organisations are complex systems in the scientific sense. It says complexity science offers analogies that help us think about human interaction differently. The human is not modelled by complexity. Complexity is a resource for thinking about the human.

Griffin, Shaw and the practice

Douglas Griffin's The Emergence of Leadership (2002) developed CRP's account of leadership and ethics. Leadership is not a property of individuals but a pattern of recognition emerging in interaction: the leader is the one others recognise as such, in particular situations, for particular reasons. Ethics is not a matter of applying principles from outside but of the ongoing negotiation of what is good in local interaction. Griffin's distinction between systemic self-organisation, in which the system is imagined as a whole that organises itself and the observer stands outside it, and participative self-organisation, in which there is no whole apart from participation and no outside, is one of CRP's sharpest contributions.

Patricia Shaw's Changing Conversations in Organizations (2002) developed the practice. Organisational change is not a programme designed and implemented but a change in the patterns of conversation through which an organisation exists. The consultant is not an expert who diagnoses and prescribes. She is a participant who joins the conversation and helps it notice itself. Shaw's work on "open space" and on working "live" with groups, attending to what emerges in the room and not following a plan, is the closest CRP comes to a method.

Mowles and Flinn

Chris Mowles's books, Rethinking Management (2011), Complexity: A Key Idea for Business and Society (2015) and Managing in Uncertainty (2015), have carried CRP into the 2010s and 2020s, engaged it with pragmatism and critical theory, and defended it against charges of practical emptiness. Mowles's motto, taken from Donna Haraway, is staying with the trouble: remaining in the difficulty of a situation rather than escaping into a framework that promises to resolve it. In his debate with Snowden, which To the Strait-Jacketed, the Other Always Seems Baggy examined, Mowles took the side of the conversation against the instrument.

Kevin Flinn's Leadership Development: A Complexity Approach (2018) and his later work, which The Compass and the Loom draws on, made recognition the test of leadership. A leader is someone recognised as such, and recognition cannot be commanded. It can only be earned in the conversation. The Tidal Loom's final principle, recognition, not alignment, comes from here. The test is whether people say we did this ourselves.

CRP's Account of the Human Distinction

CRP agrees with anthro-complexity that human complexity differs in kind from non-human complexity, and that models drawn from ants, starlings and agent-based simulations mislead when applied to organisations. Its grounds are different, and the difference can be stated precisely.

Anthro-complexity: humans hold multiple identities, act with intention, and create abstractions that change constraints. These are capacities of persons, which make human systems different.

CRP: humans are formed in the conversation of gestures and the figuration of interdependence. Identity, intention and abstraction are not capacities persons bring to interaction. They are produced in interaction. What makes human systems different is that their members' selves are social processes, formed by taking the attitude of the other, so the patterns of interaction are also the patterns of selfhood. Human interaction is reflexive all the way down, because the self that interacts is itself an internalised interaction.

On the CRP account, Snowden's three marks are effects of the human difference, not its ground:

  • Multiple identities are the multiple "me"s formed by taking the attitudes of the different others and groups one belongs to. A person has as many social selves, as William James already said, as there are groups whose opinion she cares about. Identities are not held. They are called out by others' responses.
  • Intentionality is the anticipation of others' responses to one's gesture. An intention is a gesture rehearsed in the inner conversation, its response already imagined. Intentions are social before they are private.
  • Abstraction is the significant symbol, the gesture whose meaning is shared because each party can take the attitude of the other towards it. Abstractions change constraints because they are shared, and they are shared because they are social.

And the asymmetry principle (only humans can change the story they tell about the other) is, on the CRP account, a statement of Mead's significant symbol. Humans can change the story because the story is a social object, maintained in the conversation and changeable by it.

The Disagreement

The disagreement between anthro-complexity and CRP has a history, carried on in posts, papers, conference exchanges and the occasional sharp aside. Its outline, as far as the record allows it to be reconstructed fairly, is this.

Snowden's charge against CRP is, in substance, that it is practically empty and philosophically one-sided. It says everything is conversation, and therefore has nothing to say about the material, technological and institutional constraints that shape conversation. It refuses instruments and therefore cannot work at scale. Its "staying with the trouble" can become a refusal to act, a privileging of reflection over intervention that leaves practitioners with nothing to do but talk. And it draws on complexity science only by analogy, so it lacks the scientific grounding anthro-complexity claims through Juarrero, Prigogine and the natural sciences.

CRP's charge against anthro-complexity, as Stacey, Mowles and others have made it, is that its instruments reinstate the manager outside the system. Cynefin, read as a framework for diagnosing situations and choosing responses, presupposes a decision-maker who stands back from the situation and categorises it. SenseMaker presupposes an analyst who sees patterns no participant sees. Estuarine mapping presupposes a facilitator who convenes and frames. In Stacey's terms, anthro-complexity keeps the split causality of systems thinking: complex for the system, rationalist for the practitioner.

The straitjacket-and-baggy debate dramatised the disagreement. Each side sees the other's form as "baggy" or "boxed", while its own form feels like freedom. CRP finds anthro-complexity's instruments boxed, all frameworks and triads and maps. Anthro-complexity finds CRP's conversations baggy: unstructured, interminable, unaccountable. To the Strait-Jacketed argued that each is right about the other and blind about itself, because no one sees the form they are standing in.

Assessing the Disagreement

The disagreement is the clearest available test of where the human distinction lies, because each side's charge against the other identifies a real weakness.

CRP's weakness is real. It is thin on instruments, and a perspective that cannot be put to work at scale leaves practitioners in large organisations with little to do. Its reliance on complexity science by analogy only does leave it without a theory of material constraint. Stacey's later work acknowledged the importance of "the material", but CRP has nothing comparable to Juarrero's account of how constraints work. Its account of power, drawn from Elias, is subtle about interpersonal figurations and less developed about technologies, infrastructures and institutions. And "staying with the trouble" can slide into what The Compass and the Loom called the objection that complexity forbids action. If every choice reinstates the upper box, what is left to do?

Anthro-complexity's weakness is also real, and it is the one this book examines. Its instruments do reinstate positions outside the system: the signifier designer, the dashboard holder, the map convener, the categoriser. Its theory of the human does ground the distinction in properties of persons, which makes it vulnerable to the CAS theorist's reply (Chapter 7) that the difference is one of degree. And its scientific grounding, while real at the level of general complexity, does not reach the human cluster. Juarrero's constraints, Prigogine's dissipative structures and Kauffman's adjacent possible are not theories of human reflexivity. The human cluster rests on claims about human capacities that are not drawn from the natural sciences at all.

Each has what the other lacks. CRP has a relational account of the human distinction, grounded in Mead and Elias, that anthro-complexity needs. Anthro-complexity has a theory of constraint, grounded in Juarrero, and a repertoire of instruments that CRP lacks. Book Eight will propose that a reflexive anthro-complexity would combine them: Juarrero's constraints understood through Mead's gestures, SenseMaker's distributed sensing redesigned to honour Elias's figuration, estuarine mapping that includes the mapper.

CRP and Anthro-Complexity Compared

Question Anthro-complexity Complex responsive processes
Is human complexity different in kind? Yes Yes
Ground of the difference Properties of persons: identities, intention, abstraction Relations between persons: gesture, figuration, the social self
Relation to complexity science Grounded in it (natural-science origins) Analogical (complexity as resource for thought)
Theory of constraint Strong (Juarrero) Weak
Theory of power Present in critique of norms; absent from human cluster Central (Elias's figuration)
Position of practitioner Participant in intent; designer of instruments in practice Participant without exemption
Teleology Formative for system, rationalist for practitioner (CRP's charge) Transformative throughout
Instruments Many: SenseMaker, estuarine mapping, triads, SwarmCompass, WRAS Few: reflective practice, working live, narrative inquiry
Scale Large populations; organisations; nations Groups; teams; conversations
Characteristic risk Reinstating the steersman Paralysis; staying with the trouble indefinitely
Characteristic image The estuary; the sailor waiting for the tide The conversation; the room
Test of success Shifting constraints; more stories like these Recognition: "we did this ourselves"

What CRP Lends This Disputation

This disputation is not a CRP critique of anthro-complexity. It draws on CRP's resources, Mead, Elias, Stacey's teleologies, Griffin's participative self-organisation, Flinn's recognition, but it accepts anthro-complexity's charge against CRP as well. A perspective without instruments cannot serve organisations at scale, and a perspective without a theory of constraint cannot account for the material world that conversation takes place in. The aim is not to replace anthro-complexity with CRP but to use CRP's relational account of the human to complete anthro-complexity's human cluster, and so to redesign its instruments around the figure of the human its theory already points towards.

There is a further reason CRP matters here, and it touches the November post directly. Snowden's critique of Ing's physician, the physician takes the pulse; the body does not, is a CRP critique. It objects to split causality: the body is formative, healing itself, while the physician is rationalist, reading and diagnosing. Snowden's answer, distributed sensing, is anthro-complexity's answer to the charge CRP has made against it for years. The November post therefore enacts the disagreement between anthro-complexity and CRP in the course of a critique of a third party. Snowden takes CRP's side against Ing, and Book Four will ask whether he can take it consistently, all the way to his own instruments.

The adversary on Snowden's behalf can say that this whole chapter is beside the point, because the November post is not about CRP and Snowden is under no obligation to engage with every neighbour in every post. He has engaged with CRP elsewhere, often and pointedly. His decision to develop anthro-complexity through instruments and not only through reflection is a considered choice, made with full awareness of CRP's objections. Treating CRP's account as the "relational ground" anthro-complexity lacks is to take one side of a long-running argument and present it as the truth.

The adversary is right that Snowden owes CRP nothing in this post, and right that the choice of instruments over pure reflection is considered. The adversary is also right that this book takes a side, as every disputation does. But the chapter's claim is narrower than the adversary suggests. It is not that CRP is right and anthro-complexity wrong. It is that CRP has a relational account of the human distinction, anthro-complexity's own human cluster points towards a relational account (the asymmetry principle, the critique of the physician), and the two accounts can be combined. That is a claim about what anthro-complexity's premises commit it to, not about which school is better. partial

ONE SCHOOL HAS THE INSTRUMENTS —

ONE SCHOOL HAS THE RELATION —

NEITHER CAN SEE THE ROOM IT IS STANDING IN.

Addition: The Baggy and the Boxed, Revisited

The disagreement between anthro-complexity and CRP has a recent instance that To the Strait-Jacketed, the Other Always Seems Baggy examined in detail: an exchange between Dave Snowden, in an essay titled Neither Baggy Nor Boxed, and Chris Mowles, in an essay titled Staying with the Trouble. This addition revisits the exchange in the light of the present disputation, because the disputation's argument illuminates it, and it illuminates the argument.

The exchange, in outline

In the reading of To the Strait-Jacketed, the exchange concerned the right degree of structure in practice. Snowden argued that practice needs structure, frameworks, methods and constraints, to be effective, and that CRP's emphasis on unstructured conversation leaves practitioners with too little to work with. He positioned anthro-complexity as neither baggy nor boxed: enough structure to act, not so much as to lose complexity. Mowles argued that the structures practitioners bring to situations shape what they see and do, often in ways that obscure the situation's real difficulty, and that the practitioner's task is to stay with the trouble, to remain in the difficulty rather than escaping into a framework that promises to resolve it.

To the Strait-Jacketed drew a predicate from the exchange: to the strait-jacketed, the other always seems baggy. Each side sees the other's form as "baggy" or "boxed", while its own form feels like freedom. Anthro-complexity, inside its frameworks, experiences them as enabling constraints and sees CRP's conversation as formless. CRP, inside its conversation, experiences it as freedom and sees anthro-complexity's frameworks as strait-jackets. Neither sees the form it is standing in.

What the present disputation adds

The present disputation's analysis adds three things to that reading.

First, the exchange is an instance of the claim to "neither" (Chapter 20). Snowden's title, Neither Baggy Nor Boxed, positions anthro-complexity outside two errors, from a vantage from which both can be seen. Chapter 39 read this vantage through Wigner. It is a description from a position, presented as a description from outside. To the Strait-Jacketed's predicate is Wigner's friend's reply. The friend inside the lab sees Wigner's description as baggy, incomplete and missing what she saw. Wigner sees the friend's as boxed, confined to one outcome when the lab is in superposition. Both are correct relative to their positions.

Second, the exchange is about where the human distinction lives. Anthro-complexity's structures (frameworks, signifier sets, maps) are instruments for sensing and intervening in human systems at scale. They are built for possessors (Thesis I): individuals with narratives to collect, constraints to map and situations to categorise. CRP's conversation is a practice for participating in human systems from inside. It is built for relations: gestures and responses, figurations and power ratios, the generalised other. The exchange about structure is, at bottom, an exchange about the ground of the human distinction. Is the human a possessor whose properties can be sensed by instruments, or a relation that can be participated in only from inside? The present disputation's answer is that the human is a relation, and that instruments can be designed for relations, so the exchange does not have to be settled by choosing between structure and conversation.

Third, the Tidal Loom is a synthesis the exchange did not see. The Loom has structure, a warp laid by convenors, and it has conversation, a weft woven by everyone. It has no position outside the cloth. It is neither baggy nor boxed in a different sense from Snowden's. It does not stand between the two as a vantage. It weaves them together as a practice. The warp is anthro-complexity's structure, and the weft is CRP's conversation. Neither is complete alone.

Each side's blind spot, stated precisely

The present disputation can state each side's blind spot more precisely than To the Strait-Jacketed did, using the vocabulary developed since.

Anthro-complexity's blind spot is its own instruments as actors. It sees CRP's conversation as baggy because, from inside the frameworks, conversation looks like an absence of structure. It does not see that its frameworks are themselves gestures in the conversation of the organisations it works in, shaping what is said, who says it and what is heard (Theses II, V). Its structure is not outside the conversation. It is a move in it, and an unusually powerful one.

CRP's blind spot is constraint at scale. It sees anthro-complexity's frameworks as boxed because, from inside the conversation, frameworks look like impositions. It does not see that conversation itself has constraints, constitutive constraints, habitus and the generalised other, that shape it as powerfully as any framework, and that at the scale of large organisations and populations those constraints cannot be addressed by conversation alone (Chapter 11). Its freedom is not freedom from structure. It is a structure it does not see, which is Freeman's tyranny of structurelessness (Chapter 23).

The exchange as a conversation of gestures

There is a final point, which To the Strait-Jacketed made and the present disputation endorses. The exchange between Snowden and Mowles is itself a conversation of gestures, in Mead's sense. Each essay is a gesture that calls out a response, and its meaning is completed by the response. Snowden's Neither Baggy Nor Boxed means something different after Mowles's reply than before it. Mowles's Staying with the Trouble means something different in the light of Snowden's essay. And both mean something different after To the Strait-Jacketed and this disputation have read them. The exchange is not a debate with a winner. It is a conversation in which each move changes the meaning of the others, as the meaning of the dog's bared teeth depends on the other dog's response.

This disputation is a move in that conversation. Its meaning will be completed by the responses it calls out, from Snowden, from Mowles, from practitioners in both schools, or from no one. If it calls out no response, its meaning will be that of a gesture unanswered. That too is a meaning, and the author accepts it.

TO THE STRAIT-JACKETED, THE OTHER ALWAYS SEEMS BAGGY —

TO THE BAGGY, THE OTHER ALWAYS SEEMS BOXED —

THE LOOM HAS WARP AND WEFT, AND NO ONE OUTSIDE THE CLOTH.

Addition: Cross-Lights From Book Nine

The process schools speak for themselves. This chapter described complex responsive processes from outside. Two later chapters let the school speak. Chapter 51 uses A Web of Relations, which sets the Designer of Conditions against the Dweller in the Conversation, keeps a ledger for each, and proposes formation as the bridge from a room to an organisation. Chapter 57 uses Kevin Flinn's book on organisational development, which shows the practice itself: large groups without an agenda, practical judgement, and writing.

What changes here. The school is no longer open to the simple charge that it cannot act. Mowles has turned to practical wisdom. It is open to a different question, put in Chapter 57: who is in the room, who formed the convenor, and what would show the practice to have failed.

Chapter 12

Critical Systems Thinking and the Pluralist Temptation

4,042 words, about 18 minutes of reading

The Neighbour in the Postscript

Snowden's postscript contains a sentence that, read carefully, opens a door the post itself keeps shut. Systems thinking, it says, is not monolithic. Mike Jackson's critical systems thinking engages with complexity far more directly, and the post's criticisms are not meant to apply to every part of the field. The concession is generous and accurate. Critical systems thinking (CST), as Michael C. Jackson and his colleagues at Hull developed it from the 1980s onwards, does engage directly with complexity. Jackson's Critical Systems Thinking and the Management of Complexity (2019) addresses complexity theory and complexity-based approaches among the systems methodologies it surveys. CST also does something no other neighbour examined so far does with the same explicitness: it places power at the centre of its account of what kind of situation one is in.

This chapter examines CST as the neighbour in the postscript. It sets out CST's core moves: the system of systems methodologies, the appeal to Habermas, boundary critique, and the pluralist meta-methodology. It identifies what CST offers this disputation, which is a vocabulary for power that anthro-complexity's human cluster lacks. It also identifies the temptation CST shares with Cynefin: that a framework for choosing frameworks reinstates, at a higher level, the outside position it was meant to discipline. The Connoisseur of Partial Truths, an earlier treatise in this series, examined that temptation in Jackson at length. This chapter draws on it and does not repeat it.

The System of Systems Methodologies

CST's founding gesture was a grid. In 1984 Jackson and Paul Keys proposed a "system of systems methodologies" that classified problem contexts on two dimensions:

  • the systems dimension: whether the system in question is relatively simple or complex;
  • the participants dimension: whether the people involved are in a unitary relation (they share values and goals), a pluralist relation (they have different values but can reach accommodation), or a coercive relation (they have conflicting interests and some can impose their will on others).

The grid gives six cells, and each cell is suited to different systems methodologies. Hard systems thinking and operational research suit simple-unitary contexts. System dynamics and the viable system model suit complex-unitary contexts. Soft systems methodology and strategic assumption surfacing suit pluralist contexts. Critical systems heuristics and other emancipatory approaches suit coercive contexts.

The grid's significance is not its particular allocations, which have been revised many times. It is that it made the relations among participants, including relations of coercion, a dimension of the problem context on a par with the complexity of the system. A situation is not just simple or complex. It is also unitary, pluralist or coercive, and the second dimension is about power. That is the move this disputation finds missing from anthro-complexity's human cluster. Cynefin's domains are defined by the relation of cause and effect. They have no participants dimension. A complex situation may be unitary, pluralist or coercive, and Cynefin's decision model for it (probe, sense, respond) is the same in each case. CST would say that a probe in a coercive context is a different act from a probe in a unitary one, because in a coercive context someone decides what is probed, who bears the risk of the probe, and who benefits from what it reveals.

Habermas and the Three Interests

CST's philosophical grounding, in its classical form, was Jürgen Habermas's theory of knowledge-constitutive interests, from Knowledge and Human Interests (1968). Habermas argued that human knowledge is shaped by three fundamental interests:

  • the technical interest in predicting and controlling nature, which constitutes the empirical-analytic sciences;
  • the practical interest in mutual understanding through communication, which constitutes the historical-hermeneutic sciences;
  • the emancipatory interest in freedom from domination and distorted communication, which constitutes critical theory and psychoanalysis.

CST mapped these interests onto systems methodologies. Hard systems approaches serve the technical interest, soft approaches the practical, and critical approaches the emancipatory. A complete systems practice would serve all three, using each kind of methodology for the interest it serves. CST later moved away from Habermas towards a broader pluralism, but the three-interest structure left a lasting mark. It insisted that methodologies are not neutral tools. Each embodies an interest, and choosing a methodology is choosing whose interest the inquiry serves.

For this disputation, the Habermasian moment matters because it names something anthro-complexity's instruments embody without naming. SenseMaker, estuarine mapping and the rest serve interests. Mostly they serve the practical interest in mutual understanding, by making a population's narratives and constraints visible to itself. But they can serve the technical interest in prediction and control, when patterns are used to steer and constraints are mapped to be manipulated. And they can serve, or betray, the emancipatory interest, depending on whether they reveal the distortions of communication that power produces or reproduce them. Which interest an instrument serves depends not on its design alone but on who commissions it, who reads it, and what is done with what it reveals. Habermas's framework makes the question unavoidable.

Boundary Critique

The most practically powerful contribution of CST, and the one most directly relevant here, is boundary critique, developed by Werner Ulrich in Critical Heuristics of Social Planning (1983) and extended by Gerald Midgley in Systemic Intervention (2000).

Ulrich's starting point is that every systems inquiry draws a boundary. It decides what is inside the system and what is outside, whose concerns count and whose do not, what is to be improved and what is to be taken as given. The boundary is not given by the situation. It is a judgement, and a judgement with ethical and political content. Different boundaries make different things visible and different improvements possible. The "expert" who draws the boundary has a power those inside and outside it may not have. Ulrich's critical systems heuristics offers twelve boundary questions, organised under four headings:

Source of Questions about what is and what ought to be the case
Motivation Who is (ought to be) the client or beneficiary? What is (ought to be) the purpose? What is (ought to be) the measure of improvement?
Control Who is (ought to be) the decision-maker? What resources are (ought to be) controlled by the decision-maker? What conditions of success are (ought to be) outside the decision-maker's control?
Knowledge Who is (ought to be) considered a professional or expert? What expertise is (ought to be) consulted? What is (ought to be) assumed to be the guarantor of success?
Legitimacy Who is (ought to be) witness to the interests of those affected but not involved? What secures (ought to secure) the emancipation of those affected from the premises and promises of those involved? What worldview is (ought to be) determining?

The phrase that has become the signature of boundary critique is those affected but not involved: the people whose lives a system will change but who have no part in designing it. Ulrich's heuristics are designed to make them visible, and to give them, or their witnesses, a voice in the boundary judgement.

Midgley's development of boundary critique added the concept of marginalisation. When boundaries are drawn, some stakeholders and concerns fall into the margin between a narrow and a wide boundary. They are neither fully inside nor fully outside, and they are often made "profane", treated as irrelevant or illegitimate, by those whose boundary prevails. Midgley's systemic intervention is a practice of attending to the margins and exploring the boundary judgements that create them.

Boundary critique applied to anthro-complexity's instruments

Boundary critique is the instrument this disputation most needs and anthro-complexity most lacks. Apply Ulrich's twelve questions to a typical SenseMaker deployment:

  • Client. Who is the beneficiary? Typically the organisation that commissioned the study, represented by its leadership. Who ought to be? On anthro-complexity's Pelagian principles, the population whose narratives are gathered.
  • Purpose. What is the purpose? Typically to give leadership insight into the organisation's culture, risks or attitudes. What ought it to be? On anthro-complexity's principles, to help the population see itself and act on what it sees.
  • Decision-maker. Who decides what is done with the patterns? Typically leadership. Who ought to? Anthro-complexity's principles suggest the population, or at least the population together with leadership.
  • Expert. Who designs the signifiers? Typically the Cynefin practitioner, with the client. Who ought to? Anthro-complexity's principles suggest the population should have a voice in what counts as salient.
  • Witness for the affected. Who speaks for those affected by the study but not involved in its design, the narrators themselves? Typically no one, formally. Who ought to? Ulrich would say someone must.
  • Worldview. What worldview determines the study? The worldview embodied in the signifier set, which encodes what the designers expected, hoped to find and would prefer not to find. What ought to? A worldview the population can recognise as its own.

The application shows that in each case the "is" and the "ought" diverge, and that the "ought" is supplied by anthro-complexity's own principles: Pelagian, distributed, self-sensing. Boundary critique does not impose foreign values on anthro-complexity. It reveals the gap between anthro-complexity's values and the boundary judgements its instruments embody in practice. That is what boundary critique is for.

The Compass and the Loom's Tidal Loom can be read as a set of answers to Ulrich's "ought" questions. Reading of the reader answers the expertise question: convenors signify too. Declination before direction answers the worldview question: designers state their expectations. Return as design requirement answers the client question: narrators receive patterns before anyone else. Consent to the harvest answers the legitimacy question: no narrative becomes data without informed consent and a share in the outcome.

The Pluralist Turn and Its Temptation

From the 1990s CST moved from the system of systems methodologies towards a broader pluralism. Flood and Jackson's Creative Problem Solving: Total Systems Intervention (1991) proposed a meta-methodology, TSI, for choosing and combining methodologies according to the problem context. It moved through phases of creativity (surfacing the issues), choice (selecting appropriate methodologies) and implementation (applying them), with critical reflection throughout. Jackson's later work developed this into "critical systems practice", a pluralism that combines methodologies from different paradigms in a single intervention, guided by critical awareness of their strengths, weaknesses and the interests they serve.

The pluralist turn has a temptation, and The Connoisseur of Partial Truths examined it at length. The meta-methodologist who chooses among methodologies stands, in principle, above them. She surveys the methodologies, judges their fitness for the context, and combines them. She is the connoisseur of partial truths, who knows each methodology's partiality and so knows better than any of them. The figure is the systems version of the steersman: the one who reads the situation from above and chooses how to act on it. The meta-methodology is meant to be critical and reflexive, and its practitioners are meant to examine their own choices. But the structure of the meta-methodology, choose the methodology for the context, places the chooser outside the methodologies chosen and outside the context diagnosed.

Cynefin's mirror

Cynefin faces exactly the same temptation, and the parallel is close:

Critical systems practice (Jackson) Cynefin (Snowden)
What is classified Problem contexts (simple/complex × unitary/pluralist/coercive) Situations (clear/complicated/complex/chaotic)
What is chosen Systems methodology suited to context Decision model suited to situation
Who chooses The critical systems practitioner The decision-maker / sense-making group
Position of the chooser Above the methodologies (meta-methodology) Before the situation (categorisation)
Safeguard Critical reflection on choices and interests Aporetic/confused domain; collective sense-making
Characteristic risk The connoisseur of partial truths The steersman who reads, then acts from above

Both frameworks are frameworks for choosing frameworks. Both place the chooser in a position the frameworks chosen do not include. Both have safeguards: CST's critical reflection, Cynefin's aporetic domain and the insistence on collective sense-making. Both have been criticised by CRP for keeping the split causality of systems thinking. The difference that matters for this disputation is that CST's grid has a participants dimension, unitary, pluralist or coercive, and Cynefin's does not. CST's chooser is at least required to ask whether the situation is coercive. Cynefin's is not.

What CST Lends This Disputation

CST lends this disputation four things.

First, a participants dimension. The relations among participants, including relations of coercion, are a dimension of the situation as basic as its causal structure. A reflexive anthro-complexity would add this dimension to Cynefin. The question would be not only "what is the relation of cause and effect here?" but "what are the relations of power among those affected?" The answer would shape the decision model: a probe in a coercive context needs safeguards that a probe in a unitary one does not.

Second, the interest question. Every instrument serves an interest, and choosing an instrument is choosing whose interest it serves. A reflexive anthro-complexity would ask of every deployment: technical, practical or emancipatory?

Third, boundary critique. Ulrich's twelve questions, applied to anthro-complexity's instruments, reveal the gap between the "is" and the "ought" of each deployment and supply the "ought" from anthro-complexity's own principles. A reflexive anthro-complexity would build boundary critique into every instrument's design.

Fourth, a warning. The meta-methodology reinstates the steersman at a higher level, and the same is true of any framework for choosing frameworks, Cynefin included. A reflexive anthro-complexity would need a safeguard against its own meta-level, and the safeguard cannot be another framework. It has to be a practice of including the chooser among the chosen, the reader among the read.

What CST Does Not Lend

CST does not lend this disputation its account of the human. CST's human, in its Habermasian form, is a communicating subject whose communication may be distorted by power and freed by critique. In its pluralist form, the human is a stakeholder with interests, values and worldviews that methodologies must accommodate. Neither account is relational in Mead's or Elias's sense. Both take the person, with interests, values and communicative capacities, as given, and ask how systems practice can serve her. CST's human is closer to anthro-complexity's than to CRP's: a bearer of properties, not a pattern of relation. CST adds power to the properties, the capacity to impose or resist, but it does not make power the medium through which persons are formed, as Elias does.

CST also does not lend its confidence in critique. Habermas's emancipatory interest presupposes that distorted communication can be recognised as distorted, from a standpoint, the ideal speech situation, that is not itself distorted. That standpoint is the outside position again, in its most refined form. The critical theorist who reveals the distortions of power stands where power does not distort. CRP's Elias would say there is no such place. Every critic is in a figuration, and every critique is a move in it. This disputation, as Chapter 1 said, is itself a move in the field it describes.

Jackson's defender has a strong reply. CST's critical reflection is not a claim to stand outside. It is a practice of continuously examining one's own choices and the interests they serve, and of revising them when they are found wanting. The connoisseur charge confuses a meta-methodology with a meta-position. The critical systems practitioner does not stand above the methodologies. She moves among them, aware of each one's partiality, including the partiality of her own awareness. The same reply is available to Cynefin: the decision-maker does not stand outside the situation, but moves within it, using the framework to make sense of where she is.

The reply wins half. It is right that critical reflection, practised honestly, is a practice within the situation and not a view from above, and right that both CST and Cynefin, at their best, are used that way. It is wrong that the structure of a meta-methodology is neutral between the two readings. A framework that classifies contexts and assigns methods to them is a framework built for someone who classifies and assigns. Its structure invites the reading from above, whatever its practitioners intend. The safeguard against that invitation cannot be the practitioner's good intentions alone. It has to be built into the structure, as the Tidal Loom tries to do: convenors signify, designers declare, narrators receive first. partial

The Neighbours Mapped

With this chapter, Book Two's survey of the neighbours is complete. Before moving on, the map can be drawn, with the warning of Chapter 10 that every map of neighbours is drawn by a neighbour:

Neighbour Shares with anthro-complexity Differs from anthro-complexity Lends this disputation
Complex adaptive systems Emergence; nonlinearity; no central control; natural-science grounding Agents with rules; simulation; modeller outside The distinction of kind is relational (Ch. 8)
Systems thinking (Ing) Propensity; retrospect; Gibson; Ingold; turbulent field; kairos Wholes presumed; human downplayed; dyadic; eurhythmia Pepper's maxims cut both ways (Ch. 10)
Complex responsive processes Human complexity differs in kind; no outside position (in intent) Relational ground; transformative teleology; few instruments Mead, Elias, teleologies, recognition (Ch. 11)
Critical systems thinking Frameworks for choosing frameworks; engagement with complexity Participants dimension; power; interests; boundary critique Boundary critique; the interest question (Ch. 12)

Book Three now asks the question the map makes possible. With the neighbours placed, why does anthro-complexity treat human complexity differently? Is it right to, and is it right for the reasons it gives?

WHO IS AFFECTED BUT NOT INVOLVED? —

WHO DESIGNED THE TRIAD? —

WHO WITNESSES FOR THE NARRATOR?

Addition: Boundary Critique Applied to Estuarine Mapping and SwarmCompass

The chapter applied Ulrich's twelve boundary questions to a typical SenseMaker deployment. This addition applies them to the two other instruments the disputation has examined most closely, estuarine mapping and SwarmCompass, to show that boundary critique generalises. Each application asks what the instrument's boundary judgements are in typical use, and what anthro-complexity's own principles say they ought to be.

Estuarine mapping

Boundary question What is the case (typical) What ought to be (by anthro-complexity's principles)
Client: whose interests are served? The commissioning organisation, through its leadership or a change team All whose constraints are mapped, including those who bear the costs of shifting them
Purpose: what is the map for? To identify constraints that can be shifted to enable a desired change To show those inside the system where they have room to act, on their own terms
Measure of improvement Constraints shifted; change achieved Room to act increased for those who were constrained (von Foerster's imperative)
Decision-maker: who decides what to do with the map? Leadership or change team Those whose constraints are mapped, with leadership
Resources controlled by decision-maker Budget, authority to change formal constraints Shared: participants control some constraints (habits, practices); leadership others (rules, budgets)
Conditions outside decision-maker's control Mapped as "cannot be changed" Dated limits: "cannot be changed until/unless", with who judged so
Expert: who is the professional? The facilitator; the method The knife-holders, who know the grain, alongside the facilitator
Expertise consulted The participants in the room Participants, and those affected but not present
Guarantor of success The method's track record The participants' recognition: we found our own room
Witness for those affected but not involved None formally Someone named to ask, at each stage: who is not here, and what would they say?
Emancipation of those affected Not addressed Those affected can see the map, contest it, and propose their own constraints
Worldview Constraint energetics; the estuary Constraint energetics and constitutive constraints, closure, hysteresis, power

The application shows a pattern. Estuarine mapping's method is participatory, since participants place the constraints, but its boundary judgements in typical use are the commissioner's. The client is the organisation, the purpose is the commissioner's change, the decision-maker is leadership, and the affected-but-not-involved have no witness. The participatory method sits inside a non-participatory boundary.

SwarmCompass

The Compass and the Loom examined SwarmCompass at length, and boundary critique sharpens several of its findings.

Boundary question What is the case (on The Compass and the Loom's reading) What ought to be
Client The organisation "adrift without direction", represented by its leadership The members whose noticing is gathered
Purpose To find direction and align the organisation to it To help members recognise a direction they are already making (道, Chapter 37)
Measure of improvement Alignment Recognition: we did it ourselves
Decision-maker Leadership, informed by the swarm The swarm, with leadership as one voice
Expert The instrument; "the 17%" who see what others miss Frames, not seers: different people see different gorillas
Guarantor of success The bee, the gorilla and Lincoln warrants Testing at the rendezvous: do independent actions fit?
Witness for those affected but not involved None Those whose noticing becomes training data
Emancipation Not addressed Consent to the harvest; share in the outcome
Worldview Read, then act from above; the steersman The practitioner inside

The application confirms The Compass and the Loom's central finding in Ulrich's vocabulary. SwarmCompass's boundary judgements place the client, the decision-maker and the expert at the top of the organisation, and the members whose noticing is gathered outside the boundary of those whose interests the instrument serves. The members are affected but not involved. Their noticing is the instrument's raw material, and their interests are not its purpose.

Midgley's margins

Gerald Midgley's development of boundary critique adds the concept of marginalisation. When a narrow boundary and a wide boundary are both in play, stakeholders and concerns fall into the margin between them, and the margin is a site of conflict. Those who hold the narrow boundary tend to make the marginal elements profane, irrelevant or illegitimate. Those who hold the wide boundary tend to make them sacred, the very point of the inquiry. The conflict between sacred and profane is ritualised, Midgley argued, in ways that sustain the narrow boundary.

Applied to narrative instruments, Midgley's margin is the space of stories that do not fit the signifier set, narrators who do not respond, and groups who are under-represented. Those who hold the narrow boundary, the commissioners interested in the patterns they asked about, treat the margin as profane: noise, low response rates, outliers to be set aside. Those who hold the wide boundary, narrators whose experience the signifiers do not capture, treat it as sacred: the stories that matter most. The Tidal Loom's principles, the fourth vertex, routing the atypical to humans and the Good–Turing estimate of the missing mass, are ways of honouring the margin, treating it as sacred rather than profane.

The lesson of the applications

Boundary critique, applied to three instruments, shows the same pattern three times. Anthro-complexity's methods are participatory within boundaries drawn by commissioners. The participation is real. Narrators signify, participants map and swarms notice. But the boundary judgements that determine whose interests the participation serves, who decides what is done with it and who speaks for those affected are made outside the participation, by those who commissioned it.

This is the clearest statement in the disputation of the gap between anthro-complexity's intent and its structure (Chapter 6). The intent is participatory. The structure places participation inside a boundary the participants did not draw. Ulrich's remedy, making the boundary judgements explicit and opening them to those affected, is the remedy this disputation has proposed throughout under other names: declination, return before action, the witness for the affected, consent to the harvest. Boundary critique gives the remedies a single discipline, with twelve questions, that any practitioner could ask at the start of any deployment.

PARTICIPATION INSIDE A BOUNDARY THE PARTICIPANTS DID NOT DRAW —

ASK THE TWELVE QUESTIONS — NAME A WITNESS FOR THE AFFECTED —

TREAT THE MARGIN AS SACRED.

Addition: Cross-Lights From Book Nine

Critical pluralism and the elephant. This chapter credits critical systems thinking with its pluralism of methods. Chapter 56 examines a recent statement of a similar position from the complexity side, which recommends critical pluralism among five orientations. The author's objection is the parable of the blind men and the elephant. A pluralism of models gathers more reports from outside. The parable depends on a narrator who can see the whole animal. The amendment proposed is a pluralism of participations.

Cole on diagnostic and dialogic work. Flinn reports Mark Cole's criticism of the division of organisational development into a diagnostic kind for problems of cause and effect and a dialogic kind for wicked ones. Cole calls it a false binary in which both halves reproduce power. The criticism applies to any scheme that assigns methods to kinds of problem, the System of Systems Methodologies and Cynefin included.

Chapter 13

The Scientific Reason: Humans Are Not Ants

4,408 words, about 19 minutes of reading

Five Reasons, Not One

Book Three asks why anthro-complexity treats human complexity differently. This chapter answers that the question has five answers, not one. Anthro-complexity's human distinction rests on five reasons, which are often run together and need to be separated:

  1. The scientific reason. The best available science of human cognition and social behaviour shows that humans differ from other complex-system agents in ways that make agent-based and swarm models mislead.
  2. The reflexive reason. Human systems are partly made of their members' descriptions of them, so descriptions change the systems they describe.
  3. The anti-computational reason. Human cognition is not the execution of a programme, so models that represent it as one misrepresent it.
  4. The ethical reason. Because humans act with intention and respond to being acted on, intervention in human systems is both possible and dangerous, and needs an ethics that intervention in non-human systems does not.
  5. The positional reason. A school of thought needs a border, and the human distinction gives anthro-complexity one against systems thinking, CAS and the flat ontologies.

Each reason has its own logic, evidence and vulnerabilities. Chapter 18 argues that the five do three different jobs (ontological, methodological and institutional) and that the jobs pull against one another. This chapter takes the first reason, the scientific one. It asks what the science actually shows, and finds that it supports anthro-complexity's distinction. But on every one of the three marks, the strongest science grounds the distinction relationally, in shared intentionality, collective acceptance and social categorisation, not in properties individuals carry.

The Claim

The scientific reason is stated in the November post in compressed form: humans are not ants, and models drawn from insect behaviour mislead when applied to organisations. Elsewhere Snowden has developed the claim at length, drawing on cognitive neuroscience, evolutionary psychology and the study of decision-making. In outline:

  • Human cognition is pattern-based. People do not reason from rules to cases. They match present situations against stored fragments of past experience and act on the first fit, a mode of cognition Gary Klein studied in expert decision-makers under the name recognition-primed decision-making.
  • Human cognition is framed. What people perceive, attend to and remember depends heavily on context, priming and the frames through which situations are presented, as the inattentional-blindness studies discussed in Chapter 9 dramatise.
  • Human identity is multiple. People belong to many groups, occupy many roles and switch between identities according to context.
  • Human action is intentional. People act towards imagined futures, and imagined futures change behaviour now.
  • Human collectives are narrative. Groups maintain shared stories about who they are, how things work and what matters, and the stories constrain and enable what members do.

Each claim is plausible, and each has a scientific literature. The question for this chapter is what that literature says about the ground of each claim.

The First Mark: Identity

Social identity and self-categorisation

The science of multiple identity most directly relevant to anthro-complexity's first mark is the social identity tradition founded by Henri Tajfel and John Turner in the 1970s, and its development as self-categorisation theory by Turner and colleagues from the 1980s.

Tajfel's "minimal group" experiments showed that people assigned to groups on trivial grounds (preference for one abstract painter over another, or a coin toss) favoured their own group in allocating rewards, even with no history of interaction and no conflict of interest. Group membership, however arbitrary, was enough to produce in-group favouritism. Tajfel and Turner's social identity theory explained this. People derive part of their self-concept from the groups they belong to, and they are motivated to see those groups positively, especially in comparison with other groups.

Self-categorisation theory took the further step that matters here. It holds that a person's identity at any moment is a categorisation: the person sees herself as an instance of some category, and which category is salient depends on the comparative context. The same person may categorise herself as an individual (different from others in her group), as a member of a group (similar to her group, different from other groups), or as a human being (similar to all humans, different from other species), depending on what she is being compared with. Turner's "meta-contrast principle" specifies the rule. A collection of people is seen as a group to the extent that the differences among them are smaller than the differences between them and others in the context. Change the context, and the salient categorisation changes.

What this shows

Self-categorisation theory strongly confirms anthro-complexity's first mark. People do hold multiple identities and do switch between them by context, often without awareness. A nurse who sees herself as a clinician when compared with administrators may see herself as a union member when compared with management, and as a mother when the conversation turns to childcare. The switching follows the comparative context, and it changes perception, judgement and behaviour.

But notice what self-categorisation theory says about the ground of multiple identity. Identities are not properties a person holds independently of context and then activates. They are categorisations produced by the comparative context. And the context is social: it consists of the other people present, the groups they represent, and the comparisons the situation invites. A person does not have a union identity that the context switches on. The union identity is her categorisation of herself in a context where union and management are the salient contrast. Remove the contrast and the identity is not dormant. It is not salient, and in an important sense it is not there.

This is a relational account of identity, and it is the account the best science supports. Identity is produced in the relation between a person and a comparative context of others. The first mark, read through the science, is relation-located.

The consequence for instruments

The consequence for anthro-complexity's instruments is immediate. If identity is produced by comparative context, then the context in which a narrative is collected produces the identity from which it is told. A SenseMaker prompt that invites a story about "your experience as an employee" makes the employee identity salient. One that invites a story about "a time you felt proud" may call out a professional, familial or national identity. The signifier set, which offers dimensions of comparison, is itself a comparative context. It shapes which identity the narrator signifies from. The instrument does not discover which of a narrator's multiple identities a story comes from. It participates in producing it.

The Second Mark: Intentionality

Shared intentionality

The science of human intentionality most relevant to anthro-complexity's second mark is the comparative developmental psychology of Michael Tomasello and his colleagues, developed over three decades of comparing human children with great apes. Tomasello's central finding, set out in The Cultural Origins of Human Cognition (1999), Origins of Human Communication (2008), A Natural History of Human Thinking (2014) and Becoming Human (2019), is that what most distinguishes human from ape cognition is not individual intelligence, at which apes are very capable, but shared intentionality: the capacity and motivation to form joint goals, joint attention and joint commitments with others.

From around nine months, human infants engage in joint attention: they look where an adult looks, point to show the adult something, and check that the adult is attending to the same thing. Apes follow gaze, but they do not, on Tomasello's evidence, point to share attention for its own sake or check that attention is shared. By the second year, children engage in joint action with shared goals. If an adult stops participating in a cooperative game, the child tries to re-engage the adult instead of continuing alone, as if the child understood that we were doing this together. By three or four, children understand collective intentionality: social norms, conventions and institutions that apply to "us" and that "we" enforce on one another.

Tomasello calls the result "we-intentionality", intentional states whose subject is plural. We intend to play the game, we are going to the shop, we do things this way here. His argument is that human thinking, communication, morality and culture all depend on it. Human thinking is a form of internalised social coordination, a perspective-taking dialogue with imagined others, much as Mead had proposed seventy years earlier.

What this shows

Tomasello's work strongly confirms anthro-complexity's second mark. Humans do act with intention, in a sense apes do not. Apes act purposively and anticipate, but they do not form shared goals and commitments, and their purposiveness is closer to Holland's anticipation through internal models than to the open-ended, revisable intention anthro-complexity describes.

But notice what Tomasello says about the ground of distinctively human intentionality. It is shared. What makes human intention different from ape purposiveness is not that individual humans have more complex private intentions. It is that humans form intentions together: joint goals, shared attention, collective commitments. Individual human intentions are, on Tomasello's account, built on and out of shared ones. A child learns to intend by intending with others. The adult's private intentions are internalised forms of the joint intentions she once formed in cooperation. The distinctively human form of intentionality is, at its root, relational.

The consequence for instruments

If distinctively human intention is shared intention, then the intentions that matter in a human system are not the private intentions of individuals but the joint intentions of groups: what we are trying to do. These are not properties of any individual. They exist in the relations among the group's members, maintained by their mutual expectations and commitments. A narrative instrument that collects individual stories and aggregates them sees patterns across individuals, the distribution of experiences in a population. It does not see joint intentions, which exist between individuals. A thousand individual stories about "what I am trying to do" do not add up to "what we are trying to do". The difference between them is the difference between an aggregate and a commitment.

The Third Mark: Abstraction and Narrative

Searle's social ontology

The science (strictly, the philosophy) most directly relevant to anthro-complexity's third mark is John Searle's social ontology, developed in The Construction of Social Reality (1995) and Making the Social World (2010). Searle asked how there can be facts that exist only because we believe they exist: money, marriage, property, government, corporations, universities. A piece of paper is money, a ceremony makes a marriage, a vote creates a president. These facts are as objective as facts about mountains, since it is not a matter of opinion whether this paper is legal tender. Yet they depend on human attitudes in a way facts about mountains do not.

Searle's answer has three elements. First, collective intentionality: humans (and, Searle says, some animals) can share intentional states, believing, desiring and intending things together. Second, the assignment of function: humans can assign functions to objects that the objects do not have by virtue of their physical properties alone. A stick becomes a lever, a stone a boundary marker. Third, and distinctively human, status functions. Humans can assign to objects or people a status, and with it a function, that they can perform only because of the collective acceptance of the status. The paper can buy things only because we collectively accept it as money. The person can command the army only because we collectively accept her as president. Searle's formula is X counts as Y in context C. Status functions carry deontic powers, rights, duties, obligations and permissions, that give people reasons for action independent of their desires.

Status functions are created and maintained by language. In Making the Social World Searle argues that all institutional facts are created by a single kind of speech act, the status function declaration, which makes something the case by representing it as the case. "I now pronounce you married." "The meeting is adjourned." "This company is hereby incorporated." The declaration changes the constraints within which people act, because it creates deontic powers that did not exist before.

What this shows

Searle's social ontology is the most rigorous available account of anthro-complexity's third mark. Humans do create abstractions that change the constraints they operate within. Status functions are such abstractions, and they are among the most powerful constraints in human life. Laws, rules, roles, budgets, contracts and hierarchies are all status functions, created by declaration and maintained by collective acceptance. The asymmetry principle is a special case of Searle's point. The planners can change the story they tell about the river because the river's status (protected, navigable, a boundary, a hazard) is a status function, and status functions can be changed by declaration.

But notice, once more, the ground. Status functions exist only by virtue of collective acceptance. An individual cannot create a status function alone. A declaration that no one accepts creates nothing. The power of abstraction to change constraints is therefore not a power individuals have. It is a power that exists in the relation between a declaration and the community that accepts it. Searle is emphatic that collective intentionality cannot be reduced to individual intentionality plus mutual belief. It is irreducibly plural. The third mark, read through Searle, is relation-located.

The consequence for instruments

If abstractions change constraints only through collective acceptance, then the question of whether a new narrative will change a constraint is the question of whether the narrative will be accepted. Acceptance is a matter of power, legitimacy and the figuration of the community. A senior executive's declaration that "we are now an agile organisation" may or may not change constraints, depending on whether it is collectively accepted. A frontline worker's story about how things really work may change constraints if it spreads and is taken up, or vanish if it is not. Anthro-complexity's question, how do we get more stories like these and fewer like those?, is at bottom a question about collective acceptance: which stories will be accepted, by whom, and with what deontic consequences. That question is political as well as epistemic.

Searle and the declarative force of frameworks. Searle's account has a consequence for complexity frameworks that is easy to miss. When a framework is adopted by an organisation, its categories can become status functions. If an organisation adopts Cynefin and declares a situation "complex", the declaration, collectively accepted, creates deontic powers: a licence to probe instead of plan, permission to fail safely, release from the demand for a business case. The framework's category has become an institutional fact. It is no longer only a description of the situation. It is a constraint within the situation, with rights and duties attached. This is the precise sense in which, as The Compass and the Loom puts it, any framework applied to a human system becomes an actor within it. The framework acts by declaration. Searle's formula makes the mechanism exact: this situation (X) counts as complex (Y) in this organisation (C), and the counting creates powers.

Narrative as Social Object

The narrative element of the third mark has its own literature. Jerome Bruner's Actual Minds, Possible Worlds (1986) and Acts of Meaning (1990) distinguished two modes of thought: the paradigmatic, which seeks general truths through logic and evidence, and the narrative, which seeks verisimilitude through stories of human intention and action. Bruner argued that the narrative mode is how humans make sense of the social world, and that a culture's stock of narratives, its "folk psychology", is the medium through which its members understand themselves and one another. Walter Fisher's "narrative paradigm" in communication theory proposed that humans are story-telling beings, storytellers who judge arguments by their narrative coherence and fidelity rather than by formal logic.

The narrative literature supports anthro-complexity's emphasis on narrative as the vehicle of human meaning, and SenseMaker's design reflects it. But the same literature, read carefully, also grounds narrative relationally. For Bruner, narratives are cultural: they draw on a shared stock of plots, characters and evaluations, and they are told to someone, in a context, for a purpose. A story is a social act. Its meaning depends on its audience, and it is shaped by the teller's anticipation of the audience's response, which brings us back to Mead. A micro-narrative collected by SenseMaker is not a window into the narrator's private meaning. It is a story told to an instrument, in response to a prompt, within a frame, anticipating some audience. Its meaning is relational from the start.

The Scientific Reason Assessed

Mark Supporting science What it confirms Where it grounds the mark Consequence for instruments
Multiple identities Social identity theory; self-categorisation (Tajfel, Turner) People switch identities by context In the comparative social context The prompt and signifier set produce the identity signified from
Intentionality Shared intentionality (Tomasello) Human intention differs in kind from ape purposiveness In joint goals, attention and commitment Aggregated individual stories do not show joint intentions
Abstraction & narrative Social ontology (Searle); narrative mode (Bruner) Abstractions create constraints; narrative is the medium of meaning In collective acceptance; in the audience Whether a story changes a constraint depends on acceptance, and acceptance on power

The scientific reason holds. Anthro-complexity is right that the best available science shows humans to differ from other complex-system agents in ways that make ant, swarm and agent-based models mislead. Self-categorisation theory, shared intentionality and social ontology are among the most robust and theoretically developed bodies of work in the human sciences, and each confirms one of anthro-complexity's marks.

The scientific reason also relocates the distinction. On every mark, the science grounds the human difference not in a property individual humans carry but in a relation among humans: the comparative social context of self-categorisation, the shared intentionality of joint commitment, the collective acceptance of status functions. The science does not say humans have multiple identities, intentions and abstractions and therefore form distinctive social systems. It says humans form distinctive social relations (comparative, shared, collectively accepting) and therefore have the identities, intentions and abstractions they have.

That is the reversal this book has been anticipating since Chapter 4, and the science makes it. The human distinction is real, and it is relational. Anthro-complexity states it in a grammar of possession (humans hold, humans act with, humans create) that the science does not support, and the grammar matters, because instruments built for possessors differ from instruments built for relations.

Thesis 3 confirmed and qualified. Snowden is right that human systems differ in kind from other complex systems, and the best available science confirms it. holds

Thesis 4 supported. The reason anthro-complexity gives for the difference is misplaced. Self-categorisation theory, shared intentionality and social ontology each ground the relevant mark in relations among persons, not in properties of persons. holds

The adversary will say this is a quarrel about grammar. Snowden knows identity is contextual: he has written about it for twenty years, and his crews and roles are contextual by definition. He knows intention is social, because his work on narrative is about shared meaning. He knows abstractions need acceptance, because the whole point of "more stories like these" is that stories spread. To say humans "hold" multiple identities is shorthand, not ontology.

The adversary is right that Snowden knows all this, and right that his practice often reflects it. Crews are contextual. SenseMaker gathers shared meaning. But grammar is not neutral, and the November post states the human distinction in the grammar of possession at exactly the moment it is defining anthro-complexity's identity against a neighbour. When a school's self-definition puts the human distinction in what persons have, its instruments will be built to gather what persons have: individual narratives, individually signified, aggregated into distributions. The relational facts the science identifies, comparative contexts, joint commitments and collective acceptance, are not gathered by aggregating individuals. They need different instruments. The quarrel about grammar is a quarrel about what gets built. partial

IDENTITY IS A CATEGORISATION IN A CONTEXT —

INTENTION IS FIRST A WE —

ABSTRACTION IS A DECLARATION SOMEONE ACCEPTED.

Addition: Weick and the Sensemaking Tradition

The scientific reason has a further body of evidence that the chapter above did not take up. It comes from organisation studies itself, and it is closer to anthro-complexity's own vocabulary than any of the sciences cited so far. Karl Weick's work on sensemaking, developed from The Social Psychology of Organizing (1969, revised 1979) through Sensemaking in Organizations (1995) and many studies of organisations in crisis, is the most influential account in management scholarship of how people in organisations make sense of what is happening to them. Anthro-complexity uses the word "sense-making" constantly. Weick's account of what sensemaking is supports anthro-complexity's human distinction and, like the sciences of Chapter 13, grounds it relationally.

Seven properties

Weick identified seven properties of sensemaking.

  1. Grounded in identity construction. Sensemaking begins with a sensemaker, and who she is shapes what she makes of events. But identity is not fixed. It is constructed in the process of sensemaking, as the sensemaker discovers who she is by seeing how others respond to her and how she responds to events.
  2. Retrospective. People make sense of what has already happened. They can know what they have done only after they have done it. Weick's famous formula, adapted from E. M. Forster, is how can I know what I think until I see what I say? That is anthro-complexity's retrospective coherence, at the level of the individual mind.
  3. Enactive of sensible environments. People do not simply discover their environments. They enact them. Their actions create the conditions they then make sense of. A manager who expects resistance acts in ways that produce it, and then makes sense of the resistance she produced.
  4. Social. Sensemaking is never solitary. Even when alone, people make sense in anticipation of others' responses, in terms they share with others and for audiences real or imagined. This is Mead's internalised conversation of gestures, in organisational form.
  5. Ongoing. Sensemaking never starts and never stops. People are always in the middle of things, always already making sense.
  6. Focused on and by extracted cues. People extract cues from the flow of experience, small, familiar structures from which they develop a larger sense of what is going on. What they extract depends on context and on what they are looking for.
  7. Driven by plausibility rather than accuracy. People settle on accounts that are plausible enough to act on, not on accounts that are accurate. Accuracy is often unattainable and always slower. Plausibility lets action proceed.

Mann Gulch

Weick's most famous study, "The Collapse of Sensemaking in Organizations: The Mann Gulch Disaster" (1993), examined the 1949 fire in Montana in which thirteen smokejumpers died. Weick, drawing on Norman Maclean's book Young Men and Fire, argued that the disaster was a collapse of sensemaking. The crew had arrived expecting a routine "10 o'clock fire", one that would be under control by ten the next morning. When the fire blew up and raced towards them, the crew's sense of what was happening, and of who they were as a crew, collapsed. Their foreman, Wagner Dodge, lit an escape fire in the grass ahead of them and lay down in its ashes, ordering his men to join him. They did not. The order made no sense in their frame, and the crew's structure, its roles and its sense of itself as a crew had disintegrated in the panic. Dodge survived in his escape fire. Two others survived by running for a ridge. Thirteen died.

Weick drew a lesson that bears on anthro-complexity's human distinction. The crew's sensemaking was social. It depended on their roles, their trust in their foreman and their shared sense of the situation. When the social structure collapsed, sensemaking collapsed with it. Dodge's escape fire was an act of genuine novelty, an improvisation that saved his life. But it could not save the crew, because novelty needs a social structure within which it can be taken up. A gesture whose meaning is not completed by the response, an order not followed, saves only the one who makes it. Weick drew lessons about the resources that sustain sensemaking under pressure: improvisation, bricolage (Weick also used Lévi-Strauss's term), respectful interaction, and the wisdom of knowing what one does not know.

What Weick adds to the scientific reason

Weick's work confirms each of anthro-complexity's three marks, and grounds each relationally.

Mark Weick's property Ground
Multiple identities Grounded in identity construction Identity constructed in sensemaking, through others' responses
Intentionality Enactive; retrospective Intention discovered in action; how can I know what I think until I see what I say?
Abstraction and narrative Social; driven by plausibility Shared plausible accounts, taken up by others, govern action

Weick also adds something the earlier sciences did not: the enactment of environments. People's actions create the conditions they then make sense of. That is descriptive reflexivity (Chapter 14) at the level of practice. Organisations do not simply respond to environments described to them. They produce the environments their descriptions describe. Weick's enactment is the organisational form of the self-fulfilling prophecy and of Barnesian performativity (Chapter 34).

Weick and the instruments

Weick's account bears on anthro-complexity's instruments in three ways.

Sensemaking is ongoing; instruments are episodic. A SenseMaker round captures stories at a moment. But sensemaking never stops. The stories captured are snapshots of a process that continues, and the process will absorb the instrument's returned patterns as cues. The instrument is not outside the sensemaking. It is one of its cues.

Sensemaking seeks plausibility; patterns offer it. When patterns are returned to a population, they offer a plausible account of what is happening, and populations, Weick predicts, will tend to adopt plausible accounts and act on them. A pattern's plausibility, its clarity and its visual authority on a dashboard may lead it to be adopted whether or not it is accurate. That is the mechanism of Barnesian performativity in narrative sensing.

Sensemaking collapses when its social structure does. Mann Gulch shows that sensemaking under pressure depends on roles, trust and respectful interaction. Instruments deployed in crisis, as anthro-complexity's often are (in public health, counter-terrorism, organisational turmoil), will capture sensemaking in a state of partial collapse. What they capture is not "what people experience" but what people can say when the structures that let them make sense are under strain. A reflexive instrument would attend to the state of the social structure in which sensemaking takes place, and not only to its outputs.

HOW CAN I KNOW WHAT I THINK UNTIL I SEE WHAT I SAY? —

SENSEMAKING IS SOCIAL, RETROSPECTIVE, ENACTIVE —

AND THE ESCAPE FIRE SAVES ONLY THOSE WHO FOLLOW.

Chapter 14

The Reflexive Reason: Looping Kinds and the Double Hermeneutic

4,604 words, about 20 minutes of reading

The Deepest of the Five

Of the five reasons anthro-complexity has for treating human complexity differently, the reflexive reason is the deepest. It is also the one the November post states least directly. It appears not as one of the three marks but as a consequence of the third ("abstractions and shared narratives that change the very constraints they operate within") and in the asymmetry principle ("only one of them can change the story it tells about the other"). The Compass and the Loom states it outright: any framework applied to a human system becomes an actor within it, so human systems are reflexively complex, and describing them changes them.

This chapter argues that reflexivity is the ground of the other reasons. The scientific reason, read through self-categorisation theory, shared intentionality and social ontology, turns out to rest on it. The anti-computational, ethical and positional reasons each depend on it in ways later chapters will show. Reflexivity is also the reason that turns back on anyone who states it. A theory that says human systems are changed by their descriptions is itself a description of human systems, and it changes them.

The chapter sets out five forms of social reflexivity from the literature, asks whether reflexivity is uniquely human, takes up the problem of realism flagged in Chapter 2, and ends with the lesson of second-order cybernetics, which met this problem fifty years ago and drew a conclusion anthro-complexity has not yet drawn.

Five Forms of Social Reflexivity

"Reflexivity" names a family of related phenomena, and they need to be told apart. Five are well established in the social sciences.

1. The self-fulfilling and self-defeating prophecy

Robert K. Merton named the self-fulfilling prophecy in a 1948 essay. His central example was a run on a solvent bank. A rumour that the bank is insolvent leads depositors to withdraw their money, and the withdrawals make the bank insolvent. The prophecy, false when made, becomes true because it was made. Merton defined the phenomenon as a false definition of the situation evoking new behaviour that makes the originally false conception come true. He traced it to W. I. Thomas's theorem: if men define situations as real, they are real in their consequences.

The self-defeating prophecy is the converse. A prediction leads to behaviour that prevents what was predicted. A forecast of traffic congestion leads drivers to take other routes, so the congestion does not happen. A forecast of an epidemic leads to precautions that prevent it. The prophecy, true when made, becomes false because it was made.

Both phenomena turn on public prediction. A private prediction does not change behaviour. A published one can. In the 1950s Herbert Simon, and separately Emile Grunberg and Franco Modigliani, asked whether a correct public prediction is possible at all when publishing it changes the outcome. They showed that it can be, under certain conditions, as a fixed point: a prediction which, once known, produces exactly the behaviour it predicts. But the fixed point may not exist, may not be unique, and may be unstable. In general, the social scientist who publishes a prediction is an actor in the system she predicts.

2. Looping kinds

Ian Hacking's work on "making up people" (1986), the looping effects of human kinds (1995) and interactive kinds (in The Social Construction of What?, 1999) identified a subtler form of reflexivity. Classifications of people, such as "child abuser", "multiple personality", "autistic", "refugee" or "high-potential employee", do not just describe people who already exist. They make up people, in the sense that people classified come to experience themselves, and to behave, in terms of the classification. Then the change in the people changes the classification: the experts revise their criteria, the category shifts, and the new category makes up people differently. Hacking called this the looping effect.

Hacking distinguished indifferent kinds, whose members are unaffected by being classified, from interactive kinds, whose members are affected. Quarks are an indifferent kind. Classifying something as a quark does not change it, because quarks are not aware of the classification. "Women refugees" is an interactive kind. Being classified as one changes how a person is treated, how she sees herself and what she does, and her changed behaviour feeds back into what the classification means. Interactive kinds are moving targets. They change as they are studied, because studying them is part of what changes them.

3. The double hermeneutic

Anthony Giddens, in New Rules of Sociological Method (1976) and The Constitution of Society (1984), named the double hermeneutic. The natural sciences have a single hermeneutic: they interpret a world of objects that do not themselves interpret. The social sciences have a double one. They interpret a world of subjects who already interpret their own world, and the social scientists' interpretations are taken up by those subjects and change their self-interpretations. Concepts like "sovereignty", "citizenship", "class" and "organisational culture" began as social-scientific concepts and have become part of how people understand and conduct their lives. The social sciences, Giddens argued, are not outside the world they study. They are among its constituent practices.

Giddens drew a consequence that matters for anthro-complexity. The social sciences seem to produce fewer "discoveries" than the natural sciences, not because they are less successful but because their successes are absorbed into the world they study and become common sense. A social-scientific concept that works becomes part of how people think, and so no longer seems like a discovery. The same may be true of complexity concepts in organisations. "Complex", "emergent", "safe-to-fail" and "probe" have entered managerial vocabulary, where they now shape the situations they once described.

4. Reflexivity in markets

George Soros, drawing on Karl Popper, built a theory of financial markets around reflexivity, set out in The Alchemy of Finance (1987) and later work. Market participants' views of the market have two functions: a cognitive function, understanding the market, and a manipulative function, acting on it. The two interfere. Participants act on their understanding, their actions change the market, and the changed market changes their understanding. Prices do not converge on a fundamental value that exists independently of participants' views. Participants' views help determine fundamental values (a company whose share price rises can raise capital more cheaply, which improves its fundamentals), so there is no independent anchor. Booms and busts are reflexive processes in which views and fundamentals reinforce each other until the gap between them becomes unsustainable.

Soros's theory is open to criticism as economics, but it names a form of reflexivity that Arthur's El Farol model (Chapter 7) formalises in a limited way. Its significance here is that it makes reflexivity a property of the system, not just of individual participants. A reflexive market is not one in which some traders are self-aware. It is one whose dynamics are driven by the interference between cognition and manipulation across all participants.

5. Performativity

The fifth form, performativity, was developed in the sociology of finance by Michel Callon and Donald MacKenzie. Economic models do not only describe markets. They perform them, shaping the markets they model. MacKenzie's An Engine, Not a Camera (2006) is the classic study. The Black–Scholes–Merton options-pricing model, published in 1973, was at first a poor fit to observed prices. As traders adopted it, using it to price options and identify mispricing, prices converged on the model's predictions. The model became more accurate because it was used. After the crash of 1987, prices departed systematically from the model in a pattern, the "volatility smile", that persisted. MacKenzie called this counter-performativity: use of the model, in conditions it did not anticipate, helped produce outcomes that contradicted it.

The Compass and the Loom draws the general lesson. Instruments are never neutral. Published assessments can push organisations towards them, away from them, or into oscillation around them. Named observables become targets. Snowden's own acumen framework is safe as a sensitising concept but unsafe as a score.

The five compared

Form Key source What reflexes Mechanism Unit
Self-fulfilling/defeating prophecy Merton; Thomas; Simon; Grunberg & Modigliani A prediction Public prediction changes behaviour Event
Looping kinds Hacking A classification of people Classified people change; classification changes Category
Double hermeneutic Giddens A social-scientific concept Concepts enter lay self-understanding Concept
Market reflexivity Soros Participants' views Cognition and manipulation interfere System
Performativity Callon; MacKenzie A model Model in use shapes the market it models Model / instrument

All five share a structure. A description of a human system (a prediction, classification, concept, view or model) enters the system, through publication, adoption or use, and changes it. The changed system then makes the description more true (performativity, self-fulfilment), less true (counter-performativity, self-defeat), or different (looping, the double hermeneutic). In every case the describer is not outside the system. Her description is one of its constituent processes.

Is Reflexivity Uniquely Human?

Chapter 9 conceded that first-order reflexivity, behaviour shaped by others' likely inferences about it, is not unique to humans. The cleaner wrasse behaves more cooperatively when watched by potential clients. Many social animals show audience effects, tactical deception and reputation management. If reflexivity in that sense is the human distinction, it is a difference of degree.

The five forms above, however, are not first-order reflexivity in the wrasse's sense. Each involves a description: a prediction, classification, concept, view or model, which is a public, shareable, symbolic representation of the system. It enters the system as a description and changes it as a description. The wrasse adjusts its behaviour to its audience's inferences, but it does not publish a theory of cleaning that cleaners and clients then take up. No animal is known to produce descriptions of its social system that the system's members adopt and are changed by. That requires what Mead called significant symbols, what Tomasello calls shared intentionality and what Searle calls status-function declarations: representations whose meaning is shared, so that a description offered by one member can be taken up by all.

Call this descriptive reflexivity: the reflexivity of a system changed by public descriptions of itself. Descriptive reflexivity is, as far as is known, uniquely human. It is the difference of kind. And it is relational by definition, because a description changes a system only when it is shared, taken up and acted on by members of the system.

The reflexive ground. What distinguishes human from other complex systems in kind is descriptive reflexivity: human systems are changed by public descriptions of themselves, which their members take up and act on. Descriptive reflexivity depends on shared symbols, shared intentionality and collective acceptance, so it is relational, not a property of individuals. Anthro-complexity's three marks (multiple identity, intention, abstraction) are the capacities that make descriptive reflexivity possible, but the distinction of kind lies in the reflexive relation, not in the capacities.

The Problem of Realism

Chapter 2 flagged a tension in Element 1. Anthro-complexity is realist about complexity: some systems really are complex, whatever anyone thinks of them. If human systems are descriptively reflexive, if what people think about them changes them, how can one be a realist about their complexity?

The tension is real, but not fatal. Hacking's work shows how to handle it. Interactive kinds are real. Being classified as a refugee has real consequences, and the people so classified are really affected. But they are not indifferent. Their reality includes their response to being described. A realist about interactive kinds says that there is a fact of the matter about a human system at any moment, including the fact that it is responding to descriptions of itself, and that this fact can be known, with difficulty and provisionally. What she cannot say is that the fact is independent of descriptions. The descriptions are part of what the fact is about.

For anthro-complexity, this means a situation's complexity is real but not indifferent. Whether a situation is complex is a fact about it. Whether it is described as complex is another fact, which may change the first. A situation described as complex, and collectively accepted as such (Searle), becomes a situation in which people are licensed to probe, fail safely and refuse plans. Those licences change how it unfolds. They may make it more complex, as experiments multiply and interactions proliferate, or less, as the licence to probe lets people find stable patterns that were invisible under the demand for a plan.

The realist anthro-complexity theorist can live with this. She can say that a situation is complex, that saying so changes it, and that a good practitioner takes both facts into account. What she cannot say is that her categorisation is a neutral reading of a fact that would have been the same without it. The categorisation is a move. Cynefin's own logic, that categorising is a decision, already acknowledges this. The question is whether its instruments do.

Anthro-Complexity's Reflexive Reason, Stated

With the five forms and the reflexive ground in hand, anthro-complexity's reflexive reason can be stated more precisely than the November post does.

  1. Human systems are descriptively reflexive. Public descriptions of them, taken up by their members, change them (Merton, Hacking, Giddens, Soros, MacKenzie).
  2. Therefore human systems cannot be modelled as CAS models them, from a position outside the system whose descriptions do not affect it. A published model of a human system is a description that enters the system (Chapter 8, Dimension 10).
  3. Therefore human systems cannot be predicted in the forward-causal sense. Predictions are descriptions, and published predictions change outcomes (the Simon and Grunberg–Modigliani problem).
  4. Therefore intervention in human systems must proceed by probe, sense and respond: small actions whose effects, including effects on how people describe the system, are sensed and responded to.
  5. Therefore any framework applied to a human system becomes an actor within it.

The argument is valid, and its premises are well grounded. It is anthro-complexity's strongest argument, and it explains why human complexity must be treated differently.

The Reflexive Reason Turned Back

Step 5 is where the argument turns. Any framework applied to a human system becomes an actor within it. Cynefin is a framework. SenseMaker is a framework embodied in software. Estuarine mapping is a framework embodied in a practice. SwarmCompass and WRAS are frameworks embodied in products. Each, when applied, becomes an actor in the system it is applied to.

What kind of actor? The five forms answer:

Form Anthro-complexity instrument as actor
Self-fulfilling prophecy Categorising a situation as complex licenses experiment, which may produce the emergence that confirms the categorisation
Looping kinds SenseMaker signifiers, once known, become categories narrators use to understand their own experience. "Is this a story of trust or of compliance?" becomes how people sort their experiences
Double hermeneutic "Safe-to-fail", "probe", "weak signal", "constraint" enter organisational vocabulary and reshape the conversations they were meant to observe
Market reflexivity Patterns fed back to a population change the narratives the population tells; the next round of sensing reads the effect of the last
Performativity An estuarine map, once accepted, defines what counts as costly to change, and the organisation behaves accordingly; the map makes the terrain it maps

None of this is a criticism in itself. Every framework applied to a human system is an actor in it. CAS models, systems frameworks, CRP conversations and this disputation are all actors. Snowden's insight is correct and general. The criticism is that anthro-complexity's instruments are not designed as actors. They are designed as sensors: instruments that read the system and report what they find. SenseMaker reads narratives. Estuarine mapping reads constraints. Cynefin reads situations. Each is presented as a way of seeing what is there, so that action can be taken on what is seen. None is presented as a way of acting on the system by describing it, though on anthro-complexity's own reflexive reason that is what each is.

The Lesson of Second-Order Cybernetics

The problem has been met before. In the 1960s and 1970s cybernetics underwent a reflexive turn, led by Heinz von Foerster, Humberto Maturana, Francisco Varela, Gregory Bateson and Margaret Mead (no relation to George Herbert). First-order cybernetics had studied observed systems: thermostats, servomechanisms, homeostats, nervous systems, all described by an observer outside them. Second-order cybernetics studied observing systems, systems that include the observer. Its founding insight, in Maturana's phrase, was that everything said is said by an observer. Von Foerster's formulation was that the cybernetics of observed systems must give way to the cybernetics of observing systems, in which the observer accounts for herself as part of what she observes.

Second-order cybernetics drew three conclusions that matter here.

First, the observer must be included in the description. A description that does not account for its describer is incomplete, because the describer's position, distinctions and purposes shape what is described. Von Foerster's students were taught to ask of every description: who is saying this, from where, and why?

Second, objectivity is replaced by responsibility. If there is no view from nowhere, the observer cannot claim objectivity in the sense of a description independent of herself. She can take responsibility for her description: acknowledge her position, state her distinctions, and accept that her description has consequences. Von Foerster put this as an ethical imperative: act always so as to increase the number of choices. A description that closes down possibilities is an act with ethical weight, and the describer is responsible for it.

Third, the system observed is a system that observes. In social systems the people observed are themselves observers, describing the system and one another, and the scientist's description enters a field of descriptions. Niklas Luhmann, who took this up in his systems theory of society, made "second-order observation", the observation of observations, the defining operation of modern social systems.

Anthro-complexity and second-order cybernetics

Anthro-complexity's reflexive reason is a second-order cybernetic insight: any framework applied to a human system becomes an actor within it. But anthro-complexity has not drawn second-order cybernetics' conclusions about itself. Its instruments do not include the observer in the description. SenseMaker shows patterns in narratives, but not the designer's position, distinctions or purposes in choosing the signifiers. Its instruments do not replace objectivity with responsibility. They are presented as ways of seeing what is there, not as descriptions whose consequences the describer owns. And they do not treat the system observed as a system that observes. Narrators self-signify, which is a step in that direction, but the narrators' observations of the instrument (what they think the study is for, what they expect will be done with it, how they shape their stories accordingly) are not part of what is sensed.

That is the gap between anthro-complexity's reflexive reason and its instruments, and it is the gap Book Four explores. The reflexive reason is first-rate. It is the deepest of the five reasons, the one that grounds the difference of kind, and it is correct. It is stated as a fact about human systems, applied to other people's frameworks (Ing's eurhythmia, the systems-change industry, the CAS models), and not yet applied to anthro-complexity's own.

Proposition granted: Any framework applied to a human system becomes an actor within it.

Then it must be granted that Cynefin, SenseMaker and estuarine mapping, when applied, are actors in the systems they are applied to. — Granted.

Then it must be granted that their effects as actors (on what people notice, how they describe their experience, what they treat as costly to change, what they are licensed to do) are part of what happens when they are applied. — Granted.

Then it must be granted that an account of what these instruments reveal is incomplete unless it includes what they do. — Granted.

Then it must be granted that the instruments should be designed to sense their own effects as actors: to record how probes change the prober, how signifiers shape signification, how maps make terrain. — Granted in principle. This is The Compass and the Loom's demand that instruments should record how probes change the prober.

Then the respondent must say whether the instruments as currently designed do this. — Doubted. Self-signification is a partial step. The narrator's own interpretation is captured. The designer's position is not, the instrument's effect on narration is not, and the effect of feedback on subsequent narration is not separately tracked. The gap between the granted principle and the current design is what Book Four examines.

EVERYTHING SAID IS SAID BY AN OBSERVER —

EVERY FRAMEWORK APPLIED BECOMES AN ACTOR —

INCLUDING THIS ONE, INCLUDING YOURS.

Addition: Luhmann and the Blind Spot

The chapter ended with second-order cybernetics, von Foerster's cybernetics of observing systems and Maturana's everything said is said by an observer, and noted that Niklas Luhmann took up the reflexive turn in his systems theory of society. Luhmann deserves more than a mention. His theory is the most systematic attempt in social theory to build a whole account of society on second-order observation, and it offers a precise vocabulary for what this disputation has called the exemption: the blind spot that every observation has, and that no observer can see.

Observation as distinction

Luhmann, following Spencer-Brown (Chapter 18), defined observation as the use of a distinction to indicate one side rather than the other. To observe is to draw a distinction, such as system/environment, true/false, legal/illegal or complex/ordered, and to indicate one side. Every observation is therefore selective: it sees what its distinction makes visible, and it does not see what the distinction excludes.

Above all, an observation cannot see its own distinction while using it. The distinction is the blind spot of the observation. An observer using the distinction legal/illegal can see what is legal and what is illegal. She cannot, while using that distinction, see the distinction itself, its contingency, its history or the alternatives it excludes. To see the distinction she must use another distinction, observing her first observation from a second-order position. But the second-order observation has its own blind spot, its own distinction, which it cannot see while using it. There is no observation without a blind spot, and no position from which all blind spots are visible.

Second-order observation

Luhmann argued that modern society is distinguished by the prevalence of second-order observation: the observation of observations. In science, scientists observe one another's observations through peer review and replication. In the economy, market participants observe one another's observations through prices, which are observations of what others will pay. In politics, politicians observe public opinion, which is an observation of politicians. In the mass media, journalists observe what others are observing. Modern society is a society of observers observing observers.

Second-order observation has a characteristic power. The second-order observer can see what the first-order observer cannot, namely the first-order observer's distinction, her blind spot. The second-order observer can say: you see it that way because you use that distinction; with another distinction, you would see it differently. That is the critic's power, and the social scientist's, the consultant's and the analyst's. But the second-order observer has her own blind spot, which a third-order observer could see, and so on.

The exemption, in Luhmann's terms

Luhmann's vocabulary states the exemption of Thesis II precisely. Anthro-complexity's instruments are second-order observers. They observe a population's observations, its narratives, which are its members' observations of their experience. The instruments can see the population's distinctions. They can show, through patterns, that a population sees its situation in terms of compliance and fear, or trust and autonomy. That is their power.

But the instruments have their own blind spot: their own distinctions, the signifier set, the triad vertices and the categories of the map. While using them, the instruments cannot see them. An instrument that shows a population's distinctions cannot, at the same time, show its own. The pattern is visible, and the frame that produced the pattern is the blind spot. That is why declination tables, the reading of the reader and frame rotation are needed. They are ways of making the instrument's own distinctions visible, from a further position, the third order, so that the population, or the practitioner, can see the instrument's blind spot, as the instrument saw the population's.

And the disputation itself is a third-order observer. It observes anthro-complexity's instruments, which observe populations' narratives, which observe experience. It can see the instruments' blind spot, their distinctions. It has its own blind spot, its own distinctions (possessive/relational, inside/outside, estuary/meshwork), which it cannot see while using them. The adversary's Article 10, you are standing outside, is a fourth-order observation of the disputation's blind spot. Luhmann would say there is no end to this, and that the regress is not a defect but the condition of modern observation.

What Luhmann adds

Luhmann adds to the disputation a principle and a caution.

The principle. Every observation has a blind spot, and the blind spot is its own distinction. No instrument, framework or disputation can see its own distinctions while using them. Reflexive practice is therefore not the elimination of blind spots, which is impossible, but the institutionalisation of second-order observation: building into practice the means by which one's distinctions can be observed by others, from other positions, with other distinctions. Declination, rotation, return and the reading of the reader are institutions of second-order observation.

The caution. Luhmann's theory is famously austere about human beings. For Luhmann, social systems consist of communications, not people. People are in the environment of social systems, as psychic systems whose consciousness is not part of the social system's operations. Luhmann's theory would seem, therefore, to stand at the opposite pole from anthro-complexity's human distinction, and from this disputation's account of persons as formed in relation. The caution is that second-order observation, borrowed from Luhmann, should not bring Luhmann's anti-humanism with it. The disputation takes from Luhmann the logic of observation and the blind spot. It does not take his exclusion of persons from social systems, which Archer's emergent personhood (addition to Chapter 2) and Mead's social self both reject.

Luhmann, Snowden and the blind spot

There is an irony in the encounter between Luhmann and anthro-complexity that should be stated. Luhmann's theory is a systems theory, the most elaborate in twentieth-century social thought, and it begins from the system/environment distinction that Snowden's November post argues complexity should not presume. Yet Luhmann's account of observation, distinction and the blind spot is among the most rigorous available statements of the reflexivity that anthro-complexity's own asymmetry principle implies. A systems theorist supplies anthro-complexity with the logic it needs to see its own blind spot. That is a case of the neighbours borrowing from each other, which is what the November post's conclusion recommends. The better relationship, in its words, is that of neighbours who borrow from each other, not parent and child.

EVERY OBSERVATION HAS A BLIND SPOT, AND THE BLIND SPOT IS ITS OWN DISTINCTION —

NO ONE SEES THE FORM THEY ARE STANDING IN —

INSTITUTIONALISE THE OBSERVATION OF YOUR OBSERVATIONS.

Addition: Cross-Lights From Book Nine

Snowden's own statement of the reflexive reason. In January 2024 he wrote that a complex system can be understood only by acting on it, so that all forms of diagnosis are interventions and all interventions should be designed as diagnostic instruments. That is the argument of this chapter in a sentence, by the author it is addressed to. Chapter 56 draws the consequence: his instruments cannot merely read, and should be designed to show what they are doing.

Jaeggi on problems. Rahel Jaeggi holds that problems are at once given and made. Something goes wrong and announces itself. What the problem is has to be worked out, with descriptions that society supplies. That is the looping of kinds described at the level of a single diagnosis.

Chapter 15

The Anti-Computational Reason: Against the Programme

4,096 words, about 18 minutes of reading

The Claim and Its Target

The third reason anthro-complexity has for treating human complexity differently is negative. Human cognition is not the execution of a programme, so models that represent it as one misrepresent it. The November post states the reason against Mayr's teleonomy: a programme is a computational metaphor, and much of the anthro-complexity series has been an argument against picturing the human brain as a thing that runs programmes. The target is wider than Mayr. It includes the computational theory of mind that dominated cognitive science from the 1960s to the 1990s, the agent-based models of CAS theory, and every management practice that treats people as information processors whose inputs can be designed and whose outputs predicted.

This chapter examines the anti-computational reason in three steps. First, it sets out the tradition of argument against the computational theory of mind and finds it largely sound. Second, it notes a complication. The most influential contemporary alternative, predictive processing, is itself a computational theory of a kind, so the anti-computational reason must be stated with care to avoid rejecting the very neuroscience it draws on. Third, it asks where the best anti-computational science locates the distinctively human, and finds again that the answer is relational. Embodied, enactive and distributed accounts of cognition place it between organism and world, and between people, not inside a head. A final section notes the tension between anthro-complexity's anti-computational ontology and the increasingly computational character of its practice, which Thesis VI (Chapter 24) develops.

The Computational Theory of Mind

The computational theory of mind holds that thinking is computation: the manipulation of symbolic representations according to formal rules. Its classical statement came from the convergence, in the 1950s and 1960s, of Turing's theory of computation, Newell and Simon's work on problem-solving programs, Chomsky's generative linguistics and Fodor's philosophy of mind. On this view the mind is to the brain as software is to hardware. Cognitive processes are programs running on neural machinery, and they can in principle be specified independently of the machinery, as a program can be specified independently of the computer that runs it.

The theory had real successes. It made cognitive science possible as a discipline, gave psychology a vocabulary for internal processes that behaviourism had forbidden, and produced artificial intelligence. Its characteristic picture of the human (a symbol-processor taking inputs from the senses, computing over representations, and producing outputs as actions) became the default picture in much of management science, decision theory and organisational design.

The Arguments Against

The arguments against the computational theory are many, and several are directly relevant to anthro-complexity.

Dreyfus: the background

Hubert Dreyfus's What Computers Can't Do (1972, revised 1979 and 1992) argued, from Heidegger and Merleau-Ponty, that human intelligence depends on a background of embodied, situated, practical know-how that cannot be represented as explicit rules. When we cope skilfully with the world, walking through a crowd, holding a conversation or using a hammer, we do not consult representations. We respond directly to the situation as it shows up for us, shaped by bodily skills and cultural practices we could not articulate. Attempts to formalise this background produce an infinite regress, because each rule needs further rules for its application. Dreyfus predicted that symbolic AI would fail at tasks requiring common sense, and on its own terms it did.

Dreyfus's argument supports Snowden's emphasis on tacit knowledge, mētis and phronesis. What humans know is largely background, embodied and situated, not representable as rules.

Searle: syntax is not semantics

John Searle's "Chinese room" argument (1980) imagined a person in a room who receives Chinese characters, follows rules in a book for manipulating them, and passes out other characters. To an observer outside, the room seems to understand Chinese. The person inside understands nothing. Searle concluded that symbol manipulation alone, syntax, cannot produce understanding, semantics, so a computer running a program does not understand, however well it simulates understanding.

The argument has been much disputed, but its core point, that following rules for manipulating symbols is not the same as grasping their meaning, supports anthro-complexity's insistence that human meaning-making is not computation.

The enactivists: cognition is embodied action

Francisco Varela, Evan Thompson and Eleanor Rosch's The Embodied Mind (1991) proposed enaction as an alternative to computation. Cognition is not the representation of a pre-given world by a pre-given mind. It is the enactment of a world through the history of structural coupling between an organism and its environment. Perception is not the reconstruction of an external world in internal symbols but a form of skilled action, an active exploration of the environment guided by sensorimotor expectations. Meaning is not computed from representations. It is brought forth in the organism's engagement with what matters to it.

Enactivism has since developed into a substantial research programme. Thompson's Mind in Life (2007) extended it to the continuity between life and mind. Di Paolo and colleagues developed its account of agency and sensorimotor life. De Jaegher and Di Paolo's "participatory sense-making" (2007) extended it to social cognition. That last development matters below.

Embodied and extended cognition

Andy Clark and David Chalmers's "The Extended Mind" (1998) argued that cognitive processes are not confined to the brain. When a person uses a notebook to remember, or a calculator to compute, the notebook and calculator are, under the right conditions, part of the cognitive system. Cognition extends into the body and the world. Clark's later work and the wider movement of "4E cognition" (embodied, embedded, enactive, extended) have made this view mainstream in philosophy of mind.

Developmental systems: against the genetic programme

A less familiar line of argument targets the metaphor at the root of Mayr's teleonomy, the genetic programme. Susan Oyama's The Ontogeny of Information (1985) and the developmental systems theory that followed argued that the genome is not a programme for building an organism. Development is the product of interaction among many resources (genes, cellular machinery, maternal environment, temperature, nutrition, behaviour), none of which contains the "information" for the outcome on its own. Information is constructed in development, not transmitted by genes. The genetic programme is a metaphor that privileges one developmental resource and obscures the others.

Developmental systems theory supports Snowden's suspicion of teleonomy at its root. If even organisms are not run by programmes, the case that humans are is weaker still.

The Complication: Predictive Processing

The anti-computational reason has a complication that its proponents must face. The most influential contemporary framework in cognitive neuroscience, often cited in anthro-complexity's discussions of how the brain works, is predictive processing. Associated with Karl Friston's free-energy principle, Andy Clark's Surfing Uncertainty (2016) and Jakob Hohwy's The Predictive Mind (2013), predictive processing holds that the brain is fundamentally a prediction machine. It constantly generates predictions about incoming sensory signals from an internal generative model, compares them with the actual signals, and updates the model to minimise prediction error. Perception is "controlled hallucination": the brain's best guess about the causes of its sensory input, corrected by error signals. Action is a way of making the world conform to predictions. Attention is the weighting of prediction errors by their expected reliability.

Predictive processing has three features that bear on the anti-computational reason.

It is computational. It describes the brain as performing approximate Bayesian inference over a hierarchical generative model. That is computation in a precise mathematical sense, though not symbolic rule-following. Friston's free-energy principle is a variational principle from statistical physics, applied to self-organising systems. It is not the computational theory of mind in Fodor's sense, but it is a computational theory.

It supports much of what anthro-complexity says about cognition. Pattern-based cognition, the dominance of prior expectation over sensory evidence, susceptibility to framing and priming, and the brain as an organ that acts on expectations: all are central to predictive processing. When Snowden says human cognition matches the present against fragments of past experience and acts on the first fit, he is describing something close to the predictive brain's reliance on its generative model.

Its leading philosophical interpreter is surfing. The title of Clark's book is Surfing Uncertainty. The image of cognition as surfing, riding the wave of prediction error, matching one's model to the flow of the world, is Ing's image of propensity as reading the wave, here applied to the brain. The coincidence is instructive. The surfing image places the surfer, the brain, in a relation to the wave, the world, in which the surfer reads and the wave is read.

Stating the anti-computational reason with care

The complication means the anti-computational reason must be stated carefully. Anthro-complexity cannot reject all computational accounts of cognition, because the neuroscience it draws on is computational in the predictive-processing sense. What it can reject, and what the November post's target, Mayr's programme, actually is, is the account of cognition as the execution of a specifiable program: a fixed set of rules or instructions that determines behaviour given inputs, and that a modeller could write down. Predictive processing is not that. The generative model is learned, continuously updated, shaped by embodied action and context-sensitive in its weighting of evidence. It is a computational dynamics, not a program.

The careful statement of the anti-computational reason is therefore:

The anti-computational reason, stated with care. Human cognition is not the execution of a specifiable program, and cannot be modelled as rule-following agents whose rules a modeller could write down. It may be described in computational terms as embodied, continuously learning, context-sensitive inference, but such descriptions do not yield programs that predict behaviour, because the generative models they posit are shaped by histories of embodied and social engagement that no modeller can specify.

Stated this way, the reason holds. It rules out the CAS agent, the agent-based model's rule-follower and the management science of designable inputs and predictable outputs. It does not rule out computational neuroscience.

Where the Anti-Computational Science Locates the Human

The most important question for this disputation is where the anti-computational science locates what is distinctively human. The answer, again, is relational.

Participatory sense-making

Hanne De Jaegher and Ezequiel Di Paolo's "participatory sense-making" (2007) extended enactivism from individual to social cognition. Classical theories of social cognition, both the "theory theory" (we understand others by theorising about their minds) and "simulation theory" (we understand others by simulating their minds in our own), treat social understanding as something an individual does: one mind modelling another. De Jaegher and Di Paolo argued that social understanding emerges in interaction. When two people interact, the interaction itself acquires an autonomy, a dynamics of its own that neither controls, and meaning is generated in the coordination of the interaction, not in either participant's head. Two people trying to pass each other in a narrow corridor and repeatedly stepping the same way are caught in an interaction dynamic that neither intends and both participate in. The meaning of the encounter, its awkwardness and humour and eventual resolution, is made between them.

The Compass and the Loom cites participatory sense-making as support for its claim of collective acumen: meaning sits in coordination between agents. The point here is sharper. Participatory sense-making is enactivism's account of distinctively human, or at least distinctively social, cognition, and it is relational. The anti-computational science that anthro-complexity draws on, when it turns to the social, places meaning between people.

Distributed cognition

Edwin Hutchins's Cognition in the Wild (1995), examined in Chapter 31, made the same move through ethnography. Hutchins studied navigation on a US Navy ship and showed that the cognitive task of fixing the ship's position was accomplished not by any individual but by a system of people, instruments, charts and procedures. Cognition was distributed across the socio-technical system. No one person knew where the ship was. The ship's crew, with its instruments and practices, did. Hutchins argued that much human cognition is like this: socially distributed, mediated by artefacts, and accomplished in coordination. The individual cognitive capacities of the navigators mattered, but the distinctively human achievement, the reliable fixing of position at sea, belonged to the system.

Vygotsky and the social origin of higher functions

Behind both participatory sense-making and distributed cognition stands Lev Vygotsky's principle that higher mental functions appear first between people, as social interaction, and only later within the individual, as internalised processes. A child first solves problems with an adult's help, in what Vygotsky called the zone of proximal development, and only later alone. Thought is internalised dialogue. Voluntary attention is internalised social direction of attention. Vygotsky's principle is the developmental psychologist's version of Mead's claim that mind is internalised social process, and of Tomasello's claim that human thinking is internalised shared intentionality.

The convergence

The anti-computational sciences converge on a relational account of the distinctively human:

Source Anti-computational claim Where the human is located
Dreyfus Intelligence depends on a background that cannot be represented In shared cultural practices and bodily skills
Enactivism Cognition is enactment through structural coupling Between organism and world
Participatory sense-making Social meaning emerges in interaction Between people, in the coordination
Extended mind Cognition extends into artefacts In the coupling of person and tool
Distributed cognition Cognitive tasks are accomplished by socio-technical systems Across people, instruments and practices
Developmental systems No programme; information constructed in development In the interaction of developmental resources
Vygotsky Higher functions appear first between people In social interaction, later internalised

Every row places cognition, and specifically distinctively human cognition, in a relation: between organism and world, between people, across people and artefacts. None places it in a property of an individual head. The anti-computational reason, followed through the science that supports it, arrives at the same place as the scientific reason in Chapter 13 and the reflexive reason in Chapter 14. The human distinction is relational.

The Tension With Practice

A final section notes the tension Thesis VI will develop. Anthro-complexity's anti-computational ontology is well grounded. Its practice has grown increasingly computational. SenseMaker is software. Narratives are collected as data, signified as coordinates and aggregated into statistical patterns. Recent instruments, as The Compass and the Loom documents, move towards a human/AI hybrid in which human noticing is gathered as input for automated systems. The brochure's deliverables, on that treatise's reading, turn human noticing into training data for automation, so that human "superpowers" are captured in order to be automated.

This is not a simple contradiction. Anthro-complexity can coherently hold both that humans are not computers and that computers can help humans see patterns in their own experience. The question is what happens to the human in the translation from narrative to data. On the anti-computational science's own account, the human is relational: meaning is made between people, cognition is distributed across people and artefacts, and sense-making is participatory. When narratives are extracted from the relations in which they were made, coded on signifiers and aggregated, the relational character of the meaning is precisely what is lost. A micro-narrative in a database is not participatory sense-making. It is a trace of it, detached from the interaction that gave it meaning.

The Compass and the Loom names the specific risk for a practice built on weak signals. Recursive training erases the tails of the distribution, which are where weak signals appear. A model trained on aggregated narratives learns what is typical. The atypical narrative, which may be the weak signal, is underweighted, and if the model's outputs feed back into what is collected or attended to, the atypical becomes less visible with each iteration. The anti-computational ontology says the human is not a computer. The computational practice risks making the human more like the computer's model of the human, more typical and less surprising, by feeding the model's expectations back into what humans are asked to notice.

Hutchins offers the right framing, and The Compass and the Loom adopts it. Cognition is distributed across arrangements of people and tools, so the question is how to arrange what passes between them. The anti-computational reason does not forbid tools. It asks how tools and people should be arranged so that the relational character of human cognition is supported and not stripped away. Chapter 41 will take up The Compass and the Loom's answer, the grating, a checkpoint where a human must sense before a machine decides, and its insistence that mesh size matters.

The adversary will point out that SenseMaker was designed precisely to preserve the human in the translation. Self-signification keeps the narrator's interpretation attached to the narrative, so the data is not raw text coded by an analyst but meaning-as-signified by the person who had the experience. The aggregation reveals patterns humans can then explore together. The computation serves the relation and does not replace it. The human/AI hybrid, on this reply, is an extension of distributed cognition, Hutchins's ship with better instruments.

The reply wins half. Self-signification does keep something the coding of narratives by analysts loses: the narrator's own placing of her story. And the aspiration to serve relation with computation is the right one. But self-signification keeps the individual's interpretation, not the relational meaning. A narrator signifies alone, in response to a prompt and a frame she did not choose. The participatory sense-making in which her experience had its meaning is not captured. And the aggregation, however carefully designed, produces patterns that are seen by someone, usually first and most fully by whoever holds the dashboard. Hutchins's navigation team shared its instruments: everyone could see the chart. A distributed cognition in which the chart is seen first by the captain and later, if at all, by the crew is a different arrangement. The adversary's reply is sound in aspiration and incomplete in arrangement. partial

NOT A PROGRAMME IN A HEAD —

BUT A MEANING MADE BETWEEN —

AND WHAT IS MADE BETWEEN IS LOST WHEN IT IS STORED ALONE.

Addition: The Anti-Computational Reason After Large Language Models

The anti-computational reason was formulated, in the tradition from Dreyfus to the enactivists, against symbolic artificial intelligence: programs that manipulate explicit representations according to explicit rules. Its arguments were that intelligence depends on an unrepresentable background, that syntax is not semantics, and that cognition is embodied enactment, not computation over symbols. By the mid-2020s, the most visible form of artificial intelligence was no longer symbolic. Large language models, trained on vast corpora of text by statistical learning, produce fluent narrative, answer questions, summarise documents and imitate styles, without any explicit rules or symbolic representations that a programmer wrote. The anti-computational reason must be reconsidered in their light, because they bear on it in an unexpected way.

What large language models are not

Large language models are not programs in Mayr's sense, or in the classical AI sense. No one wrote the rules by which they produce text. They are not symbol manipulators of the Fodorian kind. Their internal states are high-dimensional vectors, not discrete symbols with assigned meanings. They are not rule-followers in Holland's sense. Their "rules", if the word applies, are distributed across billions of parameters learned from data, not stated in a form any human could read.

In this respect large language models are closer to what the anti-computational tradition praised than to what it attacked. Snowden's account of human cognition is that it is pattern-based, matching the present against fragments of the past and acting on the first fit. That is not a bad description of how a language model generates text. It matches the present context against patterns learned from a vast store of past text and produces the most plausible continuation. Weick's sensemaking is driven by plausibility, not accuracy (see the addition to Chapter 13). Language models are driven by plausibility in a precise statistical sense.

What follows for the anti-computational reason

If language models are pattern-based, context-sensitive and plausibility-driven, and human cognition is too, has the anti-computational reason been refuted? Has computation, in a new form, captured what the reason said it could not?

The disputation's answer is that the reason has been relocated, and that the relocation confirms this book's argument. Language models show that pattern-based, context-sensitive and plausibility-driven output can be produced by computation. What they do not show, and what the anti-computational sciences of Chapter 15 identified as distinctive, is the relational character of human meaning.

Address. A language model's output is not addressed in Bakhtin's sense. It responds to a prompt, but it does not anticipate a particular listener's response from within a relationship, and it does not stand in a history of addressing that listener. Its "addressee" is the statistical average of addressees in its training data.

Stake. The Compass and the Loom's QBist distinction applies: an AI can output probabilities but lacks the stake that makes them its own. A language model's continuations have no consequences for it. It is not changed by the response its output calls out, except through retraining, which is a change made to it, not by it. A human narrator is a reagent. Her story changes her relations and her standing, and what she can say next.

Participatory sense-making. A language model does not make sense with another in an interaction whose dynamics neither controls. Its interaction with a user is turn-taking, each turn conditioned on the history. But it does not have the mutual coordination, the shared stakes or the jointly constituted meaning that De Jaegher and Di Paolo identified as participatory sense-making.

Shared intentionality. Tomasello's we-intentionality, joint goals and commitments that bind the parties, is not something a language model forms. It can produce text that describes joint commitments. It cannot be committed.

Collective acceptance. Searle's status functions exist by collective acceptance. A language model's output can be accepted by humans, and then it can create status functions. A model-generated policy adopted by an organisation becomes a constraint. But the acceptance is human. The model does not accept, and its outputs change constraints only through human uptake.

The relocated reason

The anti-computational reason, relocated, can be stated as follows.

The anti-computational reason, relocated. Computation can now produce pattern-based, context-sensitive, plausibility-driven output resembling human narrative. What it cannot produce is the relational character of human meaning: address within a history of relationship, stake in the consequences, participatory sense-making, shared intentionality and collective acceptance. The human difference that matters is not in the form of cognition, pattern versus rule, but in its relations. The reason, once directed against symbolic AI, now applies with full force to any practice that treats narrative output as equivalent to human meaning.

Consequences for the harvest

The relocation sharpens Thesis VI. If language models can generate plausible narratives, a narrative instrument coupled to them faces a new risk: synthetic narratives. Model-generated stories could be used to augment small samples, fill gaps or simulate populations that were not sensed. They would be plausible and pattern-consistent, and they would have no address, no stake, no participatory sense-making and no relation to any actual person's experience. They would also be drawn from the typical, the statistical centre of the training data, so the tails would be thinned (Chapter 24). A narrative instrument that admitted synthetic narratives would be sensing a population that does not exist, made of the model's expectations of what people say.

The grating (Chapter 41) applies with new force. The checkpoint between narrative and data must now also verify that the narrative is a narrative, told by a person, addressed and at stake. The consent tiers (Chapter 45) must state whether any part of the study's output will be generated or augmented by models. And the reading of patterns must distinguish, where the instrument is coupled to models, between patterns in what people said and patterns in what the model expects people to say.

The irony, stated once

There is an irony here, which should be stated once and then left. The anti-computational tradition argued for decades that human cognition is pattern-based and context-sensitive, not rule-based, and that this is why computers could not replicate it. Computers have now been built that are pattern-based and context-sensitive, and they replicate much of what the tradition said they could not. The tradition was right about the form of human cognition and wrong that the form was what made it human. What makes it human, on the argument of this book, is the relation: the address, the stake, the shared commitment and the collective acceptance. That was the relational account all along, which the sciences of Chapter 13 and the thinkers of Book Five had been describing, and which the possessive grammar of the human cluster obscures.

THE MACHINE NOW MATCHES PATTERNS —

IT DOES NOT ADDRESS, IS NOT AT STAKE, CANNOT BE COMMITTED —

THE HUMAN WAS NEVER IN THE PATTERN; SHE WAS IN THE RELATION.

Chapter 16

The Ethical Reason: Possible and Dangerous

4,482 words, about 19 minutes of reading

Possible and Dangerous

The November post states its ethical reason in one sentence, at the close of the section on humans, animals and intention:

Snowden grants that attention matters, and that much of his own work is about shifting what people attend to. But a theory that sets intention aside, he argues, cannot explain how human systems differ from ecological ones, and that difference is what makes intervention in human systems both possible and dangerous.

Possible and dangerous. Intervention in human systems is possible because humans act with intention: they can be engaged, persuaded, invited and enlisted, and they can choose to change. It is dangerous for the same reason. Humans respond to intervention as intervention. They read it, interpret it, resist it, game it, comply with it outwardly while subverting it inwardly, and tell stories about it that outlast it. An ecosystem can be damaged by intervention, but it cannot be wronged. A human system can.

This chapter examines the ethical reason. It argues that the reason is sound and that it has consequences anthro-complexity has partly drawn, through the Pelagian stance, safe-to-fail probing and disintermediation, and partly not. The undrawn consequences concern consent, the distribution of the risks of probing, the ethics of changing conditions instead of persuading, and the responsibility of the practitioner for the limits of her instruments. The most serious of these is a tension at the heart of anthro-complexity's theory of intervention. The principle that behaviour is the shadow of conditions recommends a mode of intervention that works by bypassing intention, at exactly the point where the ethical reason says intention is what makes intervention in human systems different.

The Ethical Difference

Why does intention make intervention ethically different? Three answers from the philosophical tradition converge.

The Kantian answer. Persons are ends in themselves, never merely means. To intervene in a human system is to act on persons, and acting on persons requires respecting their capacity to set their own ends, their autonomy. An ecosystem has no ends of its own in this sense, so intervention in it raises questions of prudence and of duties to future persons, but not of respect for the ecosystem's autonomy. A human system is made of beings with ends, and intervention that treats them merely as means to the intervener's ends wrongs them.

The Habermasian answer. Human action can be strategic, oriented to success through influencing others' behaviour, or communicative, oriented to reaching understanding through reasons others can accept. Strategic action treats others as objects in the actor's environment, to be influenced. Communicative action treats them as subjects who can be addressed. Both are possible in human systems, and the difference between them is ethical. Communicative action respects others' capacity to evaluate reasons. Strategic action does not, and is legitimate only within limits that communicative action has set.

The responsibility answer. Hans Jonas, in The Imperative of Responsibility (1979), argued that the scale and unpredictability of modern technological action require a new ethics of responsibility for consequences we cannot foresee. Cilliers, in the complexity literature, drew a parallel conclusion. Because our models of complex systems are necessarily incomplete, we are responsible for the limits of our models and for the consequences of acting on them. Responsibility does not require knowing the consequences. It requires acknowledging that we act without knowing them, and owning what follows.

Each answer grounds the ethical reason. Intervention in human systems acts on beings with ends, who can be addressed or merely influenced, and whose responses cannot be foreseen. That is why it is possible (they can be addressed) and dangerous (they can be wronged, and the consequences are unforeseeable).

The Consequences Anthro-Complexity Has Drawn

The Pelagian stance

Anthro-complexity's Pelagian stance (Element 20) is an ethical consequence of the human distinction. If the people in a system have the capacity to heal themselves, intervention that supplies what they "lack" from outside fails to respect that capacity. The Pelagian practitioner works with what is there: she amplifies patterns that people themselves want more of and dampens patterns they want less of. This is Kantian in spirit. It treats the system's members as setters of their own ends, whose capacity for self-direction is to be supported, not overridden.

Safe-to-fail probing

The safe-to-fail probe is an ethical consequence of retrospective coherence and the danger of intervention. Because the consequences of intervention in complex human systems cannot be foreseen, interventions should be small, reversible and designed so that failure does no great harm. Many small, safe probes are ethically preferable to one large, irreversible programme. This is Jonas's responsibility applied to practice: act in ways whose unforeseeable consequences can be contained.

Disintermediation

Disintermediation, removing the expert who stands between people and their own meaning, is an ethical consequence of the human distinction and the critique of the physician. Self-signification lets narrators interpret their own stories instead of having experts interpret for them. This is Habermasian in spirit: it treats narrators as subjects who can speak for themselves, not as objects whose speech must be decoded.

These three consequences are real, and they distinguish anthro-complexity sharply from the systems-change industry Snowden criticises, from consultancies that diagnose and prescribe, and from behavioural interventions designed by experts and imposed on populations. Anthro-complexity's ethics of intervention is among the most thoughtful in organisational practice.

The Consequences Not Yet Drawn

The first undrawn consequence concerns consent. If intervention in human systems acts on persons who can be wronged, the persons acted on have a claim to consent. Consent does not have to be formal, but it has to be real: they know what is being done, by whom, for what purpose and with what likely consequences, and they have a genuine opportunity to decline.

Anthro-complexity's instruments vary in how far they secure consent. Narrators who share stories through SenseMaker are invited to participate and can decline. In that sense participation is consensual. But consent to share a story is not consent to what is done with the story. A narrator who shares an experience in response to a prompt does not usually know which signifiers will be applied, what patterns her story will contribute to, who will see the patterns, what decisions will be made from them, or whether her story will become training data for an automated system. The Compass and the Loom found that the brochure vocabulary of the instruments it examined omitted questions of consent, ownership and benefit. One of the Tidal Loom's principles responds directly: consent to the harvest. No narrative becomes training data without informed consent and a share in the outcome.

The consent question is not marginal. It follows directly from the ethical reason. If humans differ from ecological systems in that they can be wronged by intervention, a practice that gathers their narratives must secure their consent to what is done with them, not merely to the gathering.

Safe for whom?

The second undrawn consequence concerns the distribution of risk. A safe-to-fail probe is safe for someone. The question the ethical reason forces is: for whom? In a human system, probes are designed by some people and experienced by others. The designers, typically managers and consultants, bear the reputational risk of a failed probe, which is usually small, since the probe was designed to be small. The people in whose working lives the probe takes place bear its practical risk: the disruption, the extra work, the uncertainty, and the exposure if the probe goes wrong in a way that affects them. The Compass and the Loom concluded that "distributed" risk postures move exposure down the power gradient, and that risk lives in whoever is left holding it.

Snowden has addressed this in other writing. He insists that probes be designed with those affected and that safe-to-fail means safe for the system, not only for the designer. But the insistence is a principle, and the ethical reason requires a mechanism: a way of tracking whose anxiety a probe creates and where its exposure lands. The Tidal Loom's principle risk follows anxiety proposes one. Track whose anxiety a probe creates and where its exposure lands on the power gradient. Without such a mechanism, "safe-to-fail" is a promise made by those who design probes on behalf of those who experience them.

The ethics of the shadow

The third undrawn consequence is the most serious, because it concerns the central principle of anthro-complexity's theory of intervention: behaviour is the shadow of conditions. If you want different behaviour, change the constraints and conditions people act within, not the behaviour directly.

Consider the principle in the light of the ethical reason and the Habermasian distinction. To change someone's behaviour by changing their conditions is to influence them without addressing them. It is strategic action, not communicative. The person whose conditions are changed is not given reasons to behave differently. She finds herself in a situation in which different behaviour comes naturally, as a shadow changes shape when the light moves. She may not know her conditions were changed, or by whom, or why. She may not have agreed to the change, and she may not even notice that her behaviour has changed.

This is the structure of nudge. Thaler and Sunstein's Nudge (2008) proposed that "choice architects" could improve people's decisions by changing the conditions of choice (defaults, framing, the order of options, the salience of information) without restricting options or changing incentives. The proposal was influential and controversial. Its defenders called it "libertarian paternalism", improving outcomes while preserving freedom of choice. Its critics called it manipulation: influencing people through mechanisms that bypass their rational agency, without their knowledge or consent. The central ethical objection to nudge is that it works because it bypasses deliberation. A nudge that people noticed and reflected on would often lose its force.

Snowden has been a vocal critic of nudge in its standard form, and anthro-complexity's practice differs from it in important respects. Changes of conditions in anthro-complexity are meant to be explored with the people affected, not designed by experts and imposed. Estuarine mapping is done by participants. Probes are co-designed. The aim is to shift constraints so that people's own propensities can flourish, not to steer them towards outcomes chosen by a choice architect. These differences are real.

But the principle that behaviour is the shadow of conditions has the same structure as the principle behind nudge, and the ethical objection to nudge applies to it in principle. Changing conditions to change behaviour, rather than addressing people and giving them reasons, is a mode of influence that bypasses intention. And the November post's ethical reason holds that intention is what makes intervention in human systems different from intervention in ecological ones. An ecosystem can only be influenced by changing its conditions. A human system can be addressed. A theory of intervention whose central principle is to change conditions rather than address persons treats human systems, at the point of intervention, as if they were ecological. It honours intention in its ontology and bypasses it in its method.

Proposition granted: Intention is what makes intervention in human systems both possible and dangerous.

Then it must be granted that intervention which addresses intention (giving reasons, inviting, persuading, asking) differs ethically from intervention which bypasses it (changing conditions so that behaviour changes without deliberation). — Granted. This is the Habermasian distinction.

Then it must be granted that the principle "behaviour is the shadow of conditions; change the conditions, not the behaviour" recommends the second mode of intervention. — Granted, with qualification. The principle recommends changing conditions, and in anthro-complexity's practice conditions are often changed with the participation of those affected. Where they are, the change is communicative: it is discussed, agreed and understood.

Then it must be granted that where conditions are changed without the participation, knowledge or agreement of those affected, the principle recommends a mode of intervention that bypasses the very feature, intention, that the ethical reason says distinguishes human systems. — Granted.

Then the respondent must say whether the principle, as stated, requires participation, knowledge and agreement. — It does not. As stated in the November post, the principle is a claim about the mechanism of behaviour change (behaviour follows conditions), not about the ethics of changing conditions. It is silent on who changes them and with whose knowledge.

Conclusion. The shadow principle needs an ethical supplement that the ethical reason itself supplies: change the conditions, with the knowledge and participation of those whose conditions they are. Without the supplement, the principle is a theory of nudge. With it, it is a theory of shared self-governance.

The supplement is what the Tidal Loom's declination before direction and reading of the reader attempt to provide. Those who change conditions declare what they are doing and why, and expose their own reading of the situation to the same scrutiny as everyone else's. A change of conditions that has been declared, discussed and agreed is communicative action, even though it works through conditions. A change of conditions made silently by those with the power to make it is strategic action, however benign its purpose.

Responsibility for the instruments

The fourth undrawn consequence concerns the practitioner's responsibility for her instruments. Cilliers's ethics of complexity holds that because models of complex systems are necessarily incomplete, the modeller is responsible for what her model leaves out, and for the consequences of acting on an incomplete model. Anthro-complexity's instruments are models in Cilliers's sense. A signifier set models what is salient. An estuarine map models what is costly to change. A Cynefin categorisation models what kind of situation this is. Each is incomplete, each leaves something out, and each, when acted on, has consequences for what it left out.

The ethical reason requires that the practitioner own these limits. Owning them means more than acknowledging in general terms that all models are incomplete. It means stating, for each deployment, what the instrument was designed to see and what it was not, what the designers expected to find and what they hoped not to find, and what was left out and why. The Compass and the Loom's declination table (each signifier set states what its designers expected, hoped for and would prefer not to find) is a mechanism for this. Without such a mechanism, the practitioner's responsibility is acknowledged in principle and evaded in practice.

Von Foerster's Imperative

Chapter 14 introduced Heinz von Foerster's ethical imperative: act always so as to increase the number of choices. It offers a test for interventions in human systems that is close to anthro-complexity's own values. An intervention that increases the choices available to those affected respects their agency. One that decreases them, however benign its purpose, constrains it.

Apply the test to anthro-complexity's practices:

Practice Increases choices? Decreases choices? Net assessment
Safe-to-fail probe (co-designed) Opens new possibilities through experiment — Increases
Safe-to-fail probe (designed above, experienced below) Opens possibilities for designers Imposes disruption on those experiencing it Depends on distribution
Self-signification Narrator chooses how to place her story Signifier set limits the dimensions available Increases, within a frame
Pattern feedback to population Population sees itself and can act — Increases, if feedback is real and timely
Pattern feedback to leadership only Leadership sees and can act Population cannot see what is known about it Decreases for population
Estuarine mapping (participatory) Participants see where they have room to act — Increases
Estuarine map used by leadership to target constraints Leadership's options increase Participants' constraints may be changed without their say Depends on use
Changing conditions (declared, agreed) Conditions changed openly; participants can respond — Increases
Changing conditions (silent) Designers achieve outcomes Those affected do not know their choices were shaped Decreases

The table shows that anthro-complexity's practices are not inherently choice-increasing or choice-decreasing. It depends on how they are deployed: who designs, who sees, who decides, with whose knowledge. That is the central lesson of the ethical reason. The ethics of intervention in human systems is not settled by the design of the instrument. It is settled by the arrangement of the deployment: the relations among the people who design, use and are affected by it.

The Ethical Reason, Relationally

The ethical reason, like the scientific, reflexive and anti-computational reasons before it, turns out to rest on relations. Intervention in human systems is dangerous not because individual humans have intentions that can be frustrated, though they do, but because intervention is a relation between intervener and intervened, and the relation can be one of address or of influence, of shared governance or of steering, of consent or of imposition. The ethics of intervention is the ethics of that relation.

Anthro-complexity's ethical consequences so far drawn (the Pelagian stance, safe-to-fail probing, disintermediation) each improve the relation in one respect. The Pelagian stance respects the system's capacity for self-direction. Safe-to-fail probing limits the harm intervention can do. Disintermediation lets people speak for themselves. The undrawn consequences (consent, the distribution of risk, the ethics of the shadow, responsibility for the instruments) concern the respects in which the relation has not yet been improved. In each, the relation between intervener and intervened is left to the intervener's good intentions rather than built into the practice.

The ethical reason, completed. Intervention in human systems is possible and dangerous because it is a relation between beings who can address one another. The ethics of intervention is therefore the ethics of that relation: whether it is one of address or influence, consent or imposition, shared governance or steering. Anthro-complexity's Pelagian stance, safe-to-fail probing and disintermediation improve the relation. Its shadow principle, as stated, is silent on it, and so recommends a mode of influence that bypasses intention. A completed ethics would require consent to the harvest, tracking of risk along the power gradient, declaration of changes to conditions, and ownership of the instruments' limits.

The adversary will say that the nudge comparison is unfair, and in an important sense it is. Snowden has criticised nudge sharply. His objection is that nudge assumes the choice architect knows what outcomes are good and how to produce them, which in complex systems she does not. Anthro-complexity's changes of conditions are exploratory, not prescriptive. They open possibilities without steering towards predetermined outcomes. And the participation of those affected is standard practice, not an afterthought.

The adversary is right on all three counts, and the chapter accepts them. Anthro-complexity's practice is better than nudge in exactly the ways the adversary describes. But the chapter's claim is about the principle, not the practice. The shadow principle, as stated in the November post, is a claim about how behaviour changes, and it does not say that conditions should be changed only with the participation of those affected. The practice supplies the participation. The principle does not require it. When the principle travels, as principles do, into organisations whose practitioners have not been trained in anthro-complexity's practice, it travels without the participation. "Change the conditions, not the behaviour" is a sentence a chief executive can act on alone. The ethical supplement must be part of the principle, or it will be lost in transit. partial

AN ECOSYSTEM CAN ONLY BE CONDITIONED —

A PERSON CAN BE ADDRESSED —

CHANGE THE CONDITIONS, BUT SAY SO, AND ASK.

Addition: A Casuistry of Probes

Ethics is learned in cases as much as in principles. The scholastic tradition that this disputation honours developed casuistry, the reasoning from paradigm cases to new ones, as a method of practical ethics. It was later abused, and the word acquired a pejorative sense, but Albert Jonsen and Stephen Toulmin's The Abuse of Casuistry (1988) recovered it as a serious method. This addition offers five cases of safe-to-fail probing in human systems, each illustrative and invented, and reasons through the ethical questions each raises. The cases are designed to test the ethical supplement proposed in the chapter: change the conditions, with the knowledge and participation of those whose conditions they are.

Case 1: The canteen experiment

A hospital's leadership, worried about burnout, runs a probe. For one month, the staff canteen on one ward opens an hour earlier and offers free breakfast. Leadership watches whether sickness absence on the ward falls. The staff are told about the breakfast but not that it is an experiment.

Questions. Is the probe safe-to-fail? For the staff, yes: free breakfast harms no one. For leadership, yes: a small cost. Is consent needed? The staff consented to breakfast, not to being studied. Does it matter? The probe works through conditions, not through address. If sickness absence falls, it is because the conditions changed, not because the staff were persuaded.

Reasoning. This is the mildest case, and the ethical supplement asks little of it. But it asks something. The staff should know that the breakfast is a probe and what is being watched. Knowing might change the result, which is the Zeno and demand-characteristics problem. But not knowing means their sickness absence is being monitored for a purpose they were not told, which treats them as objects of study and not as participants in an inquiry. The supplement recommends telling them, and accepting the risk to the result. Verdict: declare the probe.

Case 2: The restructured team

A company's leadership believes that a team's performance is limited by its hierarchical structure. As a probe, it removes the team leader role for three months and tells the team to self-organise. The team leader is reassigned to a special project, and leadership watches what emerges.

Questions. Safe-to-fail for whom? For leadership, the probe is cheap and reversible. For the team leader, it is a public removal from her role, which colleagues will read as demotion, whatever leadership says. For the team, it is a sudden loss of structure in which, Freeman predicts, an informal hierarchy will form, unaccountable and possibly unfair (Chapter 23). The team's least powerful members, those without the status characteristics that make people heard, bear the most risk.

Reasoning. The probe fails the supplement. It changes conditions without the participation of those whose conditions they are, and its risks fall down the gradient. A probe that met the supplement would be designed with the team and the team leader, would state what is being tested and why, would track who gains and loses influence in the informal structure that forms, and would have a dispersal rule saying when and how it ends. Verdict: redesign with participation; track the gradient.

Case 3: The narrative nudge

A government agency uses SenseMaker to collect citizens' stories about a public service. The patterns show that stories of frustration cluster around one step in the process. The agency changes the step's wording on its website, without announcement, and watches whether frustration stories fall in the next round.

Questions. This probe uses a narrative instrument to identify a condition, changes the condition silently, and uses the instrument again to measure the effect. The citizens' stories were the raw material for the change. Did they consent to that use? They shared stories about frustration. Did they want the agency to change its wording, or something more substantial? If frustration falls, is it because the service improved or because the wording made frustration harder to name?

Reasoning. The case has the structure of nudge (Chapter 16), and it raises Hacking's loop (Chapter 33). The change might reduce frustration, or reduce the expression of frustration by changing the vocabulary in which it is expressed. The supplement recommends returning the pattern to citizens before acting (this is what you told us; here is what we propose to change), and asking whether the proposed change addresses the frustration. That converts a silent nudge into a public response. Verdict: return before action; ask whether the change is what was wanted.

Case 4: The waiting leader

A chief executive, advised that her organisation's resistance to a merger is a complex situation, decides to wait, holding back from announcing integration plans for six months so that emergent patterns can form. She tells no one why she is waiting.

Questions. Waiting is not a probe in the usual sense, but it is an intervention (Chapter 21). Her silence will be read: as indecision, as a secret plan, as a sign that the merger is in trouble. Staff will position themselves. Some will leave. The rumour mill will fill the silence. Is this safe-to-fail? The chief executive bears little risk from waiting. Staff bear the anxiety of uncertainty, and the most anxious, those whose jobs are most at risk, bear the most.

Reasoning. The supplement says that waiting, as a gesture, should be declared. The chief executive should say that she is waiting, what she is waiting for, and what would end the wait. That does not abolish uncertainty, but it shares the reason for it, and lets people read the wait as a deliberate choice instead of filling it with fears. Menzies Lyth's insight applies: unacknowledged uncertainty is pushed down the gradient as anxiety. Verdict: declare the wait; track whose anxiety it creates.

Case 5: The receiver's first week

A receiver appointed over a property (see Chapter 46) decides, as a probe, to visit every occupier in her first week before setting any strategy, and to ask each what she should know.

Questions. This is a probe of a different kind. It changes conditions by addressing the people affected, not by altering their environment. Its risks are small: some time, and some exposure of the receiver to hostility. Its benefits are information (the grain) and a shaping of the meaning of her appointment (the gesture). Consent is built in. Occupiers can decline to talk, and those who talk do so knowingly.

Reasoning. The case meets the supplement by design, because it works through address and not only through conditions. It is communicative action in Habermas's sense, an intervention that treats those affected as subjects who can be spoken to. It is the paradigm case against which the others should be measured. Verdict: the model.

What the cases show

Case Works through Risk falls on Supplement met? Remedy
1 Canteen Conditions No one significantly Partly Declare the probe
2 Restructured team Conditions Least powerful; team leader No Design with participation; track gradient; dispersal rule
3 Narrative nudge Conditions, via narrative Citizens' voice No Return before action; ask
4 Waiting leader Withholding Most anxious staff No Declare the wait; track anxiety
5 Receiver's first week Address Receiver (minor) Yes —

The cases show that the ethical supplement is not a burden added to anthro-complexity's practice from outside. It is the difference between probes that treat people as an environment to be conditioned and probes that treat them as participants to be addressed. Case 5 shows that the second kind is not slower or less effective. It is a probe, it generates information, and it changes conditions. It does so through address, which is what the ethical reason says distinguishes human systems from ecological ones.

DECLARE THE PROBE — TRACK THE GRADIENT — RETURN BEFORE ACTION —

DECLARE THE WAIT —

AND WHERE YOU CAN, PROBE BY ASKING.

Chapter 17

The Positional Reason: A School Needs a Border

3,691 words, about 16 minutes of reading

The Reason No One States

The first four reasons for treating human complexity differently are scientific, reflexive, anti-computational and ethical. They are reasons about the world: about what human systems are, how they behave and how they may rightly be acted on. The fifth reason is of a different kind. It concerns not the world but the field of ideas in which anthro-complexity stands. It is rarely stated, because to state it seems to accuse those who hold it of bad faith. This chapter states it anyway. It does so without accusing anyone of bad faith, because the positional reason is a structural reason, one that acts on every body of thought that must make its way in a field, whatever the intentions of those who hold it. Snowden himself supplies the analysis, in his institutional story about IBM, and the chapter extends that analysis to the case he does not apply it to.

The positional reason is simply stated. A school of thought needs a border. To exist as a distinct body of thought, with a name, a literature, a community of practitioners, a training programme and a market, anthro-complexity must be distinguishable from its neighbours. The human distinction supplies the border. Against systems thinking, anthro-complexity is the school that refuses to presume wholes and puts human intention at the centre. Against CAS, it is the school that refuses to model humans as ants. Against the flat ontologies, it is the school that keeps the human distinct. Against CRP, it is the school with instruments and a theory of constraint. Without the human distinction, anthro-complexity would be complexity science applied to organisations, a field with many occupants and no clear owner.

Snowden's Own Analysis

The November post gives the positional reason its first statement, and applies it to systems thinking. Systems thinkers, Snowden says, were always more at home in large organisations than complexity scientists. Systems thinking offers structure (models, boundaries, purposes, a method that can be taught and rolled out), and corporations like that. He was at IBM for seven years before he became an irritant. He offers this not as a judgement but as a hint about which body of thought an institution finds easy to host, and about why complexity has so often been folded into systems thinking instead of being met on its own terms.

This is a sociological argument about intellectual history, and a good one. It says that the fate of ideas is shaped by the institutions that host them, and that institutions select ideas for properties (structure, teachability, roll-out) that are not the same as truth. Complexity, offering less structure than systems thinking, has been reshaped to fit the institutions that took it up, and the reshaping took the form of absorption into systems thinking. The argument applies Element 18, behaviour as the shadow of conditions, to the behaviour of ideas. Ideas take the shape of the institutions that host them.

Element 22, resistance to commodification, gives a second statement. Careful original work is simplified, packaged and sold until what circulates bears little relation to what was written. Lewin's field theory became the three-step model. Edmondson's psychological safety is becoming a commodity. Some authors are abused by the process and others are complicit. The process is structural: it acts on every idea that enters organisational life.

Both statements are correct. Neither is applied to anthro-complexity itself. The rest of this chapter applies them.

Three Sociologies of Borders

Three bodies of sociological work help to analyse the positional reason with more precision than the November post's sketch allows.

Gieryn: boundary-work

Thomas Gieryn's concept of boundary-work (1983, developed in Cultural Boundaries of Science, 1999) describes how scientists demarcate science from non-science. Gieryn showed that the boundary between science and its others (religion, engineering, pseudoscience, politics) is not fixed by any philosophical criterion. It is drawn and redrawn rhetorically, in particular contests over authority, resources and credibility. Scientists describe science differently depending on whom they are distinguishing it from. Against religion, science is empirical and practical. Against engineering, it is theoretical and pure. Against pseudoscience, it is sceptical and methodical. Against politics, it is objective and disinterested. Each description is true enough, and each is chosen for the contest at hand.

Anthro-complexity's border-drawing has the same structure. Against systems thinking, it emphasises its natural-science origins: complexity comes from physics, chemistry and theoretical biology. Against CAS, it emphasises the human: humans are not ants, and models from the natural sciences mislead. Against the flat ontologies, it emphasises human distinctness. Against CRP, it emphasises its theory of constraint and its instruments. Each emphasis is true, and each is chosen for the contest. Taken together, they form a position that is scientific against the humanists and humanist against the scientists, which is a strong position in a field contested on both sides.

Abbott: the system of professions

Andrew Abbott's The System of Professions (1988) analysed professions as competitors for jurisdiction: the right to control a domain of problems. Professions do not have fixed tasks. They claim jurisdiction over problems, and the claims are contested by other professions. Medicine and psychology contest jurisdiction over mental illness. Law and accountancy contest jurisdiction over tax. Abbott showed that jurisdictional claims rest on abstract knowledge, a body of theory that lets a profession redefine problems and claim new ones. The more abstract a profession's knowledge, the more problems it can claim.

Abbott identified three acts through which a profession exercises jurisdiction:

  • diagnosis, which classifies the client's problem into the profession's categories;
  • inference, which reasons from the diagnosis to a course of action;
  • treatment, which acts on the problem according to the inference.

The structure is the structure of medicine, which Abbott took as the paradigm profession. The professional receives a problem, diagnoses it, infers what should be done, and treats it.

Abrahamson: management fashion

Eric Abrahamson's work on management fashion (1991, 1996) analysed the rise and fall of management techniques (quality circles, business process re-engineering, total quality management, agile, design thinking) as fashion cycles. A fashion-setting community of consultants, business schools, gurus and the business press produces techniques. Managers consume them to appear rational and progressive. Techniques spread not because they have been shown to work but because they are fashionable, and they fade when the fashion moves on. The fashion-setters compete to launch the next technique, and the competition rewards novelty, simplicity and the promise of results.

Complexity has had its fashion cycles. In the 1990s "the edge of chaos" and "the new science" were fashionable. In the 2010s it was "VUCA" and "agility". In the 2020s it is "emergence" and "sense-making". Anthro-complexity has stood partly outside these cycles. Snowden has been a fierce critic of fashion in management and of the complexity fashion in particular. But no body of thought that makes its way in the management field can stand wholly outside the fashion system, because the field is organised by it. A school that refuses fashion still has to make its refusal visible, and making a refusal visible is itself a position in the field.

The Positional Reason Applied

With these sociologies in hand, the positional reason can be applied to anthro-complexity directly.

The human distinction as jurisdictional claim

On Abbott's analysis, the human distinction is a jurisdictional claim. It defines a domain of problems (the problems of human systems as distinct from other complex systems) and claims it for anthro-complexity's abstract knowledge. Against CAS, it says computational modellers do not have jurisdiction over human organisations, because humans are not ants. Against systems thinking, it says systems practitioners do not have jurisdiction over complex human situations, because their methods presume wholes. Against CRP, it says that conversation alone cannot serve organisations at scale, so the jurisdiction needs instruments. The human distinction is the abstract knowledge that supports the claim. Human systems are different in kind, and anthro-complexity's theory of the human is what lets one deal with that difference.

None of this makes the distinction false. Jurisdictional claims are often true, and medicine's claim over infectious disease is a true one. But it means the distinction does two jobs at once. It describes the world, and it marks a territory. The two jobs can pull against each other, because a description of the world that is good for marking territory is not necessarily the best description of the world.

The physician's structure returns

Abbott's analysis has a sharper consequence. The acts through which a profession exercises jurisdiction, diagnosis, inference and treatment, are the acts of the physician. A profession exercising jurisdiction over human systems receives a client's problem, diagnoses it in its categories, infers what should be done, and acts on the inference.

Set this beside Cynefin's practice. The client presents a situation. The practitioner, or the client with the practitioner's help, categorises it as clear, complicated, complex or chaotic. That is diagnosis. The category determines the decision model: probe-sense-respond for complex, sense-analyse-respond for complicated, and so on. That is inference. The decision model is carried out through the profession's instruments (probes, SenseMaker studies, estuarine maps). That is treatment. The structure is Abbott's, and it is the physician's.

This is the structure Snowden criticises in Ing. The physician takes the pulse; the body does not. His answer is distributed sensing, in which the body takes its own pulse. But the positional reason, analysed through Abbott, shows why distributed sensing has difficulty dislodging the physician. The profession's jurisdiction depends on the physician's structure. A profession whose clients diagnosed, inferred and treated themselves would have no jurisdiction. The client pays for the diagnosis, the inference and the treatment, or for the instruments through which they are performed. Distributed sensing can distribute the act of sensing: the narrators self-signify and the participants place the constraints. It has difficulty distributing the jurisdiction, because the jurisdiction is what the profession is.

Proposition granted: The physician takes the pulse; the body does not; a practice that keeps the physician's position is Augustinian.

Then it must be granted that a profession exercising jurisdiction over human systems through diagnosis, inference and treatment keeps the physician's position, to the extent that it, and not the system, holds the categories of diagnosis, the rules of inference and the instruments of treatment. — Granted.

Then it must be granted that Cynefin's practice (categorise, choose the decision model, deploy the instruments) has the structure of diagnosis, inference and treatment. — Granted, with the qualification that categorisation is often done collectively, by the people in the system.

Then it must be granted that collective categorisation distributes the act of diagnosis without distributing the categories. The domains, the decision models and the instruments are the profession's, and the system uses them under licence. — Granted.

Then the respondent must say whether a practice in which the system diagnoses itself using the profession's categories, infers using the profession's rules and treats itself using the profession's instruments is Pelagian or Augustinian. — Both. It is Pelagian in that the system acts for itself, and Augustinian in that what it acts with is supplied from outside. This is the formula of Chapter 5's flag: Pelagian in content, Augustinian in form.

Pelagian content, Augustinian form

The formula repays expansion. The theological analogy Snowden uses is exact, and its exactness is instructive.

Pelagius held that humans have the natural capacity to choose the good. Grace assists but does not replace it. Augustine held that the capacity was so damaged by the Fall that grace, coming from outside, was necessary for any good willing. Snowden uses the terms analogically: an Augustinian intervention supplies what the system lacks, and a Pelagian intervention helps the system use what it has.

But in the historical controversy there was a further question that bears directly on anthro-complexity's position. Even Pelagius did not deny that humans needed something from outside: the law, the teaching of Christ, the example of the saints. These were, for him, forms of grace in a broad sense. They did not replace human capacity, but they informed and directed it. The question between Pelagius and Augustine was not whether anything came from outside, but whether what came from outside was necessary for the good, or only helpful.

Anthro-complexity's instruments are in exactly this position. They do not replace the system's capacity for self-sensing. They inform and direct it, and the question is whether they are necessary or only helpful. Snowden's Pelagian answer is that they are helpful. The system could sense itself without them, and the instruments make its sensing more effective. The positional reason, analysed through Abbott, suggests a different answer, or at least a pressure towards one. A profession whose instruments were merely helpful would have weak jurisdiction. Its clients could do without it. Jurisdiction is strong when the profession's knowledge is necessary: when, without the profession's categories and instruments, the system cannot sense itself adequately. The positional reason pushes anthro-complexity towards presenting its instruments as necessary, which is the Augustinian position, while its theory holds that they are only helpful, which is the Pelagian one.

The Drift as Positional Effect

Chapter 6 traced a drift in the Cynefin lineage, from ontology through epistemology, instruments and energetics to products, and noted that each step answers a question the previous step raised and moves closer to the client's question. The positional reason explains why the drift has that direction. A body of thought that must make its way in the management field is pulled towards the client's question, because the client is the one who sustains it. Each step of the drift is a step towards greater jurisdictional strength. The products of Stage Six are more clearly necessary than the knowledge-management practices of Stage One. They promise solutions to recognisable problems ("adrift without direction", strategy and risk unconnected) that the client cannot solve unaided.

The drift is not a fall from grace. Every body of thought that enters practice undergoes something like it, and the alternative, remaining at Stage One or refusing instruments as CRP largely has, carries its own costs: practical marginality, the inability to work at scale, dependence on the practitioner's personal skill. Snowden chose the drift with open eyes, and the choice has let anthro-complexity reach organisations, governments and populations that CRP has not.

But the drift has a consequence for the human distinction, and it is the one this book is concerned with. At each step the human moves further from the centre. At Stage One the human is the knower and narrator. At Stage Six the human is sensor and substrate. The positional reason explains why. The client's question is always asked from the position of someone who acts on the system, and a body of thought pulled towards the client's question is pulled towards that position. The human the instruments serve is the client. The human the instruments sense is the population. The theory of the human, restated at Stage Five, describes the population as narrators, intenders and abstractors. The instruments, built at Stages Three, Four and Six, serve the client as reader, steerer and decider.

The Immune Response and Its Limits

Element 22, resistance to commodification, is anthro-complexity's immune response to the positional pressure. Snowden's critique of Lewin's reconstruction, of Edmondson's commodification, of the systems-change industry and of management fashion are all forms of it. They are vigilance against the process by which careful thought is packaged and sold.

The immune response is real and valuable. Anthro-complexity has resisted simplification more successfully than most bodies of thought in the management field. Cynefin has been vulgarised, as every successful framework is, but Snowden has spent twenty years correcting the vulgarisations, revising the framework (the renaming of domains, the introduction of the aporetic and liminal, the insistence on dynamics), and refusing the two-by-two simplification consultants keep trying to impose on it.

But an immune response has limits. It can attack what it recognises as foreign. It has difficulty recognising what it has come to treat as self. Anthro-complexity's critique of commodification is directed outward: at Lewin's reconstructors, at Edmondson's packagers, at the systems-change industry. Applied inward, the same critique would ask whether SwarmCompass and WRAS are careful thought packaged and sold, whether the accreditation programmes simplify what they teach, and whether "anthro-complexity" itself is becoming a brand. Snowden's distinction between authors who are abused and authors who are complicit is the right instrument for the inward question. It is the hardest one to apply, because no one sees the form they are standing in.

The adversary will say this chapter is ad hominem in sociological dress. To analyse a body of thought by its position in a field is to change the subject from whether its claims are true to why its holders might hold them. Snowden's human distinction is either well grounded or not, and Chapters 13 to 16 have argued that it is. Its usefulness as a border is irrelevant to its truth. Bourdieu, Gieryn and Abbott can explain the career of any idea, true or false, and explaining a career is not refuting a claim.

The adversary is right that position does not refute truth, and the chapter does not claim it does. The human distinction is true, as Chapter 13 concluded. But the positional reason is not offered as a refutation. It is offered as an explanation of a gap: the gap between anthro-complexity's theory of the human and its instruments' placement of the human, identified in Chapter 6 and documented in Chapters 13 to 16. The positional reason explains why the gap opened (the drift towards the client's question) and why it is hard to close (jurisdiction depends on the physician's structure). It does not say the theory is false. It says the theory and the instruments are pulled in different directions by different forces, and that a practitioner who wants to close the gap needs to understand the forces. Snowden's own institutional story about IBM is an analysis of exactly this kind, applied to systems thinking. Applying it to anthro-complexity is not ad hominem. It is the same argument, made without exemption. partial

The Border the Human Distinction Draws

One more feature of the positional reason bears on the argument of Book Four. A border marks an inside and an outside. The human distinction, as a border, marks anthro-complexity's inside (the study of human complexity) against its outside (complexity in general, systems thinking, CAS). It also does something else. It draws a border within the systems anthro-complexity studies, between the humans in them and everything else: the river and the planners, the narrators and the constraints, the actants who narrate and the actants who do not.

Within human systems there is a further border that the human distinction, as stated, does not draw. It is the border between the humans who study the system and the humans who are studied, between the planners and the planned-for, the convenors and the convened, the readers and the read. The asymmetry principle says that only humans can change the story they tell about the other. But within a human system, some humans are better placed than others to change the stories, because they hold the instruments, convene the sessions, design the signifiers and receive the patterns. The human distinction draws its border between humans and rivers. It does not draw the border between humans and humans, which is where, in a human system, the asymmetry of story-changing power actually lies.

That border is the subject of Book Four. Each of the eight theses that follow is, in one way or another, about it: the border between the humans who describe and the humans described, between the position of the reader and the position of the read. Anthro-complexity's theory erases this border in principle, since every human is a narrator, an intender and an abstractor. Its instruments redraw it in practice, because someone designs, someone convenes and someone receives. The positional reason explains why. A body of thought that must make its way in a field must have clients, and clients occupy the position of the reader.

The Five Reasons Reviewed

With the positional reason in hand, the five reasons can be reviewed together before Chapter 18 asks what jobs they do.

Reason Claim Verdict Ground, when followed through
Scientific The science shows humans differ in kind from other complex-system agents holds Relational: self-categorisation, shared intentionality, collective acceptance
Reflexive Human systems are changed by descriptions of themselves holds Relational: descriptive reflexivity requires shared symbols and uptake
Anti-computational Human cognition is not a specifiable program holds (stated with care) Relational: participatory sense-making, distributed cognition, Vygotsky
Ethical Intervention in human systems is possible and dangerous holds Relational: the ethics of intervention is the ethics of the intervener–intervened relation
Positional A school needs a border; the human distinction supplies it holds as explanation Draws the border between humans and non-humans, not between describing and described humans

All five reasons hold. Four of them, followed through, ground the human distinction relationally. The fifth explains why the distinction is stated as it is: in a grammar of possession that marks anthro-complexity's border with its neighbours, and not in a relational grammar that would mark the border between describers and described within the systems it studies.

A SCHOOL NEEDS A BORDER —

THE BORDER IT DRAWS IS BETWEEN HUMANS AND RIVERS —

THE BORDER IT NEEDS IS BETWEEN READERS AND READ.

Addition: Cross-Lights From Book Nine

The positional reason, acknowledged. Snowden's trialectic says that most consultancies default to the deficit mode because the market rewards it and clients expect it, and that the encounter mode is often declined as professionally risky. That is the argument of this chapter from the practice's own author.

Classification struggles. The Zen Garden study applies Bourdieu's idea that groups contest how the world is sorted. Snowden's refusal to be a branch of the systems tree is a justified move in such a contest. So is his placing of the Stacey matrix near systems thinking, and so is this book's sorting of its thinkers.

Jaeggi's test. A commercial form that keeps a practice from doing what its own norm requires is, in her terms, a blockage to learning that goodwill cannot remove. Chapter 54 lists it as one of four.

Chapter 18

The Three Jobs of a Distinction

3,488 words, about 15 minutes of reading

Draw a Distinction

George Spencer-Brown's Laws of Form (1969) opens with an injunction: Draw a distinction. The book builds a calculus from that one act. To draw a distinction is to split a space into two sides, a marked side and an unmarked side, and to indicate one of them. Everything else follows. Spencer-Brown's calculus became, through Heinz von Foerster and above all Niklas Luhmann, a foundation for second-order systems theory. Luhmann took from it the insight that every observation is the drawing of a distinction, and that the distinction itself cannot be observed while it is being used. To see with a distinction is to be blind to it. The eye does not see itself seeing.

Spencer-Brown also introduced, late in the book, an operation he called re-entry: the re-entry of a distinction into the space it distinguishes. A distinction between inside and outside re-enters when the inside is itself divided by the same distinction, so that the inside contains its own inside and outside. Re-entry is the formal structure of self-reference. Luhmann made it central: social systems distinguish themselves from their environment, and then re-enter the distinction, representing the system/environment distinction within the system, as the system's own description of itself.

This chapter applies Spencer-Brown's injunction to anthro-complexity's human distinction. The distinction between human and non-human complexity is a drawn distinction. It splits complex systems into two sides and indicates the human side. Like every distinction, it does work, and the chapter argues that it does three jobs at once. The jobs pull against one another, and the pull explains the gap between theory and instruments that the previous chapters have documented. The chapter ends with re-entry. The human/non-human distinction, re-entering the human side, divides humans into those who describe and those described, and that re-entered distinction is what Book Four examines.

Three Jobs

Job one: ontological

The first job of the human distinction is to describe the world. It says that human complex systems differ in kind from other complex systems. This is a claim about what there is, and the previous chapters have tested it. The scientific reason (Chapter 13) found that the best science confirms it. The reflexive reason (Chapter 14) found its ground in descriptive reflexivity. The anti-computational reason (Chapter 15) found that human cognition is not a specifiable program. The ethical reason (Chapter 16) found that intervention in human systems is a relation between beings who can address one another. In its ontological job, the human distinction is well grounded, and its ground is relational.

Job two: methodological

The second job is to justify a method. If human systems differ from other complex systems, they need different methods of inquiry and intervention. Anthro-complexity's methods are narrative-based (SenseMaker), participatory (estuarine mapping), probe-based (safe-to-fail experiments) and sense-making-oriented (Cynefin). The human distinction justifies each. Because humans narrate, collect narratives. Because humans interpret their own experience, let them signify it. Because human systems are unpredictable, probe instead of plan. Because humans make sense of situations, give them frameworks for doing so.

The methodological job is done well in one respect. It rules out methods that ignore the human distinction: agent-based simulation used as if humans were ants, survey instruments that reduce meaning to ratings, expert diagnosis that overrides the people's own sense-making. But it pulls the ontological job in a particular direction. Narrative methods collect narratives from individuals. Self-signification is done by individuals. To justify individual-level methods, the human distinction is most naturally stated at the individual level: individuals narrate, individuals interpret, individuals hold identities and intentions. The methodological job pulls the ontology towards the grammar of possession.

Job three: institutional

The third job, analysed in Chapter 17, is to mark a border, which in Abbott's terms is a jurisdictional claim. The human distinction marks anthro-complexity's territory against CAS, systems thinking, the flat ontologies and CRP. It gives the school a name, an identity and a claim over a domain of problems.

The institutional job also pulls the ontology. A border must be sharp to function as a border. A distinction of degree makes a poor border, since if humans differ from ants only in degree, the CAS modellers can claim jurisdiction by enriching their agents. The institutional job therefore pulls the ontology towards sharp exceptionalism: humans are not ants, full stop. The science (Chapter 13) and the comparative biology (Chapter 9) both suggest a more graded picture. First-order reflexivity is found in fish, and joint attention has precursors in apes. Only descriptive, second-order reflexivity is clearly a difference of kind.

How the Jobs Pull

The three jobs interact, and the interactions explain much of what the previous chapters found.

Job two pulls job one towards possession

Narrative instruments work at the level of individual narrators. Each narrator shares a story and signifies it, and the stories are aggregated. For these instruments to be justified by the human distinction, the distinction must be one that individuals carry: individuals have the narratives, the identities and the interpretations that the instruments gather. A relational distinction, one in which identities are produced by comparative context, intentions are shared and meanings are made in participatory sense-making, would justify different instruments: instruments that gather relations, not individuals. The methodological job, given the instruments that exist, favours the possessive statement of the distinction.

Job three pulls job one towards sharp exceptionalism

A school needs a sharp border, so the institutional job favours stating the distinction as absolute (humans are not ants) rather than graded (humans differ from ants in degree on some dimensions and in kind on one). The sharp statement is rhetorically effective. It is also slightly false, and the falseness is exploitable. A CAS theorist who knows the comparative literature can point to the cleaner wrasse and the El Farol model and say the distinction is one of degree, and the possessive statement of the distinction has no reply. The relational statement does have one. Descriptive reflexivity, the uptake of shared descriptions that change constraints, is a difference of kind no enrichment of agents can reach. But the relational statement is not the one the institutional job favours, because it is less quotable and its border runs in an unexpected place.

Job one, taken seriously, undermines jobs two and three

This is the most important interaction. If the ontological job is done properly, if the human distinction is grounded in descriptive reflexivity, then two consequences follow that pull against the other jobs.

Against job two. If human systems are descriptively reflexive, every method of inquiry is an intervention. A narrative instrument that collects stories is not merely gathering what is there. It is a description entering the system: its prompts make identities salient, its signifiers become categories, its patterns are fed back and change subsequent narration. The instruments must then be redesigned as actors, accounting for their own effects, and not as sensors that read what is there. The methodological job, as currently done, justifies instruments as sensors. The ontological job, done properly, requires instruments that are designed as actors.

Against job three. If human systems are descriptively reflexive, the human distinction itself is a description that enters the field. Anthro-complexity's border with its neighbours is not a fact about the world that the school has discovered. It is a description of the field that changes the field. It shapes how CAS theorists, systems thinkers and CRP practitioners position themselves, how clients understand what they are buying, and how practitioners understand what they are doing. The border is performative. A school whose ontology says that its own descriptions are performative cannot treat its border as a neutral fact. It must treat the border as a move in the field. Jurisdiction, as Abbott analyses it, requires the opposite. It requires that the profession's categories be taken as accurate descriptions of the problems they classify, not as moves in a contest.

The jobs displayed

Job What it needs from the distinction How it pulls the distinction What job one, done properly, does to it
Ontological Truth — —
Methodological Justification for narrative, participatory, probe-based methods Towards possession (individuals carry what instruments gather) Requires instruments designed as actors, not sensors
Institutional A sharp border; jurisdiction Towards sharp exceptionalism (humans are not ants) Makes the border a performative move, not a neutral fact

Theses 21 and 22

Thesis 21. The distinction between human and other complexity does three jobs at once (ontological, methodological, institutional), and taking the first seriously undermines the other two. holds

The ontological job, done properly, grounds the distinction in descriptive reflexivity. Descriptive reflexivity makes every instrument an actor, which undermines the methodological justification of instruments as sensors. It also makes every border a performative move, which undermines the institutional treatment of the border as fact. The methodological and institutional jobs, in turn, pull the ontological statement towards possession and sharp exceptionalism, which is where the CAS theorist can attack it.

Thesis 22. The drift from ontology to epistemology, energetics, decision and finally product is the path along which reflexivity is lost. holds

At Stage One, reflexivity is present in practice. The narrator and the listener are in conversation, and the listener's presence shapes the telling. At each later stage the reader moves further from the read: the leader categorises, the dashboard aggregates, the map represents terrain, and the product steers. The theory restated at Stage Five recovers reflexivity in principle, as any framework applied to a human system becomes an actor within it. But the instruments carry the earlier stages' positions, and those positions are not reflexive.

The adversary will say that this analysis proves too much. Every school's central distinction does several jobs. Darwin's distinction between natural and artificial selection described the world, justified a method (comparative observation of variation) and marked a border (against special creation and Lamarckism). The jobs pulled against each other, and Darwinism survived. A distinction that does several jobs is not compromised by doing them. To demand that the human distinction do only its ontological job is to demand that anthro-complexity stop being a practice.

The adversary wins half. It is right that every working distinction does several jobs, and right that doing them does not by itself compromise the distinction. The demand is not that the human distinction stop doing its methodological and institutional jobs. It is that the methodological and institutional jobs be done in a way consistent with the ontological job. Darwin's methodological job was consistent with his ontology. Comparative observation of variation is what natural selection, as an ontology, calls for. Anthro-complexity's methodological job is partly inconsistent with its ontology. Descriptive reflexivity calls for instruments designed as actors, and the instruments are designed as sensors. The inconsistency is not inherent in having three jobs. It is a feature of how these particular jobs are currently done, and it can be corrected. Book Eight proposes how. partial

Re-Entry

The chapter returns to Spencer-Brown. The human distinction draws a line between human and non-human complexity and indicates the human side. Luhmann's lesson is that the distinction cannot be seen while it is being used. Anthro-complexity, looking at complex systems through the human/non-human distinction, sees humans and non-humans, narrators and constraints, planners and rivers. What it does not see, while it uses the distinction, is the distinction itself: who draws it, from where, for what purposes.

Spencer-Brown's re-entry shows what happens when the distinction is turned on itself. Re-enter the human/non-human distinction into the human side, and the human side is divided by the same distinction. Among humans, some are positioned as describers, narrating, interpreting, changing stories, acting with intention on the system. Others are positioned as described: narrated about, interpreted, their stories gathered, their constraints mapped. The asymmetry principle (only one party can change the story it tells about the other) re-enters, too. Among humans, the describers can change the story they tell about the described more easily than the described can change the story told about them. The planners argue about the river. Within the city, the planners also argue about the residents, and the residents' capacity to change the planners' story about them is not equal to the planners' capacity to change theirs.

The re-entered distinction is the subject of the eight theses of Book Four. Each examines one aspect of it:

  • Thesis I, the possessive fallacy. The re-entered distinction is hidden by stating the human marks as possessions, since possessions belong equally to describers and described.
  • Thesis II, the asymmetry principle applied to itself. The re-entered asymmetry applies to anthro-complexity's own descriptions.
  • Thesis III, humans are not estuaries. The root metaphor of the estuary places the describer outside a non-reflexive medium and so erases the re-entry.
  • Thesis IV, the under-socialised constraint. Constitutive constraints are where describers and described are made.
  • Thesis V, the missing term is power. The re-entered asymmetry is a power asymmetry.
  • Thesis VI, the computational practice. Describers' instruments turn the described into data.
  • Thesis VII, the triad as simplex. The instrument's geometry decides what the described can say.
  • Thesis VIII, the degradation of instruments. The described learn the describers' instruments and change.

Book Three has argued that anthro-complexity is right to treat human complexity differently, and that it is right for reasons that, followed through, are relational. Book Four argues that the relational ground, once stated, re-enters, and that the re-entry divides human systems along a line anthro-complexity's theory does not draw and its instruments do not see.

DRAW A DISTINCTION: HUMAN AND RIVER —

LET IT RE-ENTER: DESCRIBER AND DESCRIBED —

THE SECOND LINE IS WHERE THE WORK BEGINS.

Addition: Three Jobs in Other Schools

Chapter 18 argued that anthro-complexity's human distinction does three jobs, ontological, methodological and institutional, and that taking the first seriously pulls against the other two. The adversary replied that every working distinction does several jobs, citing Darwin. The reply deserves a fuller answer. This addition examines three other schools whose central distinctions did several jobs, and asks how each handled the tension. The comparison shows that the tension is common, that it can be managed well or badly, and that the ways it is managed well suggest what a reflexive anthro-complexity might do.

Darwin: natural and artificial selection

Darwin's On the Origin of Species (1859), from which Snowden takes the epigraph of the November post, opens not with nature but with domestication: the breeding of pigeons, dogs and cattle by human selection. Darwin's central distinction, between artificial selection (by breeders) and natural selection (by the struggle for existence), did three jobs.

  • Ontological. It described a mechanism that, Darwin argued, operates in nature as breeders' choices operate in the farmyard.
  • Methodological. It justified a method: comparative observation of variation, especially under domestication, where variation and selection could be seen.
  • Institutional. It marked a border against special creation and Lamarckian inheritance, giving the new theory a territory.

How did Darwin handle the tension? The methodological job, observing artificial selection to understand natural selection, was consistent with the ontology, because the ontology said the two were the same kind of process. The institutional job, the border against creation, was sharp, and Darwin kept it sharp, but he did not overstate the ontology to sharpen it. He acknowledged difficulties, gaps and objections at length, in a chapter titled "Difficulties on Theory". The ontology was allowed to be uncertain where it was uncertain, and the border did not depend on certainty.

Lesson. A distinction can do all three jobs well when the method follows from the ontology and the border does not require the ontology to be overstated.

Freud: conscious and unconscious

Freud's distinction between conscious and unconscious mental processes did three jobs.

  • Ontological. It described a domain of mental life inaccessible to introspection, which shaped behaviour and symptoms.
  • Methodological. It justified a method, free association and the interpretation of dreams, slips and symptoms, by which the unconscious could be inferred.
  • Institutional. It founded a profession, psychoanalysis, with its training institutes, its lineage of analysts and its border against academic psychology and psychiatry.

How did Freud handle the tension? Less well than Darwin, on most accounts. The institutional job, the founding and defence of a profession, pulled the ontology towards orthodoxy. Dissenters (Adler, Jung, Rank) were expelled, and the theory's claims were defended against revision. The methodological job, interpretation by the analyst, placed the analyst in a position of interpretive authority over the patient, which the ontology, the unconscious as inaccessible to the patient's own introspection, seemed to justify. Later critics, from within psychoanalysis and without, argued that the analyst's authority had been overstated and the patient's own understanding undervalued. The relational turn in psychoanalysis, from Sullivan through Mitchell, moved meaning from the analyst's interpretation to the relationship between analyst and patient.

Lesson. When the institutional job pulls the ontology towards orthodoxy, and the methodological job places the practitioner in interpretive authority, the distinction's ontological truth is distorted. The relational turn that followed is a correction of exactly the kind this disputation proposes for anthro-complexity.

Kuhn: normal and revolutionary science

Thomas Kuhn's distinction between normal science (puzzle-solving within a paradigm) and revolutionary science (the overthrow of one paradigm by another), in The Structure of Scientific Revolutions (1962), did three jobs.

  • Ontological. It described the history of science as alternating between stability and rupture.
  • Methodological. It justified a method for history of science: attend to the paradigm, the shared exemplars, the anomalies and the crises.
  • Institutional. It founded, unintentionally, a field (science studies) and gave a vocabulary ("paradigm shift") to management consultants, who used it to sell change programmes.

How did Kuhn handle the tension? He was troubled by it. The institutional uptake of "paradigm shift" in management and popular culture distorted the ontology. Every change programme became a paradigm shift, and the concept lost its precision. Kuhn spent much of his later career clarifying and qualifying the concept, and was dismayed by its spread. His experience is a case of what Snowden calls commodification (Element 22): a careful piece of work simplified, packaged and sold until what circulates bears little relation to what was written.

Lesson. When a distinction's vocabulary is taken up by institutions outside its home field, its ontology is at the mercy of its users. The author can clarify, but cannot control.

The comparison

School Distinction How the jobs interacted Lesson for anthro-complexity
Darwin Natural / artificial selection Method followed ontology; border did not require overstatement Let the method follow the ontology; do not sharpen the border by overstating
Freud Conscious / unconscious Institution pulled towards orthodoxy; method placed analyst in authority; relational turn corrected The relational turn is a known correction of interpretive authority
Kuhn Normal / revolutionary Vocabulary taken up outside, and ontology distorted Commodification of the vocabulary is outside the author's control
Anthro-complexity Human / non-human complexity Method pulls towards possession; border pulls towards sharp exceptionalism; ontology, done properly, makes instruments actors and the border performative Follow Darwin: let the instruments follow the relational ontology. Follow the relational turn: move interpretation into the relation. Watch for Kuhn's fate

What a well-managed distinction would look like

The comparison suggests three conditions under which a reflexive anthro-complexity could do all three jobs well.

The method follows the ontology. If the ontology is relational, with descriptive reflexivity as the human difference of kind, the instruments should be designed as relational actors, not as individual sensors. That is the Darwinian condition, and Book Eight's protocols are an attempt to meet it.

The border does not require overstatement. Anthro-complexity's border with CAS should be drawn at descriptive reflexivity, where the difference of kind is secure, and not at the three marks, where the CAS theorist can argue for degree. A border drawn where the ontology is strongest needs no overstatement to hold.

The practitioner's authority is relational. Following psychoanalysis's relational turn, the practitioner's interpretive authority should be moved into the relationship between practitioner and population. Patterns become proposals, readings become contestable and the interpretant belongs to the community (Chapter 32).

Under these conditions the three jobs would no longer pull against one another. The ontology would justify the method, the method would embody the ontology, and the border would be drawn where both are strongest.

DARWIN LET THE METHOD FOLLOW THE ONTOLOGY —

FREUD'S HEIRS MOVED MEANING INTO THE RELATION —

KUHN WATCHED HIS VOCABULARY SOLD. CHOOSE YOUR PRECEDENT.

Addition: Cross-Lights From Book Nine

The method named. This chapter argued that the human distinction does three jobs and that the first undermines the other two. Rahel Jaeggi's account of immanent criticism, set out in Chapter 49, says what kind of argument that is. It measures a practice by a norm the practice cannot do without, and shows the norm's realisation working against it. Chapter 50 applies the method in full to Snowden's formative encounter.

A second-order problem. Jaeggi distinguishes difficulties that strike a practice from outside from problems it sets itself by its own claims. The three jobs describe a problem of the second kind. Nothing external has gone wrong. Anthro-complexity meets a consequence of its own best insight.

Chapter 19

Thesis I: The Possessive Fallacy

3,883 words, about 17 minutes of reading

Thesis I: the possessive fallacy. Anthro-complexity places human distinctiveness in properties persons have (identities, intentions, abstractions) rather than in the relational processes through which persons, identities and intentions are formed. Having refused to presume wholes, it presumes parts. This is Elias's closed person, the closed person, dressed in the language of complexity. (Theses 4 and 5 of the twenty-four.)

The Grammar of the Claim

Begin with the sentences. The November post states the human distinction as follows:

In paraphrase: human systems are complex, but they differ from other complex systems, living ones included, because humans hold multiple identities at once, act with intentionality, and create abstractions and shared narratives that change the constraints they operate within.

Look at the verbs: hold, act with, create. Each has humans as its subject and a capacity or product as its object. Humans hold identities, as one holds a passport or a share. Humans act with intentionality, as one acts with a tool or a skill. Humans create abstractions, as a maker creates an artefact. The grammar is the grammar of possession and production. The human is a subject who has capacities and makes things, and the distinctiveness of human systems follows from the capacities their members have and the things they make.

The grammar continues in the section on entanglement:

The people remain distinct, even when what they know together does not.

Here the possessive structure is explicit. Knowledge can be entangled, existing only in the interaction of people, artefacts and context. People cannot. They remain, distinct, before and after and during the interaction, bearers of the identities, intentions and abstractions that make them human. The entangled knowledge is between them. They are not between themselves.

The Compass and the Loom develops Snowden's marks into the 3Is and 3Ss, and keeps the grammar exactly: the 3Is (intelligence, identity, intention) are what agents have; the 3Ss (sapience, salience, sentience) are what they do with it. Have and do with: the grammar of property and use.

Grammar is not ontology. But a grammar used consistently at the moment a body of thought defines itself is evidence of the ontology it presupposes, and this one is consistent. The person is a bearer of properties, and the properties explain the system's distinctiveness. This chapter argues that the ontology so presupposed is the one anthro-complexity refuses everywhere else.

The Refusal Anthro-Complexity Makes Everywhere Else

Element 2, the most philosophically distinctive element of anthro-complexity's ontology, is the refusal to presume wholes or parts. Systems thinking presumes wholes and infers parts. Reductionism presumes parts and derives wholes. Anthro-complexity presumes neither. The whole is continually produced by interactions that it, in turn, modulates, and coherence has to be found in practice, at a level that may shift. Hence the method: find the level of coherence before decomposing or recombining.

Apply Element 2 to persons. If neither wholes nor parts are presumed, persons, who are parts of human systems, are not presumed either. A person is a coherence, found at a level, produced by interactions that she in turn modulates. Her identities are not held before the interactions in which they are called out. Her intentions are not formed before the interactions in which they take shape. Her abstractions do not exist before the interactions in which they are shared. The person is as much an effect of interaction as the team, the organisation or the market, and as much a cause of it.

This is not an exotic conclusion. It is what Element 2 says when it is not exempted at the level of the individual. And it is exactly what the human cluster, in its possessive grammar, exempts. The human cluster presumes the person, bearer of identities, intentions and abstractions, as the part from which human systems are made. Element 2 says no part is to be presumed. The human cluster, at the level of the person, breaks the principle the ontological cluster establishes at every other level.

Proposition granted: Anthro-complexity presumes neither wholes nor parts. Coherence is found in practice, at a level that may shift.

Then it must be granted that teams, organisations and markets are coherences found in practice, not wholes presumed. — Granted.

Then it must be granted that the individuals who make up teams, organisations and markets are not parts presumed. — Granted, as an instance of the proposition.

Then it must be granted that individuals are coherences found in practice, at a level that may shift. — Granted. A person's coherence shifts with context: she is one thing at work, another at home, a third in a crisis.

Then it must be granted that a person's identities, intentions and abstractions are not properties she holds before the interactions in which they appear, but features of the coherence found in those interactions. — Doubted by the November post, which says the people remain distinct. Granted by Element 2.

Then the respondent must choose. Either Element 2 applies to persons, and the human cluster must be restated relationally, or persons are an exception to Element 2, and the exception must be justified. — The November post does not justify the exception. It asserts it ("the people remain distinct") against Barad, whose dissolution of relata is the other extreme. But refusing Barad does not require presuming persons. It requires only that persons be found, not presumed, as distinct coherences produced in relation. That is the position of Mead and Elias, and it honours Element 2.

Elias and the Closed Person

Norbert Elias, in What Is Sociology? (1970) and the essays collected as The Society of Individuals (1987), mounted a sustained critique of what he called the closed person. The closed person is the image of the individual as a sealed container. There is an inner self, the true person, with her thoughts, feelings, intentions and identity, and an outer world, including other people, that the inner self faces across a boundary. The image is so familiar it seems like common sense. I am in here, and the world is out there. Elias argued that it is not common sense but a historical product. It arose with the civilising process he traced in his great early work, as Europeans learned over centuries to restrain their impulses, hide their bodily functions, control their emotions and experience themselves as separated from others by an invisible wall of self-restraint.

Against the closed person Elias proposed open people, open people, whose selves are formed in and through their interdependence with others. A person is not a container facing a world. She is a nexus in a figuration, a web of interdependent people, and what she is (her character, her capacities, her sense of who she is) is formed by her position in the web and by the history of her relations within it. There is no "individual" before society and no "society" apart from individuals. There are figurations of interdependent people, and the individual and the social are two aspects of the same process. Elias used the image of a dance. The dance has no existence apart from the dancers, yet no dancer can dance it alone. Each dancer's moves are shaped by the others', and the dance's pattern is no one's design.

Elias's critique applies to the possessive grammar exactly. To say humans hold multiple identities is to locate the identities inside the container: the person has them, as an inner possession, and brings them to her interactions. To say humans act with intention is to locate intention inside the container, as an inner state that precedes action and directs it. To say the people remain distinct is to affirm the container's wall: the knowledge between people may be entangled, but the people inside the walls are not. Each statement is the closed person.

Elias's alternative is not Barad's. Elias does not dissolve persons into intra-actions. Persons are real, distinct and individual, often intensely so, since the civilising process produces exactly the experience of intense individuality. But their distinctness is produced, in figurations and over time, and it is not presumed before the figuration. A person is distinct as a dancer is distinct, by her place and movement in the dance, not as a stone is distinct, by its boundary.

Mead and the Social Self

George Herbert Mead's Mind, Self, and Society (1934), assembled from his students' lecture notes, gives the relational account of the person at the level of social psychology. Mead's central claim is that the self is not given at birth but arises in social experience. Its mechanism is the conversation of gestures. A gesture is an act that calls out a response in another. When an organism becomes able to call out in itself the response its gesture calls out in the other, to take the attitude of the other towards its own gesture, the gesture becomes a significant symbol. Significant symbols make mind possible, and mind, for Mead, is the internalised conversation of gestures: thinking is talking to oneself, taking the attitudes of others towards one's own proposed acts.

The self arises from this process. Mead distinguishes the me, the organised set of attitudes of others that one takes towards oneself, from the I, the response of the organism to those attitudes, the spontaneous, novel element in conduct. The self is the relation between them. The me is social through and through: it is the community's attitudes, internalised. The I is the source of novelty, but it is a response, and it exists only in relation to the me it responds to. Above the particular attitudes of particular others stands the generalised other, the organised attitudes of the whole community, which the self takes towards itself in becoming a full member of the community.

On Mead's account, Snowden's three marks are products of the social self, not properties of a pre-social person:

  • Multiple identities are multiple me's, the attitudes of different groups taken towards oneself. A person has as many social selves as there are groups whose attitudes she takes. The selves are called out by the groups, not held independently of them.
  • Intentionality is the internalised conversation of gestures directed at the future: the rehearsal of an act with its anticipated responses. An intention is a gesture tried out in imagination against the imagined responses of others. Its content is social from the start.
  • Abstraction is the significant symbol, whose meaning is shared because each party can take the attitude of the other towards it. Abstractions are social objects. They exist in the community's shared attitudes, not in individual heads.

Mead does not deny individuality. The I is irreducibly individual, the source of every novel response. But the individual is not a container of properties. She is a social process with a novel edge.

Possessive Individualism

The possessive grammar has a political history, and the history matters. C. B. Macpherson's The Political Theory of Possessive Individualism (1962) traced, from Hobbes to Locke, the emergence of a conception of the individual as proprietor of her own person and capacities, owing nothing to society for them. On this conception, freedom is freedom from dependence on others, society is a set of market relations among proprietors, and political society is a device for protecting property, including the property each individual has in her own person. Macpherson argued that possessive individualism was the ontology of the emerging market society: the individual as owner of her labour, which she can sell, and of her capacities, which are hers to use.

The possessive grammar of the human cluster is the grammar of possessive individualism. Humans hold identities, act with intention and create abstractions, as proprietors of capacities. The grammar is not innocent, because possessive individualism carries a further implication. What a proprietor owns can be exchanged. If a person holds her narratives, interpretations and noticings as properties, those properties can be offered, collected, aggregated and, in the last stage, harvested. The narrator who owns her story can give it to a SenseMaker study. The noticer who owns her noticing can offer it to SwarmCompass. The harvest of human noticing as training data for automation, which The Compass and the Loom found in the instruments it examined, is the market completion of possessive individualism. The proprietor's capacities are collected, aggregated and turned into capital.

A relational grammar would resist this, because relations cannot be owned by one party. A meaning made in participatory sense-making belongs to neither participant. An identity called out by a group belongs to the person-in-the-group, not to the person alone. A joint intention belongs to the we. Relational goods cannot be given by one party, because one party does not have them to give. That is why the consent question is so hard for narrative instruments, and why it cannot be solved by individual consent alone. Individual consent covers what the individual owns. It does not cover the relational meaning her story carries, which belongs also to the others in it.

A Tradition Snowden Draws On: Confucian Role Ethics

The possessive grammar is a specifically Western grammar, and anthro-complexity draws heavily on Chinese thought through Jullien, so it is worth noting that the Chinese tradition offers a sharply relational alternative. Roger Ames, in Confucian Role Ethics (2011), and Herbert Fingarette before him, in Confucius: The Secular as Sacred (1972), argued that the Confucian person is not an individual with properties but a nexus of roles: son, father, minister, friend, teacher. On this reading the person is constituted by her roles and the relations they involve, and has no self apart from them. Virtue, rén 仁, is not a property of an individual but the quality of conduct in relationships. The character combines "person" (人) and "two" (二): personhood as two-ness, the human as between.

Ames calls this a "relationally constituted" conception of the person, and contrasts it with the "discrete individual" of Western liberalism. The Confucian person does not hold roles as one holds a passport. She is her roles, as lived and improved through practice. A person with more roles, more deeply lived, is more of a person, not a person with more possessions. And the cultivation of the person is the cultivation of her relations: becoming a better daughter, friend and colleague, not acquiring better properties.

The tradition matters for this disputation because it shows that the relational alternative is not a modern Western invention (Mead, Elias, enactivism), nor a postmodern dissolution (Barad). It is one of the oldest and most developed accounts of the human in world philosophy, and it belongs to the same tradition from which anthro-complexity takes propensity, 勢. A complexity of the human that took 勢 from China and left 仁 behind has taken the strategist's reading of the situation and left the person-between, who is the one being read.

Snowden's Crews and Roles: The Middle Ground

The adversary will rightly point out that Snowden's own account of multiple identities is not purely possessive. His distinction between roles and crews is contextual. A crew is a temporary identity aligned to a role, formed around a shared task and dissolved when the task is done. Crews are not held. They are formed. A person becomes a member of a surgical crew by joining the surgery, and stops being one when it ends. That is closer to Mead and Elias than the possessive grammar of the November post suggests.

The point is fair, and the crew is the best available bridge between anthro-complexity's current statement and a relational one. But note how the crew is described. It is aligned to a role. The role is the stable element: a formal position, defined in advance, that persons occupy. The crew is the temporary instantiation of roles around a task. The relational element, the crew, is anchored by a non-relational one, the role. And the person who moves between crews and roles is still described as a bearer, of multiple identities, who brings them to crews and takes them away.

A fully relational account would say more. Roles are themselves relational: a role exists only in relation to other roles (surgeon to anaesthetist to nurse) and to the expectations of those who occupy them. Crews do not instantiate pre-existing roles. They make roles, in the sense that what it is to be a surgeon in this theatre, with this team, today, is worked out in the doing. And the person who moves between crews is not a constant bearer carrying identities in and out. She is changed by each crew she belongs to, so that the person who leaves is not quite the person who arrived. "Everyone is a reagent," as The Compass and the Loom puts it.

Bakhtin: Every Utterance Is Addressed

One further resource bears directly on the instruments. Mikhail Bakhtin's theory of dialogism holds that every utterance is a link in a chain of communication. It responds to previous utterances and anticipates responses to come. An utterance has addressivity: it is directed to someone, and its form is shaped by whom it is directed to and what response it anticipates. There is no utterance that is not addressed. Even an utterance to oneself is addressed to an internalised other. And the meaning of an utterance cannot be extracted from its dialogic context, because the context, including the anticipated response, is part of what it means.

Apply Bakhtin to the micro-narrative. A micro-narrative shared in a SenseMaker study is an utterance, and it is addressed. To whom? To the instrument, and behind it to whoever the narrator imagines will read it: the organisation's leadership, the consultants, colleagues, posterity. The narrator shapes her story according to the addressee. A story told to leadership through an anonymised instrument is not the story she would tell her colleague over coffee, or her partner at home, or the union representative. Each addressee calls out a different story. The micro-narrative is not a sample of the narrator's experience. It is a response to the instrument's address, shaped by the narrator's sense of who is listening.

On the possessive grammar, the narrator has an experience and offers a story about it. The story is a report of something she owns. On Bakhtin's grammar, the narrator responds to an address with an utterance whose meaning is shaped by the address. The story is a move in a dialogue. The possessive grammar invites the instrument to treat stories as data about narrators. The dialogic grammar requires the instrument to treat stories as responses to itself, and to include itself, its prompt, its frame and its perceived audience, in what it reads.

Consequences for the Instruments

The possessive fallacy is not merely a philosophical error. It shapes what gets built. Instruments built for possessors differ from instruments built for relations, and the differences can be stated concretely.

Design question Instrument for possessors (current) Instrument for relations (proposed)
Unit of collection The individual narrative The narrative and its address: to whom, in response to what, anticipating what
Who signifies The individual narrator, alone The narrator, and also those in the story, or the group in which it was told
What is aggregated Distributions of individual significations Patterns of relation: who tells what to whom; how stories answer one another
Identity Narrator selects demographic/role categories The prompt and context that called out the identity are recorded; identity treated as produced
Intention Inferred from individual narratives Joint intentions sensed through shared tasks and collective narration
The instrument itself Outside the data Inside the data: its prompt, its signifiers, its perceived audience are part of what is read
Consent Individual consent to share Individual and relational: those in the story have a stake in its use
Feedback Patterns returned to population (sometimes) Patterns returned first to those whose relations they describe, and their response becomes part of the record

Each proposed design is feasible. Some are already partly practised in anthro-complexity's more careful deployments. Collective signification exercises, in which groups signify together, approach the relational unit. Anecdote circles, the Stage One method, collect stories in conversation and keep something of their dialogic context. The proposal is not to abandon the instruments but to redesign them around the relational account of the human that anthro-complexity's own Element 2, and the science of Chapter 13, already support.

The adversary's strongest reply is that the relational account makes instruments impossible at scale. Collecting individual narratives from ten thousand people is feasible. Collecting the relational context of ten thousand narratives (who was addressed, what was anticipated, who else was in the story, how the story answers others) is not. The possessive grammar is not a philosophical error but a practical necessity. At scale, the individual is the only unit that can be sampled. To insist on relations is to confine complexity practice to the small group, which is CRP's limitation.

The reply wins half, and it is the most serious reply to this thesis. It is right that relational data is costlier than individual data, and that some of what the relational account asks for cannot be had at scale. But it is wrong that the individual is the only unit that can be sampled at scale. Network studies sample relations at scale, and narrative methods can collect the addressee and context of a story at low cost: "to whom would you tell this story? who else was involved?" More fundamentally, the reply concedes the thesis. If the possessive grammar is a practical necessity of scale, it is not an ontological truth about humans, and the human distinction should not be stated in it. The instruments can sample individuals as a practical approximation, while the theory states the human relationally and the instruments mark what the approximation leaves out. What the possessive fallacy names is not the use of individual sampling. It is the elevation of individual sampling's unit to the definition of the human. partial

Verdict

Thesis I: the possessive fallacy. holds

The human cluster states the distinction in a grammar of possession (hold, act with, create; the people remain distinct) that presumes persons as parts, contrary to Element 2. The grammar is Elias's closed person and Macpherson's possessive individualism. Mead, Elias, Bakhtin, Confucian role ethics and the science of Chapter 13 all offer a relational alternative that keeps the distinctness of persons as produced in relation, without Barad's dissolution. The fallacy has consequences for instruments, which are built to collect what possessors have and not to sense the relations in which human meaning is made. The adversary's reply from scale is serious and partly successful. It shows that the possessive unit may be a practical necessity, but not that it is the right definition of the human.

NOT A CONTAINER FACING A WORLD —

A DANCER IN A DANCE NO ONE DANCES ALONE —

人 AND 二: THE HUMAN IS BETWEEN.

Addition: Thesis I After Book Nine, and Its Argument in Cards

The trialectic denies possession in practice: capacity is held by neither party. The post on Luhmann keeps it in the unit of collection: a narrative stays tied to its teller. The thesis is narrowed accordingly. See Chapters 48 and 52.

The three cards below set out the thesis in full: its claim and premises, the evidence with the strongest objection and the reply, and the consequence with a test and the verdict as it now stands.

Chapter 20

Thesis II: The Asymmetry Principle Applies to Itself

4,209 words, about 18 minutes of reading

Thesis II: the asymmetry principle applies to itself. If the distinctive human capacity is to tell stories that change constraints, then Cynefin, SenseMaker, estuarine mapping and the November post itself are such stories. Anthro-complexity exempts its own narrative-making from the asymmetry it calls foundational. In Stacey's terms it keeps a formative teleology for the system and a rationalist teleology for the practitioner. (Theses 6 and 7.)

The Principle Restated

In paraphrase: the river shapes the city and the city's planners argue about the river. Both are actants, but only one of them can change the story it tells about the other. That asymmetry, Snowden says, is where anthro-complexity starts.

Chapter 4 set out the principle's structure: narrative asymmetry within a symmetric ontology of action. Both river and planners act, but only the planners narrate, and because narration changes constraints, the planners have a kind of agency the river lacks. They can change the conditions of the relation by changing how they understand it.

The principle can be read in three ways, each stronger than the last.

The descriptive reading. Humans can produce descriptions of the things they interact with, and non-humans cannot. This reading is true, uncontroversial and too weak to ground a body of thought. Many things can be described by humans without the description mattering.

The constraint-changing reading. Human descriptions of the things they interact with can change the constraints of the interaction. The planners' story about the river (hazard, resource, amenity, sewer) changes what they do to the river and what they let it do to them, and so changes the constraints the river and the city impose on each other. This is the reading the November post intends, and it is what makes the asymmetry matter.

The reflexive reading. Human descriptions of the systems they are part of can change those systems, including the describers' place in them. The planners' story about the river changes the river's constraints on the city, and also the planners' own position: their authority, their budgets, their relations with residents, their sense of who they are. A planner who describes the river as a flood hazard becomes a flood-defence planner, with the powers and responsibilities that come with it. The description changes the describer.

The constraint-changing reading is the one anthro-complexity uses. The reflexive reading follows from it, because the planners are part of the system whose constraints their stories change. This chapter argues that anthro-complexity has not drawn the reflexive reading's consequence for itself.

The Theory as Actant

Apply the constraint-changing reading to anthro-complexity's own descriptions.

Cynefin as story

Cynefin is a story about situations. It says situations come in kinds (clear, complicated, complex, chaotic, with the aporetic at the centre), that each kind has a characteristic relation between cause and effect, and that each calls for a characteristic decision model. When an organisation adopts the story, it changes the constraints of the organisation's situations.

Chapter 13's analysis through Searle makes the mechanism precise. When an organisation declares a situation "complex", and the declaration is collectively accepted, the category becomes a status function with deontic powers. It licenses probing instead of planning, permits safe failure, relieves managers of the demand for a business case, and legitimises the language of emergence. These are changes in the constraints within which people act. A manager who would have been censured for launching an experiment without a plan is now praised for probing. A team that would have been told to deliver a specified outcome is now told to amplify what works. The story has changed the constraints.

It has also changed the story-tellers' position. The person who categorises the situation, whether a leader, a consultant or a facilitated group, acquires the authority that comes with categorisation. To say "this is complex" is to claim the right to say what kind of situation this is, and therefore what kind of response it calls for. The categoriser stands, for the moment of categorisation, in the position of the one who reads the situation, and the reading has consequences for everyone else.

SenseMaker as story

SenseMaker produces stories about stories. Its patterns (clusters of narratives on triads, distributions on dyads, outliers in the corners) are descriptions of a population's experience, and when they are fed back, to the population or to leadership, they change the constraints of the population's life.

A pattern showing that stories of "compliance" cluster in one department and stories of "trust" in another is a description, and once known it becomes a constraint. The department of compliance has been named. Its members know it, its managers know it, and those who allocate resources and attention know it. The next round of narration is shaped by the naming. Some narrators confirm the pattern, some resist it, and some tell stories designed to change it. Hacking's looping effect applies. The pattern makes up the department, and the department's response changes what the pattern means.

Estuarine mapping as story

An estuarine map is a story about constraints: which can be shifted, at what energy cost, on what timescale. Once a map is accepted by an organisation, it defines what counts as easy or hard to change. MacKenzie's performativity applies. The map does not just describe the organisation's constraint landscape. It shapes the organisation's behaviour towards its constraints, so that constraints mapped as costly are left alone and those mapped as cheap are attacked. The map makes the terrain it maps, to a degree that depends on how widely it is accepted and how authoritatively it is held.

The November post as story

The November post is a story about how complexity and systems thinking stand to each other: neighbours, not branches, tree, rock and water. It is a move in the field it describes. It changes how readers in both traditions understand the relation, how they position their own work, and how they read Ing's programme and Snowden's. The post's own image, three actants each constraining and enabling the others, describes this exactly. The post is one of the actants, constraining and enabling the others.

The slogan as story

Finally, the slogan humans are not ants is a story, and it has changed constraints. Organisations that have heard it think twice before using swarm metaphors, ant colony models and agent-based simulations as if they described their people. The slogan's performativity is beneficial: it has helped free organisations from a misleading class of models. But it is performativity nonetheless. The slogan is not a neutral report of a fact about humans. It is a description that, taken up, changes how humans in organisations are treated.

The Exemption

Each of these is an instance of the asymmetry principle. Humans (Snowden, Cynefin practitioners, SenseMaker designers, estuarine mappers) tell stories about the systems they interact with, and the stories change the systems' constraints. Anthro-complexity's descriptions are among the most powerful constraint-changing stories in contemporary organisational life.

Yet the theory does not describe its own descriptions this way. In the November post, the asymmetry principle is stated about planners and rivers, and implicitly about the humans in the systems anthro-complexity studies. It is not stated about anthro-complexity's own practitioners in relation to those systems. Cynefin is presented as a sense-making framework, a way of seeing what kind of situation one is in. SenseMaker is presented as distributed sensing, a way of seeing patterns in what people already experience. Estuarine mapping is presented as a way of seeing which constraints can be shifted. In each presentation the instrument sees. In none of them does it act on the system by describing it, though on the asymmetry principle that is what each does.

This is the exemption. The theory says humans change the constraints of the systems they describe, and it presents its own descriptions as seeings, not changings. The exemption is not stated. It is a matter of emphasis, vocabulary and the framing of instruments. But it is consistent, and it is structural.

Split Causality

Ralph Stacey's diagnosis of split causality, introduced in Chapter 11, names the exemption exactly. Most management theories combine two teleologies without noticing. The organisation is formative, unfolding what is enfolded, while the manager is rationalist, choosing and steering. The system is governed by one kind of causality and the person who acts on it by another, and the person is exempted from the causality that governs the system.

Anthro-complexity has a more sophisticated version. The system is not formative in Stacey's narrow sense, unfolding a pre-given form. It is complex: emergent, dispositional, coherent only in retrospect. But the practitioner is still exempted. She reads the system's dispositions, maps its constraints, senses its patterns, categorises its situations and chooses which constraints to shift and which probes to run. In doing these things she is described as a reader and chooser, a rationalist in Stacey's sense, and not as a participant whose readings and choices are themselves gestures in the system, called out by and calling out the responses of others.

The Compass and the Loom stated the alternative: every chooser is chosen too. The practitioner's choice of what to probe is shaped by her position in the figuration, by her clients' expectations, by the instruments available and by her own history in the field. Her probe, once made, calls out responses that change her. Transformative teleology applies to the practitioner as to everyone else.

Proposition granted: Only humans can change the story they tell about the other, and this is where anthro-complexity starts.

Then it must be granted that anthro-complexity's practitioners, being human, change the stories they tell about the systems they study. — Granted.

Then it must be granted that the changed stories change the constraints of those systems. — Granted. That is the force of the principle.

Then it must be granted that anthro-complexity's descriptions (Cynefin categories, SenseMaker patterns, estuarine maps) are constraint-changing stories, and not only seeings. — Granted.

Then it must be granted that an account of what these descriptions reveal is incomplete unless it includes what they change. — Granted.

Then it must be granted that the practitioner, as the teller of these stories, is changed by telling them and is part of the system whose constraints they change. — Granted.

Then the respondent must say whether anthro-complexity's instruments and their presentation account for this. — In part. Snowden's insistence that categorisation is collective, that SenseMaker returns patterns to populations and that estuarine maps are made by participants all distribute the story-telling. None of them describes the instruments as constraint-changing stories, or records how the stories change the systems and the tellers. The principle is stated outward, not inward.

The Claim to "Neither"

There is a characteristic gesture in Snowden's writing that this disputation, and To the Strait-Jacketed, the Other Always Seems Baggy before it, has had to examine. It is the claim to neither. In the essay that treatise examined, Neither Baggy Nor Boxed, Snowden positioned anthro-complexity between two errors. One is the baggy, unstructured conversation of CRP. The other is the boxed, over-structured frameworks of systems thinking. Anthro-complexity, he argued, is neither: it offers enough structure to act and enough openness to stay complex. The November post makes the same gesture several times. Anthro-complexity is neither systems thinking nor flat ontology, neither Barad nor Descartes, neither dyadic nor dualistic, neither Augustinian nor merely laissez-faire.

The claim to neither is often right, in that anthro-complexity does occupy a position distinct from both errors. But the gesture has a structure the asymmetry principle illuminates. To claim to be neither of two positions is to claim a vantage from which both can be seen as errors. The Compass and the Loom drew the parallel with Wigner's friend: the claim to stand "outside the lab" parallels Snowden's claim to "neither". Wigner, in the thought experiment examined in Chapter 39, stands outside the laboratory in which his friend performs a measurement, and asks what state the laboratory is in from his outside perspective. The puzzle arises because Wigner's outside perspective and the friend's inside perspective give different answers. Neither can be privileged without assuming that one observer stands outside the physics that governs the other.

The claim to neither, read through the asymmetry principle, is a story that positions its teller outside the alternatives it rejects. Like every story, it changes constraints. It positions anthro-complexity as the reasonable middle, the vantage from which baggy and boxed, systems and flat, can be judged. The positioning is a move in the field, and it is effective. But it is not neutral. To the Strait-Jacketed found the twin of the claim in its opposite. CRP sees anthro-complexity as boxed and itself as free, just as anthro-complexity sees CRP as baggy and itself as balanced. Each side sees the other's form as "baggy" or "boxed", while its own form feels like freedom. No one sees the form they are standing in.

Snowden's Own Reflexivity

Fairness demands an acknowledgement that the November post itself shows. Snowden is among the most personally reflexive writers in the management field. In the post he revises his first reading of Ing in public, on two points (the containing whole and the role of the practitioner), and says so. He notes the limits of his sources: one paper, one lecture deck, and a ten-year programme that may address his objections. He holds the dyad-and-triad point "most loosely" and asks whether the Chinese sources have resources he is missing. He opens with an invitation to correction: if I have misread either source, I would welcome the correction.

This is reflexivity of a high order, the reflexivity of an author who knows his readings are readings and holds them open to revision. It deserves to be recognised.

But it is the reflexivity of an author, not of an instrument. Snowden's personal willingness to revise is a quality of his practice as a writer and thinker. It is not built into Cynefin's structure, SenseMaker's design or estuarine mapping's protocol. A practitioner trained in the instruments who lacks Snowden's personal reflexivity will use them without it, and nothing in the instruments will prompt her otherwise. The asymmetry principle applied to itself demands structural reflexivity. Instruments themselves must be designed to account for their own effects, so that reflexivity does not depend on the virtue of the user.

Structural Reflexivity: What It Would Require

What would it mean for anthro-complexity's instruments to be structurally reflexive, to embody the asymmetry principle applied to themselves? Four requirements follow from the analysis.

First, the instrument records its own descriptions. Every deployment records not only what was found but what was said: the prompts used, the signifiers offered, the categories applied, the patterns reported, and to whom. That is the record of the stories the instrument told.

Second, the instrument records the effects of its descriptions. After a pattern is fed back, after a situation is categorised, after a map is accepted, the instrument senses what changed: in subsequent narration, in behaviour towards mapped constraints, in the authority of the categorisers. The Compass and the Loom's demand that instruments should record how probes change the prober is the core of this.

Third, the instrument includes the teller. The practitioner, convenor or designer is among those whose experience is sensed. The Compass and the Loom's principle reading of the reader states it: convenors signify their own experience, published alongside everyone else's. The teller's story about the system is part of the system's story about itself.

Fourth, the instrument declares its teller's position. Before deployment, the designers state what they expect to find, what they hope to find and what they would prefer not to find, the Tidal Loom's declination before direction. The declaration makes the teller's position visible, and so makes the story's constraint-changing direction visible.

None of these requirements is exotic, and each could be added to existing instruments at modest cost. Together they would make the asymmetry principle structural: every instrument would be designed as a story-teller that changes constraints, and would account for itself as one.

Requirement What it records Tidal Loom principle Current practice
Record descriptions Prompts, signifiers, categories, patterns, recipients — Partly (deployment records)
Record effects Changes after feedback, categorisation, mapping Noticing, not mapping; return as design requirement Rarely tracked separately
Include the teller Convenors' and designers' own experience Reading of the reader Rarely
Declare position Expectations, hopes, aversions of designers Declination before direction Rarely

The adversary will press two replies. First, structural reflexivity is a counsel of perfection. No instrument can fully account for its own effects, because its effects ramify without limit, and an instrument that tried would be paralysed by self-reference. Second, anthro-complexity already has a structural mechanism for reflexivity: the probe-sense-respond cycle. A probe is made, its effects are sensed and the response adjusts. That is recording how probes change the system, including how the instrument's descriptions change it, because the descriptions are among the probes.

The first reply wins its half. Full self-accounting is impossible, and the regress is real. But the requirements above are not full self-accounting. They are four modest additions, each of which records something specific and none of which requires infinite regress. The second reply is more interesting and wins less. Probe-sense-respond does sense the effects of probes on the system. But it is designed to sense the system's response to the probe, the experiment, not to the description, the categorisation that licensed the experiment, the signifiers that framed the sensing or the pattern that prompted the response. The descriptions are the conditions of the probe, not its content, and they are not sensed. Structural reflexivity would add the descriptions to what is probed. partial

The Asymmetry Re-entered

Chapter 18 ended with Spencer-Brown's re-entry. The human/non-human distinction, re-entered into the human side, divides humans into describers and described. The asymmetry principle re-enters with it. Among humans, describers can change the story they tell about the described more easily than the described can change the story told about them.

The re-entered asymmetry is where this thesis lands. Anthro-complexity's practitioners are describers, and the populations whose narratives they gather, whose constraints they map and whose situations they categorise are described. The practitioners' stories about the populations change the populations' constraints. The populations' stories about the practitioners, what they think the study is for, whether they trust it, how they game it, change the practitioners' constraints much less, because the instruments are not designed to sense them and the practitioners' position does not depend on them in the same way. The planners argue about the river. Within the city, the planners also argue about the residents, and the residents' arguments about the planners are not heard in the same room.

The asymmetry principle, applied to itself, does not say that anthro-complexity is wrong to describe human systems. It says that anthro-complexity's descriptions are among the most consequential stories in those systems, that they change constraints and tellers alike, and that the asymmetry between describers and described is a human-human asymmetry the principle's own logic demands be made visible.

Verdict

Thesis II: the asymmetry principle applies to itself. holds

Cynefin, SenseMaker, estuarine mapping, the November post and the slogan humans are not ants are constraint-changing stories, in the precise sense the asymmetry principle identifies. Anthro-complexity presents them as seeings, not changings, and so exempts them from the principle it calls foundational, which is Stacey's split causality in a more sophisticated form. Snowden's personal reflexivity is real and exemplary, but it is the reflexivity of an author, not of an instrument. Structural reflexivity would require four modest additions: record descriptions, record their effects, include the teller, declare the teller's position. The adversary's replies show that full self-accounting is impossible and that probe-sense-respond is a partial mechanism. They do not show that the exemption is justified.

THE PLANNERS ARGUE ABOUT THE RIVER —

THE CONSULTANTS ARGUE ABOUT THE PLANNERS —

AND THE STORY OF THE STORY CHANGES THE STORYTELLER.

Addition: A Ledger of Anthro-Complexity's Performatives

Thesis II argued that anthro-complexity's descriptions are constraint-changing stories, and that the theory exempts them from the asymmetry principle it calls foundational. This addition makes the argument concrete with a ledger: a list of anthro-complexity's characteristic descriptions, each with the constraint it changes when adopted, the deontic power it creates (Searle, Chapter 13), whose position it strengthens, and whose it may weaken. The ledger is illustrative. Practitioners will know of descriptions it omits, and of effects it misjudges.

The ledger

Description When adopted, changes the constraint that… Deontic power created Strengthens May weaken
"This situation is complex" Plans must precede action Licence to probe; permission to fail safely Those who wish to experiment; facilitators of probes Planners; those who demand accountability for outcomes
"This situation is clear" Discretion is allowed Duty to follow best practice Process owners; auditors Practitioners with local judgement
"This is a weak signal" Outliers may be ignored Duty to attend to the anomaly Those who reported it; analysts Those whose priorities it distracts from
"Safe-to-fail" Failure is censured Immunity for failed probes Designers of probes Those who experience the failure (Chapter 16)
"Amplify what works, dampen what does not" Interventions must be justified in advance Permission to scale or stop on observed effect Those who judge what "works" Those whose preferred outcomes are dampened
"Behaviour is the shadow of conditions" Individuals are responsible for behaviour Licence to change conditions rather than address persons Those with power to change conditions Those whose conditions are changed (Chapter 16)
"Humans are not ants" Swarm and agent models may guide policy Duty to treat people as narrators and intenders Narrative practitioners Agent-based modellers; swarm-metaphor vendors
"This constraint is fixed" (on a map) It may be challenged Permission to work around it Those who judged it fixed Those who wanted it changed (Chapter 22)
"These stories cluster in this department" The department is unremarkable Licence to intervene in the department Commissioners; analysts The department's members (Chapter 28)
"The system lacks something" (Augustinian) The system can heal itself Licence to supply from outside Consultants who supply The system's own capacities
"The system has what it needs" (Pelagian) Outside supply is necessary Duty to work with what is there The system's members Consultants who supply, unless they supply instruments
"Neither baggy nor boxed" The field is divided between two schools Licence to occupy the middle Anthro-complexity CRP; systems thinking
"Complexity is not a kind of systems thinking" Complexity belongs in the systems lineage Duty to treat complexity as autonomous Complexity theorists Systems theorists who absorb it

What the ledger shows

Three things stand out.

Every description has winners and losers. None of these descriptions is neutral. Each, adopted, changes a constraint, creates deontic powers and redistributes them. That is not a criticism of the descriptions, many of which are true and most of which are useful. It is the asymmetry principle stated without exemption. Descriptions in human systems change constraints, and changed constraints redistribute power.

The winners are often those who use the descriptions. In several rows, the description strengthens the position of those who apply it: facilitators of probes, analysts, designers, those who judge what works, those with power to change conditions. That is the positional reason (Chapter 17) in miniature. A body of thought's descriptions tend to strengthen the position of its practitioners. It is not bad faith. It is a structural feature of descriptions in reflexive systems, and it should be declared.

The losers are often those described. In several rows, the description weakens the position of those it describes: the department whose stories cluster, the frontline who experience the failed probe, those whose conditions are changed, those whose wanted change is mapped as fixed. That is the re-entered asymmetry (Chapter 18). Among humans, the describers change the story about the described more easily than the reverse, and the change tends to favour the describers.

The ledger as instrument

The ledger could itself be an instrument. Before a description is offered to a client or a population (a categorisation, a pattern, a map or a principle), the practitioner would complete its row. What constraint will this change? What powers will it create? Whose position will it strengthen, and whose weaken? Completing the row would not prevent the description from being offered. It would make its performativity visible to the practitioner and, if shared, to those it will affect.

That is the Tidal Loom's declination before direction applied to descriptions rather than to signifier sets. It is also Austin's felicity conditions (Chapter 34) made explicit. A performative whose effects are declared before it is uttered is one whose uptake is informed. It is a performative of address, not of capture.

EVERY DESCRIPTION HAS WINNERS AND LOSERS —

THE WINNERS ARE OFTEN THE DESCRIBERS —

COMPLETE THE ROW BEFORE YOU SPEAK.

Addition: Thesis II After Book Nine, and Its Argument in Cards

Snowden has written that all diagnosis is intervention, that no method wholly avoids becoming narrative power, and that a system cannot observe the distinction by which it observes. The thesis is partly conceded. The open matter is the designer. See Chapters 52 and 56.

The three cards below set out the thesis in full: its claim and premises, the evidence with the strongest objection and the reply, and the consequence with a test and the verdict as it now stands.

Chapter 21

Thesis III: Humans Are Not Estuaries

5,894 words, about 26 minutes of reading

Thesis III: humans are not estuaries. The theory that forbids modelling people on ants models practice on tides, estuaries, rivers, waterfalls, surfing and sailing. These images share two properties that human systems lack. The medium is not reflexive, and the reader stands outside the reading. Jullien's 勢, the source of anthro-complexity's propensity, includes the strategist in the configuration. A propensity read from outside is a landscape, not a 勢. In human affairs, the tide turns partly because someone visible was seen to wait for it. (Theses 8, 9 and 20.)

The Images

Chapter 10 ended with a question Pepper would ask: what is anthro-complexity's root metaphor? It proposed the estuary, and predicted that if the estuary is the root metaphor, it will carry its own commitments into every domain to which it is extended, including the human. This chapter tests the prediction.

Begin by collecting the images. In the November post and the anthro-complexity corpus around it, the governing images of complex systems and of practice within them are:

Image Where it appears What it pictures
The estuary Estuarine mapping; the November post A complex system of constraints with different energies and timescales, governed by tides
The turn of the tide November post, on kairos The opportune moment, when what was impossible becomes easy
The sailor November post; long-standing in Snowden The practitioner who works with propensity and does not fight the tide
The surfer Ing's lecture, endorsed by Snowden Motion read from the wave, propensity against causality
The water-skier Ing's lecture (as contrast) Motion pulled by a boat, causality
The river and the city November post, the asymmetry principle Both actants; only the planners narrate
Tree, rock, waterfall November post, closing image Three actants, none containing the others
The trellis Snowden's enabling constraints Structure that enables growth without determining it
The rainforest (against the managed garden) Snowden's broader corpus Complex ecosystems against ordered cultivation

Every one of these images is drawn from non-human nature, or from a human engaging with non-human nature: the sailor and the tide, the surfer and the wave, the gardener and the trellis. The human appears in them, if at all, as the reader of a natural medium. That medium is water, more often than not, and it does not read back.

The Estuary as Root Metaphor

Estuarine mapping is anthro-complexity's most developed recent method. Its name was chosen to evoke a specific kind of complex system. An estuary is where a river meets the sea: fresh water and salt, current and tide, sediment and scour, all interacting. Its constraints have very different energies and timescales. The shape of the coastline changes over centuries, the channels over years, the sandbanks over seasons, the flow over hours. It is governed by the tide, whose turn changes everything: what could not be done an hour ago becomes easy, and what was easy becomes impossible. A skilled pilot knows the estuary's constraints, the depth of each channel, the set of each current, and above all its timing. She knows when to wait and when to move.

The image is rich and the method is powerful, and the November post uses it to state anthro-complexity's theory of timing:

In paraphrase: kairos is the turn of the tide, the moment the flow reverses, so that what was impossible an hour before becomes easy, or the reverse. An estuary is ruled by that turn. Much of estuarine mapping, Snowden writes, concerns which constraints can be shifted now and which must wait for the tide, given their different energies and timescales. Any sailor, he adds, knows not to fight the tide but to wait for it to turn, and then move.

In Pepper's terms, this is a root metaphor at work. An image from common experience, the pilot in the estuary, is extended to cover a domain, the practice of intervention in complex human systems, and its structure organises the domain. Constraints become channels and sandbanks, with depths and costs. Timing becomes the tide, with its turn. The practitioner becomes the pilot, who reads the estuary and chooses when to move. The organisation becomes the estuary.

Pepper's theory predicts that a root metaphor carries its structure into every domain it covers, and that the structure includes features of the source that may not fit the target. The question is what the estuary carries that does not fit human organisations.

Two Properties

The estuary, and every image in the table above, has two properties that human systems lack.

Property one: the medium is not reflexive

The tide does not read the tide-table. The estuary's channels do not respond to being charted. The wave does not notice the surfer, the river does not hear the planners' arguments, and the rock does not know it is gripped by roots. The medium in every image is indifferent, in Hacking's sense. Being described does not change it. A pilot's chart of the estuary is accurate or not, and its accuracy does not depend on whether the estuary knows about it.

Human organisations are not indifferent. They are interactive kinds, made of people who read descriptions of themselves and respond. Chapter 14 traced five forms of this reflexivity (prophecy, looping, the double hermeneutic, market reflexivity and performativity) and argued that descriptive reflexivity is the human difference of kind. An estuarine map of an organisation is a chart of a medium that reads the chart. The channels move when they are charted, because the people whose habits and expectations make up the channels see the chart and respond.

The root metaphor therefore carries indifference into a domain that is not indifferent. It invites the practitioner to map an organisation's constraints as a pilot charts an estuary, as features of a medium that will be as charted whether or not the chart is seen. The practitioner who follows the invitation will be surprised when her map changes the terrain, and the surprise will be the metaphor's fault.

Property two: the reader stands outside the reading

The sailor is on the deck, not in the tide. The surfer is on the board, riding the wave, not part of it. The planners stand on the bank arguing about the river. The photographer stands outside the frame of tree, rock and water. In every image the human who reads the medium is separate from it: in contact, often intimately, but not of it. The sailor's reading of the tide does not change the tide. The surfer's reading of the wave does not change the wave. The reader is outside the reading, in the precise sense that her reading is not a force acting on what she reads.

In human organisations the reader is inside the reading. A manager who reads her organisation's dispositions is part of the organisation, and her reading is a gesture others read. A consultant who maps an organisation's constraints is, for the duration of the engagement and often long after, one of its constraints. A leader who "waits for the tide to turn" is visible, and her waiting is interpreted, and the interpretation changes the tide.

The Tide That Turns Because You Waited

This last point is the sharpest test of the metaphor, and it deserves to be worked through.

The tide turns by the moon. The sailor's waiting has no effect on it. She waits, the tide turns, and she moves. The turn is independent of the waiting.

Now consider a human organisation in which a chief executive wants to restructure a division. The constraints are formidable: the division head is popular, the restructuring will be resisted, and the board is divided. On the estuary metaphor, the chief executive should wait for the tide to turn, for the moment when the constraints weaken and the restructuring becomes easy. She waits.

But her waiting is visible. The division head notices that the restructuring, rumoured for months, has not happened. He may conclude that she has lost her nerve, and become bolder. He may conclude that she is biding her time, and start looking for another job. The board notices too. Some members conclude she is wise to wait, and others that she is weak. The division's staff, reading the signals, start positioning themselves: some to resist, some to leave, some to be on the right side when it comes. Each of these readings is a response to her waiting, and each changes the constraints. The tide may turn, or it may not. If it does, it turns partly because she waited. Her waiting was a gesture, and the turn is the response it called out.

This is Mead's point in Chapter 27, and The Compass and the Loom's statement of it: withholding is also a gesture, so "just let it happen" is not a neutral option for a visible leader. In human affairs there is no waiting that is not also acting. The leader who waits for the tide is one of the forces that turn it.

The estuary metaphor cannot represent this. Its structure (a sailor who waits, a tide that turns, independently) makes the leader's waiting a non-event, a pause in action until the medium is ready. In human systems the pause is an event, and often the decisive one.

Proposition granted: Kairos is the turn of the tide; you do not fight the tide; you wait for it to turn, and then you move.

Then it must be granted that, in the estuary, the turn of the tide is independent of the sailor's waiting. — Granted. The tide turns by the moon.

Then it must be granted that, in a human organisation, the "turn" (the moment when a constraint weakens) may depend on others' readings of the practitioner's waiting. — Granted. Withholding is a gesture.

Then it must be granted that, in a human organisation, the practitioner's waiting is an action whose consequences must be considered, and not a neutral pause. — Granted.

Then it must be granted that the estuary metaphor, which represents waiting as neutral, misdescribes the timing of human affairs at exactly the point where timing matters most. — Granted.

Then the respondent must say whether anthro-complexity's theory of kairos, stated through the estuary, accounts for the gesture of waiting. — It does not, as stated. The November post's statement is the sailor's, and the sailor's waiting is not a gesture.

Jullien Against the Surfer

The estuary metaphor's two properties, a non-reflexive medium and a reader outside, are not incidental to the propensity theory anthro-complexity takes from Jullien. They contradict it.

Jullien's 勢 (shì) is not a property of a landscape viewed from outside. In Sunzi, from whom Jullien draws the military sense of the concept, 勢 is the potential of a configuration, and the configuration includes the general. Sunzi's famous image is of round stones rolling down a mountain of a thousand feet. The potential is in the arrangement of stone and slope, and the general's art is to place his troops so that their potential is like the stones'. But the placing is part of the configuration. The general does not stand outside the battlefield reading its 勢 as a sailor reads the tide. His disposition of forces is the 勢. And the enemy reads his disposition as he reads the enemy's. Sunzi's whole doctrine of deception (appear weak when strong, appear far when near) presupposes that the general is read, and that his being read is part of the configuration's potential. All warfare is based on deception, Sunzi says. Deception is possible only because the strategist is inside a configuration of mutual reading.

Jullien makes the point explicitly in A Treatise on Efficacy. The Chinese strategist does not impose a plan on the situation. He conditions it, acting upstream so that the configuration evolves in his favour. But the conditioning is itself part of the situation, and the other side conditions too. Efficacy is not the general's reading of a passive medium. It is his participation in a configuration of mutual conditioning, in which his own disposition, visible to the enemy, is one of the factors.

The surfer, by contrast, reads a wave that does not read him. Ing's image (water-skiing pulled by causality, surfing reading propensity) captures the contrast between causation and propensity correctly, but it captures propensity in its non-human form: a potential in a medium indifferent to the reader. Jullien's 勢 in its human form, the strategist's 勢, is a potential in a medium that reads the reader. The surfer is the wrong image for 勢 in human affairs. A better one is the general on a battlefield where the enemy is watching, or the player at a table where the other players are reading her face.

Chapter 35 develops this at length. The point here is that the estuary and the surfer, the governing images of anthro-complexity's practice, take Jullien's 勢 and strip out exactly the feature that makes it a concept of human strategy: the strategist's place inside a configuration that reads him. The Compass and the Loom put it compactly: 勢 cannot be read from outside, because the commander's own disposition is part of the configuration, and mapping it as energy gradients turns it into a landscape.

The Children's Party

There is an older image in Snowden's work that does not have the estuary's two properties, and the contrast is instructive. For many years Snowden illustrated the difference between ordered and complex systems with a children's party. You cannot manage a children's party by setting objectives and measuring outcomes (each child to experience six units of fun, with three games delivered on schedule). That would be a disaster. Nor do you leave the children unsupervised, which is chaos. You set boundaries ("the garden, not the road"), provide attractors (games, food, a bouncy castle), watch for patterns (a game that is working, a child who is being excluded), amplify the good ones, dampen the bad ones, and intervene quickly if something goes wrong. The children's party is complexity management in miniature.

The children's party is a human image, and it has neither of the estuary's properties. The medium is reflexive: the children read the adults. They know when they are being watched, test the boundaries, notice favouritism, and play up to or against the adults' expectations. The reader is inside the reading: the parent at the party is not a pilot on a deck but a participant, whose presence, mood and attention are part of what makes the party what it is. A parent who stands at the edge with a clipboard, reading the party's dispositions, changes the party, and every child knows it.

The children's party was an early image, from the Stage Two period of the genealogy. The estuary is a recent one, from Stage Four. The movement from the party to the estuary is the movement Chapter 6 traced: from a human image of complexity, in which the reader is inside and the medium reads back, to a natural image, in which the reader is outside and the medium is indifferent. The drift in the lineage's instruments is matched by a drift in its root metaphors. As the instruments moved the human from the centre to the periphery, the images moved from the party to the estuary.

Jacobs: The City as Organised Complexity

If anthro-complexity needs a root metaphor for human complexity that does not carry the estuary's properties, one is ready to hand, and it comes from one of the founding texts of complexity thinking about human systems.

Jane Jacobs's The Death and Life of Great American Cities (1961) ends with a chapter titled "The kind of problem a city is". In it Jacobs draws on Warren Weaver's 1948 essay "Science and Complexity", which distinguished three kinds of problem: problems of simplicity (two or three variables, as in classical mechanics), problems of disorganised complexity (millions of variables, treatable statistically, as in thermodynamics), and problems of organised complexity (a sizeable number of factors interrelated into an organic whole). Weaver argued that problems of organised complexity, in biology, medicine, economics and the social sciences, were the great challenge for science. Jacobs argued that cities are problems of organised complexity, and that urban planning had failed because it treated them as problems of simplicity (zoning, separation of uses) or of disorganised complexity (statistical averages, density targets).

Jacobs's city has neither of the estuary's properties. Its medium is reflexive: residents read the city and respond to it, and their responses make it what it is. A street is safe because people use it, and people use it because it is safe. A neighbourhood declines when residents read signs of decline and leave, which is Merton's self-fulfilling prophecy in urban form. And the reader is inside the reading. Jacobs's most famous concept, eyes on the street, is the observer inside the system. Safety on a city street is produced not by police observing from outside but by residents, shopkeepers and passers-by who watch the street because they are on it, whose watching is part of the street's life, and whose being watched by one another is part of why they behave as they do. The observer is the observed. Everyone on the street is both.

Jacobs's critique of planning is anthro-complexity's critique of systems thinking, made sixty years earlier with the city as its object. Planners presumed wholes (the neighbourhood unit, the superblock, the garden city) and designed parts to fit. They read the city from above, on maps and statistics, and imposed plans the city's residents had no part in. The plans destroyed the organised complexity that made cities work. Jacobs's alternative was not a better plan but attention to the city's own processes, the small-scale, mixed-use, fine-grained interactions through which cities organise themselves, and policy that supported those processes rather than overriding them. That is the Pelagian stance, behaviour as the shadow of conditions, and finding coherence instead of presuming it, applied to the city.

The city, read through Jacobs, is a better root metaphor for anthro-complexity than the estuary. It is a human complex system, whose medium reads back and whose readers are inside. The planners argue about the river, but in Jacobs's city they also argue with the residents, who argue back, and the city is made in the argument.

Candidate Root Metaphors Compared

Root metaphor Source Medium reflexive? Reader inside? What it pictures well What it misses
The estuary Estuarine mapping No No Constraints with different energies and timescales; timing Mutual reading; the gesture of waiting
The surfer Ing's lecture No No Propensity against causality 勢 includes the strategist
The children's party Snowden, early Cynefin Yes Yes Boundaries, attractors, amplify and dampen; the watched watcher Scale; material constraints
The battlefield (Sunzi) Jullien's source Yes Yes Mutual conditioning; deception; 勢 including the strategist Cooperation; the non-adversarial
The city (Jacobs) Death and Life Yes Yes Organised complexity; eyes on the street; planning's failure Speed; the institution's inner life
The dance (Elias) What Is Sociology? Yes Yes Figuration; no dancer dances alone; pattern without design Material constraint; scale
The conversation (Mead, Shaw) CRP Yes Yes Gesture and response; meaning completed by the other Material constraint; instruments
The game (Carse) Finite and Infinite Games Yes Yes Rules changed by play; infinite play to continue the game Constraint energetics

The table shows that the estuary and the surfer are the only candidates with neither property. Every human root metaphor (party, battlefield, city, dance, conversation, game) has a reflexive medium and a reader inside it. That is not surprising. It follows from descriptive reflexivity being the human difference of kind. But it shows that anthro-complexity, in choosing the estuary as the root metaphor of its most developed recent method, chose a metaphor that cannot represent the feature its own theory says distinguishes human complexity.

Why the Estuary Appeals

If the estuary misrepresents human complexity on the point that matters most, why was it chosen, and why does it appeal? Four reasons can be offered, and each deserves a hearing, because each points to something real.

First, the estuary represents constraint energetics better than any human image. The central insight of estuarine mapping is that constraints differ in the energy needed to change them and the time over which they can change, and that this variation matters for intervention. The estuary pictures it vividly. Coastlines, channels, sandbanks and currents change on different timescales, and a pilot who confused them would run aground. Human images are less good at this. A children's party has constraints of different durability (the garden wall, the parents' patience, the cake), but the party's short duration flattens the difference. Jacobs's city has it (buildings, street plans, institutions and habits change at different rates), but the city image foregrounds interaction over constraint. The estuary was chosen, in part, because it is the best available picture of what estuarine mapping maps.

Second, the estuary is humbling. A sailor does not command the tide. The image puts the practitioner in a position of respect towards forces larger than herself. It is a corrective to the managerial fantasy of control, in which the organisation is a machine to be steered and the manager a driver at the wheel. The estuary, like the surfer, replaces control with reading, and reading is a humbler stance.

Third, the estuary is natural, and nature lends authority. Chapter 9 observed that animal exemplars lend social proposals the authority of nature. The estuary does too. A practice modelled on how a skilled pilot reads an estuary seems grounded in something real and old, in skill and craft and mētis, unlike a practice modelled on a management framework.

Fourth, the estuary keeps the practitioner's hands clean. This reason is the least comfortable, and it follows from the other three. A practitioner who reads the estuary, waits for the tide and moves with the flow is not responsible for the flow. She works with forces she did not create and cannot command. If the tide carries the organisation somewhere bad, the tide did it. The image locates agency in the medium and gives the practitioner the modest role of timing. In a human system, where the practitioner's reading is a gesture and her waiting is an action, the modest role is not available. She is responsible for her gestures, including the gesture of waiting, and the estuary hides the responsibility by hiding the gesture.

The first three reasons are good ones, and a reflexive anthro-complexity would keep what they point to: the energetics of constraint, humility before forces larger than oneself, and the dignity of craft. The fourth reason is the one the thesis contests. A root metaphor that keeps the practitioner's hands clean is a root metaphor that hides the practitioner's agency, and in a human system the practitioner's agency is part of what is going on.

The adversary has a strong reply. Metaphors are not theories. Estuarine mapping does not claim that organisations are estuaries, any more than Cynefin's children's party claimed they were parties. The estuary is a heuristic, a way of drawing attention to constraint energetics and timing, and its limits are understood by every competent practitioner. To hold a heuristic responsible for what it does not represent is to mistake a picture for a model. Snowden is fierce about the misuse of metaphor in management. He is the last person to be accused of reifying one.

The adversary is right that the estuary is a heuristic, and right that Snowden is vigilant about metaphor. But Pepper's point about root metaphors is precisely that they do not stay heuristic. A root metaphor organises a domain by lending it structure, and the structure travels whether or not the metaphor's users intend it to. The test is not whether practitioners know that organisations are not estuaries. They do. The test is whether the method built on the metaphor accounts for what the metaphor leaves out. Estuarine mapping maps constraints by energy and time, which the metaphor pictures well. It does not, as a method, map how the mapping changes the constraints, or how the practitioner's waiting is read, which the metaphor leaves out. The heuristic's limits are known, and the method has not been extended to cover them. That is the thesis. partial

What an Estuary-Aware Practice Would Need

The thesis does not require anthro-complexity to abandon the estuary. It requires the estuary to be supplemented by an image that carries what the estuary cannot. The method built on it must then be extended to cover what the supplement adds.

Three supplements are proposed, each drawn from the human root metaphors above.

From Jacobs: eyes on the estuary. Every estuarine mapping session includes, as part of the map, the people doing the mapping and their positions. Whose constraints are these? Who would bear the cost of shifting them? Who has the power to shift them? Who is watching the mapping, and what will they make of it? The map shows not only the channels but the pilots, the passengers, the harbour authority and the people on the shore.

From Sunzi: the visible disposition. Every decision to wait, to "let the tide turn", is recorded as a gesture with likely readings. If we wait, who will notice, and what will they conclude? Waiting is treated as action, with consequences to be anticipated and sensed. The Compass and the Loom's Uncertain Compass, the compass that measures its own uncertainty, includes among its laws the Zeno law: over-frequent monitoring freezes change, so monitor less often than the change you hope to see. A visible leader's waiting is a form of monitoring, and it can freeze what it waits for.

From the children's party: the watched watcher. Every practitioner is sensed. Her presence, attention and interventions are part of what is sensed about the system. The Tidal Loom's reading of the reader states it: convenors signify their own experience, published alongside everyone else's. The parent at the party is part of the party.

These supplements are modest. Each could be added to estuarine mapping as practised, and none requires abandoning the method's core insight about constraint energetics. Together they would make the method estuary-aware. It would use the estuary for what it pictures well and correct for what it pictures badly.

The Formula of the Title

The title of this book comes from this thesis. Humans are not estuaries does not mean that estuarine mapping is useless, that constraint energetics are unreal, or that timing does not matter. It means that the medium of human complexity, unlike the estuary, reads the reader, and that the reader of human complexity, unlike the pilot, is inside the medium. Anthro-complexity's slogan says humans are not ants, which refuses a misleading model of the agents in human systems. This book's title says humans are not estuaries either, which refuses a misleading model of the medium and of the practitioner's place in it.

The two refusals belong together. The first says humans narrate, intend and abstract, which ants do not. The second says that because humans narrate, intend and abstract, the medium they make up reads every description of it, including the practitioner's, and the practitioner is one of them. The first refusal is anthro-complexity's. The second is the one its own logic requires.

Verdict

Thesis III: humans are not estuaries. holds

The governing images of anthro-complexity's recent practice (estuary, tide, sailor, surfer, river, waterfall) share two properties human systems lack: the medium is not reflexive, and the reader stands outside the reading. Jullien's 勢 includes the strategist in the configuration, so the surfer is the wrong image for human propensity. In human affairs waiting is a gesture, and the tide may turn because one waited. Snowden's earlier image, the children's party, had neither property, and the movement from party to estuary matches the drift traced in Chapter 6. Jacobs's city, with its eyes on the street, is a human root metaphor of organised complexity that anthro-complexity could adopt. The adversary's reply, that the estuary is a heuristic, is right in fact and does not meet the thesis, which concerns the method built on the heuristic and not the heuristic's acknowledged limits.

THE TIDE DOES NOT READ THE TIDE-TABLE —

THE ORGANISATION READS THE MAP —

HUMANS ARE NOT ESTUARIES.

Addition: Eyes on the Organisation

The chapter proposed Jane Jacobs's city, with its eyes on the street, as a human root metaphor of organised complexity that anthro-complexity could adopt in place of, or alongside, the estuary. This addition develops what a practice built on Jacobs's metaphor would look like, by reading four of her central observations about cities as principles for organisations.

1. Eyes on the street

Jacobs's most famous observation is that city streets are kept safe not primarily by police but by the natural surveillance of people who use them: shopkeepers watching their frontage, residents looking out of windows, passers-by going about their business. A street with many eyes on it is safe, because misbehaviour is seen. A street with few eyes, such as an empty plaza or a corridor between tower blocks, is dangerous, however many police patrol it. And the eyes are not watchers in the sense of observers outside the street. They are people on the street, part of its life, whose watching is incidental to their being there.

For organisations. An organisation's capacity to notice what is going wrong, its weak-signal detection, lies primarily not in central monitoring but in the many eyes of people who work there and notice things in the course of their work. A narrative instrument is, at best, a way of collecting what those eyes have seen. At worst it is a substitute for them, a central patrol that makes the many eyes seem unnecessary. Jacobs would say: protect and multiply the eyes. Make sure people are present where things happen, that they have reason to notice, and that what they notice is heard. The instrument is one way of hearing, and it is not the eyes themselves.

2. Mixed primary uses

Jacobs argued that a lively, safe and economically healthy district needs mixed primary uses: residential, commercial, cultural and industrial uses in the same area, bringing different people at different times of day for different purposes. Single-use zones, such as office districts empty at night or residential estates empty by day, are dead and dangerous. Mixing is what makes a district work.

For organisations. Organisations are often zoned: functions separated, levels stratified, sites specialised. Jacobs's principle suggests that the organisation's capacity for noticing and for emergent coordination depends on mixing, on the presence of different people with different purposes in the same spaces and conversations. A narrative instrument deployed in a zoned organisation will find zoned patterns, each function telling its own stories. The patterns will be read as facts about functions when they are traces of the zoning. A practice built on Jacobs's metaphor would attend to where mixing happens and where it does not, and would treat the absence of mixing as a constraint (Thesis IV).

3. Small blocks

Jacobs argued that short city blocks are better than long ones, because they multiply the routes people can take, the corners where they meet and the opportunities for chance encounter. Long blocks channel movement into a few routes and reduce encounter.

For organisations. An organisation's "blocks" are the units within which people work and between which they rarely move. Long blocks, large self-contained departments, reduce the encounters through which knowledge spreads and coordination emerges. Short blocks, smaller units with many points of contact, multiply them. This is Hutchins's overlapping horizons (Chapter 31) in urban form. A practice built on Jacobs's metaphor would look for the organisation's long blocks and ask whether shortening them, by creating corners where people from different units meet, would do more for noticing than any instrument.

4. The need for aged buildings

Jacobs argued that a district needs a mix of buildings of different ages, including old ones, because old buildings have lower rents and so house the new enterprises, marginal uses and experiments that cannot afford new buildings. A district of only new buildings excludes the new.

For organisations. An organisation needs spaces where experiments can happen cheaply: pockets of low monitoring, low overhead and low expectation, where the new can form before it is ready to be seen (Chapter 40, the decoherence law). Organisations that have rebuilt everything to the latest standard, with every process optimised, every activity measured and every space accountable, have no aged buildings, and no room for the marginal and the new. The weak signals that narrative instruments look for begin in the aged buildings. A practice built on Jacobs's metaphor would protect them.

Jacobs against the planner

Jacobs's book was an attack on the planners of her day: Le Corbusier's Radiant City, Ebenezer Howard's Garden City and Robert Moses's expressways and urban renewal. Each, she argued, imposed a plan on the city from above, based on a vision of what a city should be, and destroyed the organised complexity that made real cities work. The planners presumed wholes (the neighbourhood unit, the superblock) and designed parts to fit. They read the city from maps and statistics and did not walk its streets. They treated the city as a problem of simplicity or disorganised complexity, not organised complexity.

Jacobs's alternative was not a better plan. It was attention to the city's own processes and policies that supported them. A city, she insisted, cannot be a work of art. It is a living process that planners can help or hinder but cannot design. That is anthro-complexity's Pelagian stance (Element 20), its critique of systems thinking's presumed wholes (Element 2) and its principle of working with propensity (Element 7), stated sixty years earlier, about cities.

But Jacobs's alternative had something the estuary does not: the observer inside. Jacobs walked the streets of Greenwich Village. She lived there. She watched the sidewalk ballet, the daily choreography of people coming and going, from her own window and her own doorstep. Her knowledge of the city was a resident's, not a planner's or a pilot's. She was the fisherman in the maelstrom, the friend in the lab, the wayfarer, the cook with the knife in the ox. And her book's authority came from that position. She knew the city because she was in it.

The metaphor compared

The estuary Jacobs's city
Medium Water: indifferent to being read People: read one another and respond
Reader Pilot, on deck Resident, on the street
Safety Knowing the channels Eyes on the street
Health Reading the tide Mixed uses; small blocks; aged buildings
Failure mode Running aground The planner's vision imposed from above
Practitioner's knowledge Chart and timing Walking, watching, living there

EYES ON THE STREET, NOT POLICE ON PATROL —

MIX THE USES — SHORTEN THE BLOCKS — KEEP THE OLD BUILDINGS —

AND KNOW THE CITY BY LIVING IN IT.

Addition: Thesis III After Book Nine, and Its Argument in Cards

Snowden now writes that nobody stands outside, the practitioner least of all. The thesis stands for the images and the instruments. The Zen Garden study gives it an image: fifteen stones, and from any seat one is hidden. See Chapters 52 and 53.

The three cards below set out the thesis in full: its claim and premises, the evidence with the strongest objection and the reply, and the consequence with a test and the verdict as it now stands.

Chapter 22

Thesis IV: The Under-Socialised Constraint

3,521 words, about 15 minutes of reading

Thesis IV: the under-socialised constraint. Anthro-complexity's theory of constraint, built with Juarrero, distinguishes governing from enabling constraints and adds constructors from physics. But the constraints that make a human community what it is are constitutive. They are reproduced in the act of complying with them, and in the act of mapping them. Constructor theory, imported to supply a class of actants, is a physics of transformation without change to the transformer. In human affairs everyone is a reagent, so the import cancels the human distinction it was meant to serve. (Theses 10 and 11.)

The Achievement

This thesis must begin with an acknowledgement, because its target is anthro-complexity's greatest achievement. The theory of constraint developed by Alicia Juarrero, and taken up and extended by Snowden, is among the most important contributions to complexity thinking of the past generation. It replaces the efficient-causal picture of management (actions cause outcomes) with a constraint-based picture: actions change constraints, and constraints change the probabilities of outcomes. It explains how coherence arises in complex systems without design, through context-sensitive constraints that link events into patterns. It explains how downward causation works without mysticism. A pattern at a higher level constrains events at a lower level by changing the probabilities of their occurrence, not by exerting a force. It gives practice a target. You cannot change outcomes directly, but you can change constraints, and estuarine mapping is a method for finding which ones.

None of this is in dispute. The thesis concerns what the theory of constraint, as anthro-complexity uses it, leaves out of the human case, and what one of its imports, constructor theory, brings in that does not belong.

Juarrero's Constraints

Juarrero's Dynamics in Action (1999) and Context Changes Everything (2023) develop a philosophy of constraint whose core distinctions can be stated briefly.

A constraint is anything that changes the probability distribution of events, making some more likely and others less. Constraints are not forces. They do not push. They shape the space of possibilities within which pushing happens. A riverbank constrains the flow of water. A grammar constrains the formation of sentences. A habit constrains behaviour.

Context-independent constraints narrow possibilities without regard to context. A wall prevents passage whatever is happening on either side. Context-sensitive constraints make the probability of one event depend on another. In a language, the probability of a word depends on the words before it. In a cell, the probability of a reaction depends on which enzymes are present. Context-sensitive constraints let events become interdependent, and interdependence is how coherent patterns, which are more than collections of independent events, arise.

Snowden's practice vocabulary, developed with Juarrero, distinguishes governing constraints, which limit and channel, like rules, procedures and boundaries, from enabling constraints, which make new things possible, like a shared language, a common protocol or a trellis. Constraints vary in strength (rigid, elastic, permeable) and in energy cost to change. Estuarine mapping locates them on these dimensions.

The Missing Kind: Constitutive Constraint

The distinction

The Compass and the Loom, in its second book, identified a gap in this vocabulary. Beside governing and enabling constraints there is a third kind, constitutive constraint: the constraint that makes a community what it is.

The distinction has a well-established precedent in the philosophy of rules. John Searle, in Speech Acts (1969), distinguished regulative from constitutive rules. Regulative rules regulate activities that exist independently of the rules. "Drive on the left" regulates driving, which exists whether or not there is a rule about which side. Constitutive rules create the activities they govern. The rules of chess do not regulate a pre-existing activity of moving pieces of wood around a board. They constitute chess. Without the rules there is no chess, only wood. Searle's formula for constitutive rules, X counts as Y in context C, is the formula Chapter 13 used for status functions, and constitutive rules are how status functions are created.

Map this onto Juarrero's vocabulary. Governing constraints are regulative: they limit and channel activities that exist apart from them. Enabling constraints are partly constitutive: a shared language enables communication that could not happen without it. But the constraints that constitute a community, its identity, its sense of who belongs and what membership means, its implicit understanding of "how we do things here", go beyond enabling. They do not just make new things possible. They make the community what it is. Remove them, and the community does not lose some possibilities. It ceases to exist as that community.

Constitutive constraints are reproduced in action

The crucial property of constitutive constraints in human communities is that they are reproduced in the act of complying with them. This is Anthony Giddens's duality of structure, the central idea of structuration theory. Structure is both the medium and the outcome of the practices it organises. When a person speaks English, she draws on the rules of English, the medium, and in doing so reproduces them, the outcome. There is no English apart from its being spoken. When members of an organisation follow its unwritten norms, they draw on the norms and reproduce them. Norms that were never followed would not be norms.

Bourdieu's habitus makes the same point at the level of the body. The constitutive constraints of a social field are carried in its members as dispositions, ways of seeing, moving, speaking and judging, that are acquired through participation and reproduced in every act. Bourdieu's doxa is the domain of what goes without saying: the constitutive constraints so deeply embodied that they are not experienced as constraints at all, but as the nature of things. The Compass and the Loom summarised this: habitus and doxa are the mechanisms by which a club's rules are carried in members, not on paper. And Wittgenstein's remarks on rule-following, read through Giddens, add that rules show themselves as "this is what we do". A rule is not an instruction that is consulted and applied. It is a practice that is carried on.

Constitutive constraints are reproduced in the act of mapping them

Here the thesis bites on estuarine mapping. If constitutive constraints are reproduced in the act of complying with them, they are also reproduced, or changed, in the act of mapping them. An estuarine mapping session is a social practice. It has its own constitutive constraints: who convenes, who speaks first, whose sense of what is "fixed" goes unchallenged, what counts as a constraint and what is taken as given, how disagreements about energy cost are resolved. These are the constitutive constraints of the organisation, carried into the room by its members. The session reproduces them.

Consider a concrete case. In an estuarine mapping session, a senior manager says that the annual budget cycle is a constraint that "cannot be changed" and places it in the zone of high energy cost. No one disagrees. The constraint is mapped as fixed. But was it fixed? Perhaps the budget cycle could have been changed by a decision of the finance director, at modest cost, if anyone had proposed it. It was mapped as fixed because the senior manager said so, and because the organisation's constitutive constraints include deference to senior managers on matters of finance. The map has recorded not the budget cycle's energy cost but the organisation's habitus, the deference that makes senior managers' judgements about what is fixed go unchallenged. And by recording it as a fact about the budget cycle, the map has reproduced the deference, now with the authority of a method behind it.

The Compass and the Loom's principle dated limits (record each "cannot be changed" with a date and condition) is a partial remedy. It makes the "cannot" provisional, open to revisiting. The full remedy would map the constitutive constraint alongside the governing one: not only "the budget cycle is fixed" but "the judgement that the budget cycle is fixed was made by the senior manager and not challenged". The second is the constitutive constraint. It is often the more important of the two, and it is reproduced in the act of mapping unless the mapping is designed to see it.

Constraint Closure

There is a second sense in which anthro-complexity's constraint theory, as practised, under-socialises human constraints. Constraints in a living system are not independent. They maintain one another. Maël Montévil and Matteo Mossio, in "Biological organisation as closure of constraints" (2015), developed the concept of constraint closure: a set of constraints such that each is produced or maintained by others in the set. In a cell, enzymes constrain reactions, the reactions produce the components of membranes, the membranes constrain the concentrations in which enzymes work, and the concentrations constrain the production of enzymes. No constraint is fundamental. Each is held in place by the others. Constraint closure is, on this account, what distinguishes biological organisation from mere physical order. A living system is a set of mutually maintaining constraints.

The Compass and the Loom drew the contrast with WRAS's "substrate": constraint closure means each constraint is maintained by others in the set, which is the opposite of a substrate. A substrate is a layer beneath, which supports what is above and is not maintained by it. Constraint closure has no beneath. Everything is maintained by everything else.

Human communities exhibit constraint closure in a stronger form than cells, because human constraints include descriptions of the constraints, and descriptions maintain constraints. The budget cycle is maintained by the finance department's procedures, which are maintained by the board's expectations, which are maintained by the audit committee's descriptions of good governance, which are maintained by the regulator's codes, which are maintained by public descriptions of what corporate accountability requires. Each is held in place by the others, and several are held in place by descriptions. Change the description of good governance and the chain loosens.

An estuarine map that places constraints individually, each with its own energy cost and timescale, represents them as independent features of a landscape, like sandbanks each with its own depth. Under constraint closure, constraints are not independent, and the energy needed to change one depends on whether the others that maintain it can be loosened. The "energy cost" of a constraint is not a property of the constraint. It is a property of its position in the closure. A map that does not show the closure, which constraints maintain which, shows energies that will be wrong whenever the closure shifts.

Constructors: The Import That Does Not Fit

Constructor theory

The November post names constructors as a second kind of actant, after constraints:

In the post's own gloss, constructors, as Deutsch and Marletto use the term, are entities that bring about a transformation and keep the capacity to do so again, as a catalyst does, or an established ritual.

Constructor theory, developed by David Deutsch from around 2012 and extended by Chiara Marletto (whose The Science of Can and Can't, 2021, is its accessible statement), proposes a new mode of explanation in fundamental physics. Instead of explaining phenomena by initial conditions and laws of motion, which says what will happen, constructor theory explains them by which transformations are possible and which impossible, which says what can happen. Its basic unit is the task, a transformation of an input into an output, and its basic entity is the constructor: an entity that can cause a task to be performed and that retains the ability to cause it again. A catalyst is a constructor for a chemical reaction. It enables the reaction and comes out unchanged, ready to enable it again. A factory is a constructor for its products. A living cell is a constructor for its own replication.

Constructor theory's key move is to make the constructor's unchanged persistence definitional. Something that brought about a transformation and was used up or changed in doing so would be a reagent, not a constructor. The theory then asks what the laws of physics must be for constructors to be possible, and argues that the answer illuminates information, thermodynamics, life and knowledge. Deutsch and Marletto argue that knowledge, in particular, is information that acts as a constructor. It causes transformations (technologies, actions) while being preserved.

The import into human affairs

Snowden's example of a human constructor is "an established ritual". A wedding marries, a graduation graduates, a handover hands over, and the ritual remains available to do so again. The import is attractive. It gives anthro-complexity a class of actants that are neither constraints (which shape probabilities) nor agents (which intend), but persistent transformative capacities. Rituals, routines, institutions and practices can be understood as constructors: they bring about transformations, marrying, graduating, handing over, and persist to do it again.

Why it does not fit

The import does not fit, for a reason that strikes at the heart of the human distinction. Constructor theory defines constructors by their unchanged persistence, while anthro-complexity defines humans by their capacity to be changed by the stories they tell, the identities they move between and the abstractions they make. A human ritual is not a catalyst. It is changed by every performance.

Victor Turner's anthropology of ritual, in The Ritual Process (1969) and From Ritual to Theatre (1982), showed that rituals are not fixed scripts repeated identically. They are performances, in which the script is interpreted, adapted and transformed by the performers. Each wedding is a variation on the form of a wedding, and the variations accumulate. Weddings in 2026 are not weddings in 1926, and the change came through performances, each slightly different. Turner's concept of liminality, the betwixt-and-between phase of a rite of passage in which ordinary structures are suspended, showed that rituals contain within themselves the possibility of transformation. In the liminal phase the participants experience communitas, a bond outside the ordinary structures, and the ritual's form can be renegotiated. Max Gluckman's work on rituals of rebellion showed something subtler. Rituals that seem to challenge the social order, the licensed mockery of rulers or the temporary inversion of hierarchy, can protect the frame they appear to challenge, by channelling dissent into a bounded performance. But the protection is achieved through change: the ritual's meaning shifts as the society around it shifts, and a ritual that protected the order in one generation may subvert it in the next.

A ritual, in short, is not a constructor. It is a practice reproduced and transformed in performance, which is Giddens's duality of structure again. And its participants are not catalysts. They are reagents, changed by the transformation they perform. The bride and groom are married. The graduate is graduated. The officiant, too, is changed by having officiated, in authority, experience and relationship to the community. Everyone is a reagent, as The Compass and the Loom puts it: every participant is changed by acting.

Proposition granted: A constructor is an entity that brings about a transformation while retaining the capacity to do so again, unchanged.

Then it must be granted that a human ritual is a constructor only if it is unchanged by its performances. — Granted, by definition.

Then it must be granted that human rituals are changed by their performances. — Granted. Turner, Gluckman and the history of every ritual attest it.

Then it must be granted that human rituals are not constructors in Deutsch and Marletto's sense. — Granted.

Then the respondent may reply that "constructor" is used loosely, to mean a persistent transformative capacity that is approximately unchanged over a timescale of interest. — Granted, as a reply. But then the import is not constructor theory. It is the familiar sociological observation that institutions persist and are reproduced, which Giddens, Bourdieu and Turner explain better, because they explain the change as well as the persistence.

Then it must be granted that the import adds to anthro-complexity's ontology a model of action in which the agent is unchanged by acting, and that this model contradicts the human cluster, in which humans are changed by their stories, identities and abstractions. — Granted. The import cancels the human distinction at the point where it is introduced.

What constructor theory could do

The thesis does not claim that constructor theory has nothing to offer anthro-complexity. Its basic move, explaining by what is possible and impossible instead of by what will happen, is congenial to complexity thinking. Its account of knowledge as information that acts as a constructor is suggestive for the role of narratives and abstractions. Marletto's emphasis on "can and can't" resonates with estuarine mapping's question of which constraints can be shifted.

The best use of constructor theory in anthro-complexity would be at the level of possibility, not of actants: as a reminder that the important question is often not "what will happen?" but "what is possible here, and what is not?" That use does not require classifying rituals as constructors, and so does not import the unchanged agent. It asks what transformations the current constraints make possible and impossible, which estuarine mapping already asks, and it gives the question a foundation in physics. The import fails only when it is used to classify human practices as constructors, because that classification brings the unchanged persistence into a domain where nothing persists unchanged.

Consequences for Estuarine Mapping

The thesis has three consequences for estuarine mapping, the method in which anthro-complexity's constraint theory is most fully embodied.

Gap What current practice maps What a socialised map would add
Constitutive constraints Governing and enabling constraints, each with energy and time The constitutive constraints reproduced in the mapping itself: who judged this fixed, and was the judgement challenged?
Constraint closure Constraints as independent features Which constraints maintain which; energy as a property of position in the closure, not of the constraint
Constructors Rituals and practices as persistent transformers Rituals and practices as performances that change their performers; who is changed by this ritual, and how?

Each addition is feasible. The first is close to The Compass and the Loom's dated limits principle, extended to record who judged each limit and whether the judgement was challenged. The second could be represented on the map by links between constraints, or by a simple question for each: "what holds this in place?" The third would replace "constructor" with "practice", and ask of each practice not only what it does but what it does to those who perform it.

The adversary will say that estuarine mapping already does much of this in practice. Facilitators are trained to challenge claims that constraints are fixed. Groups argue about energy costs, and the arguments surface the deference and politics the thesis describes. The map is a starting point for conversation, not an authoritative representation, and the conversation is where the constitutive constraints become visible. As for constructors, Snowden's use is plainly loose: he means persistent transformative practices, not catalysts in the strict physical sense, and no practitioner thinks a ritual is unchanged by performance.

The adversary wins half. Good facilitation does surface constitutive constraints, and the map's value as a conversation-starter is real. But good facilitation is a quality of the facilitator, not of the method. The thesis concerns the method, what it represents and what it leaves out, because methods travel to facilitators who are not good, and the method's representations then stand without the conversation that should have qualified them. A map that represents the budget cycle as fixed, without representing the deference that made it so, will be read that way by anyone who sees the map without having been in the room. As for constructors, the loose use is the problem. A term imported from physics with a precise definition, used loosely, brings its precision as rhetorical authority and leaves its definition behind. That is the eclecticism Pepper warned against, and that Snowden himself diagnoses in Ing's absorption of complexity's vocabulary without its ontology. partial

Verdict

Thesis IV: the under-socialised constraint. holds

Anthro-complexity's constraint theory is a major achievement, and its governing/enabling distinction and energy-cost analysis are sound. It under-socialises human constraints in three ways. It lacks constitutive constraints (Searle's constitutive rules, Giddens's duality of structure, Bourdieu's habitus), which are reproduced in the act of complying with them and in the act of mapping them. It represents constraints as independent where they exhibit closure, so that energy cost is a property of position and not of the constraint. And it imports constructors from physics, whose definitional unchanged persistence contradicts the human cluster's insistence that humans are changed by acting. Rituals are performances, not catalysts, and everyone is a reagent.

THE CATALYST LEAVES UNCHANGED —

THE CELEBRANT DOES NOT —

IN HUMAN AFFAIRS EVERYONE IS A REAGENT.

Addition: Thesis IV After Book Nine, and Its Argument in Cards

The trialectic says the practitioner's formation is part of the work, which weakens the charge about the constructor. The post on Luhmann treats decision premises as actants, and its example, that nothing is decided until a senior person signs, is a constitutive constraint and a positional one. See Chapters 48 and 52.

The three cards below set out the thesis in full: its claim and premises, the evidence with the strongest objection and the reply, and the consequence with a test and the verdict as it now stands.

Chapter 23

Thesis V: The Missing Term Is Power

5,029 words, about 22 minutes of reading

Thesis V: the missing term is power. The three marks of the human (multiple identities, intention, abstraction) are capacities, and none of them is power. Snowden's power critique of eurhythmia, that talk of restoring balance tends to favour whoever defines it, applies with equal force to whoever defines the signifier set, receives the pattern and holds the residual risk. A theory of human complexity without power is CAS with better adjectives. (Theses 12, 13 and 14.)

The Sentence That Opens the Door

The November post contains one sentence about power, and it is a very good sentence. It comes in the section on eurhythmia:

Balance, Snowden writes, is a helpful metaphor for a body and a risky one for an organisation, because talk of restoring balance tends to favour whoever defines it.

The sentence identifies a mechanism of power exactly. A norm (balance, health, eurhythmia) is presented as a neutral standard against which a system's state can be measured. But in a human system the norm is not given by nature. It is defined, by someone, and the definition encodes the definer's interests. The appeal to "restoring balance" then presents the definer's interests as the system's health, and any departure from them as disorder to be corrected. Whoever defines balance wins every argument about what is out of balance.

This is a power critique of the first rank, and it is the kind Michel Foucault made central to twentieth-century social thought. Norms are instruments of power. Normalisation, the production of a standard against which individuals are measured and corrected, is one of modern power's characteristic operations. The critique applies to Ing's eurhythmia, as Snowden says. This chapter argues that it applies equally to anthro-complexity's own instruments, and that the human cluster's silence about power is what prevents the application.

The Capacities and the Missing Term

Chapter 4 set out the human cluster: humans hold multiple identities, act with intentionality, and create abstractions and shared narratives that change constraints; the asymmetry principle; entanglement without flat ontology; and the rejection of the programme. Every element is about a capacity that humans have, or a relation between humans and non-humans. None is about relations among humans in which some have more capacity than others to define, decide and impose.

Yet the third mark, abstractions and narratives that change constraints, is unintelligible without power. Chapter 13's analysis through Searle showed that abstractions change constraints only when they are collectively accepted. Collective acceptance is not distributed equally. A chief executive's declaration that "we are now an agile organisation" is far more likely to be accepted, and to create deontic powers, than a cleaner's declaration of the same thing. Not because the chief executive's abstraction is better, but because she occupies a position whose declarations are accepted. The capacity to change constraints by narrative is a human capacity. Its exercise is a function of position in a field of power.

Snowden knows this. The concern with entrainment that runs through his work, groups falling in behind the first confident voice, is a concern with power. So is his long-standing practice of disintermediation, removing the expert who stands between people and their own meaning. So is his method of ritual dissent, in which a group presents a proposal and the presenters then turn their backs while others criticise it, so that criticism can be voiced without the social cost of face-to-face confrontation. These are power-aware practices. But power is not in the theory of the human, the cluster that defines anthro-complexity. It appears in practice as something to be managed, not in ontology as something that makes human systems what they are.

Four Theories of Power

To see what the human cluster leaves out, four theories of power are needed. Each identifies a mechanism that bears on anthro-complexity's instruments.

Lukes: three dimensions

Steven Lukes's Power: A Radical View (1974; second edition 2005) distinguished three dimensions of power.

The first dimension, associated with Robert Dahl, is power in decision-making. A has power over B to the extent that A can get B to do something B would not otherwise do. Power is visible in observable conflicts over issues, and it is measured by who wins.

The second dimension, associated with Peter Bachrach and Morton Baratz (1962), is power in non-decision-making or agenda-setting. Power is exercised not only in winning decisions but in keeping issues off the agenda, so that some conflicts never surface. Whoever controls the agenda controls what can be decided.

The third dimension, Lukes's own contribution, is power in shaping preferences. The most effective power prevents conflict from arising at all, by shaping people's wants, beliefs and perceptions so that they accept their situation, either because they cannot imagine an alternative or because they see it as natural or beneficial. Lukes asked whether the supreme exercise of power is not precisely to get others to have the very desires one wants them to have.

Foucault: power/knowledge and visibility

Michel Foucault's analyses, in Discipline and Punish (1975) and The History of Sexuality (1976), made power productive as well as repressive. Power does not only forbid. It produces knowledge, subjects, categories and norms. Its characteristic modern form, disciplinary power, works through observation, normalisation and examination. Individuals are made visible, compared with a norm and corrected. Foucault's central image was Bentham's panopticon, a prison in which a central tower allows a guard to observe every cell without being seen, so that prisoners, never knowing whether they are being watched, discipline themselves. Visibility is a trap, Foucault wrote. The power is in the asymmetry of seeing. Those who are seen are subject to those who see.

Foucault's later concept of the dispositif (apparatus) generalised the analysis. A dispositif is a heterogeneous ensemble of discourses, institutions, architectures, regulations, laws, administrative measures and scientific statements, together with the relations among them, that serves a strategic function. The Compass and the Loom summarised it: a dispositif moves power into the arrangement of positions. Power is not held by individuals. It is in the arrangement.

Bourdieu: symbolic power

Pierre Bourdieu's concept of symbolic power, developed in Language and Symbolic Power (1991) and elsewhere, is the power of constructing reality, of imposing the categories through which the social world is perceived. Symbolic power is the power to name, classify and make distinctions that are then taken as natural. It is exercised most effectively when it is misrecognised, when those subject to it see the categories as simply describing how things are, not as impositions. Professions, states and schools exercise symbolic power through their authorised classifications: diagnoses, legal categories, grades. Bourdieu called the state the holder of the monopoly of legitimate symbolic violence.

Status characteristics

The fourth theory comes from experimental social psychology, the status characteristics and expectation states tradition founded by Joseph Berger, Bernard Cohen and Morris Zelditch in the 1960s and 1970s. Its central finding is that in task-oriented groups, members' external status characteristics (gender, race, age, occupation, education) shape the expectations others form about their competence. Those expectations shape who speaks, who is listened to, whose suggestions are adopted and who is credited, even when the status characteristics are irrelevant to the task and even when formal roles are equal. The Compass and the Loom summarised it: background status shapes who is heard, whatever the formal roles. The finding has been replicated across hundreds of studies. It is one of the most robust in the sociology of small groups.

Jo Freeman's essay "The Tyranny of Structurelessness" (1970, published 1972) made the political point that follows. Groups that abolish formal structure in the name of equality do not abolish hierarchy. They drive it underground, where informal elites based on friendship, confidence and status form without accountability. The Compass and the Loom again: unstructured groups grow unaccountable informal elites.

Theory Mechanism Where power lies
Lukes, 1st dimension Winning decisions In observable conflict
Lukes, 2nd dimension Setting agendas In what is never decided
Lukes, 3rd dimension Shaping preferences In what is never wanted
Foucault Visibility, normalisation, dispositif In the arrangement of seeing and being seen
Bourdieu Symbolic power; naming; classification In the categories taken as natural
Status characteristics Expectations from external status In who is heard, whatever the formal roles
Freeman Structurelessness hides informal hierarchy In the absence of declared structure

The Theories Applied to the Instruments

Each theory identifies a mechanism, and each mechanism operates in anthro-complexity's instruments as deployed. The application does not accuse anyone of exercising power badly. It shows that power is exercised, through the instruments' structure, whatever the intentions of those who use them.

The signifier set: Lukes's second and third dimensions

A SenseMaker signifier set is an agenda. It specifies the dimensions on which narrators may signify their stories: the triads (trust, compliance, fear; or autonomy, mastery, purpose), the dyads (too much control to too little), the multiple-choice questions. Whatever is not on the signifier set cannot be signified. A narrator whose story is about exploitation, when no signifier names exploitation, must place her story on dimensions that do not fit, or leave the signification incomplete. The signifier set is Lukes's second dimension. It sets the agenda of what can be said, and so keeps some issues off it.

The signifier set also operates in Lukes's third dimension. Signifiers, once used, become categories through which narrators understand their own experience, which is the looping effect of Chapter 14. A workforce that has been asked repeatedly to place its stories on a triad of "trust, compliance, fear" begins to sort its experience into those categories. Experiences that do not fit (solidarity, boredom, exploitation, humour) become harder to notice, because the frame through which experience is reflected has no place for them. The signifier set shapes not only what is said but what is noticed, and what is noticed shapes what is wanted. Lukes would call this the supreme exercise of power, and it happens whether or not the designers intend it.

The Compass and the Loom identified the mechanism precisely. Triads act as search templates and variety reducers (Ashby), and the designer's categories determine what is visible. The Tidal Loom's principle frames, not seers proposes rotating the search template, so that different people see different gorillas. No single signifier set gets to define the agenda.

The dashboard: Foucault's visibility

A SenseMaker dashboard makes a population visible. Its patterns, clusters and outliers show who is telling which stories, in which departments, at which levels. The population is seen. Who sees? Typically, and first, the client who commissioned the study: leadership, or a central function acting for it. The population may be shown the patterns, and anthro-complexity's principles say it should be, but it is shown them after leadership, in a form leadership has approved, and usually without seeing what leadership did with them.

That is Foucault's asymmetry of visibility. Those who are seen are subject to those who see, and the dashboard is a panopticon in the precise sense that the population is visible to a centre that is not visible to it. The population does not know which patterns leadership attended to, what conclusions it drew, or what it decided. It knows only that it has been seen. Foucault's point about the panopticon was that this is enough. Prisoners who know they may be seen discipline themselves. A population that knows its narratives are visible to leadership will narrate accordingly, which is the degradation Thesis VIII examines.

Boisot's I-Space adds the mechanism of diffusion. Codification strips context and makes knowledge diffusible, and the diffusion runs towards those who hold the dashboard. The narrator's rich, contextual experience is codified into a signified data point, and the data point flows to the centre. The context does not.

Categorisation: Bourdieu's symbolic power

A Cynefin categorisation is an act of symbolic power. To say "this situation is complex" is to impose a category through which the situation will be perceived. Once accepted, the category carries deontic powers (Chapter 13): licences to probe, permissions to fail and release from the demand for plans. It also carries exclusions. A situation categorised as complex is not one in which an expert's analysis will be trusted, or in which a planner's timeline will be honoured. Whoever categorises exercises the power of constructing the situation's reality.

Snowden's insistence that categorisation be collective, done by groups and not by individual experts, distributes the act. It does not distribute the categories. The domains are Cynefin's. A group that categorises its situation collectively does so in Cynefin's terms, and Cynefin's terms carry Cynefin's distinctions: between ordered and unordered, between analysis and probe, between expertise and emergence. Bourdieu would say that the categories' authority is most effective when it is misrecognised, when the group sees itself as discovering what kind of situation it is in, rather than applying a framework that makes certain kinds of situation thinkable.

The mapping session: status characteristics and the first confident voice

An estuarine mapping session, a sense-making workshop or a probe-design meeting is a task-oriented group, and the status characteristics literature predicts that background status will shape who is heard. Chapter 22's example of the senior manager whose claim that the budget cycle was fixed went unchallenged is an instance. The status characteristics tradition predicts it, and Snowden's concern with the first confident voice describes it.

Snowden's methods for countering it (ritual dissent, anonymous contribution, small-group work, turning the presenters' backs) are among the best available, and they are informed by an understanding of power in groups. But they are methods for managing the effects of status within a session. They do not represent status in what the session produces. The map that emerges from a session does not show who placed which constraint, who challenged it, or whose placements went unchallenged. The status effects are managed in the room and erased from the record.

Distributed risk: the gradient

The Compass and the Loom concluded that "distributed" risk postures move exposure down the power gradient, and that risk lives in whoever is left holding it. The mechanism is simple. When risk is "distributed" by devolving decisions to teams, encouraging experimentation, accepting safe-to-fail probes and adopting a "distributed" posture in WRAS's terms, the decisions are distributed, and so are the consequences. But the capacity to bear consequences is not distributed equally. A senior manager whose distributed experiment fails has a career that survives it. A frontline worker whose team's experiment fails may not. Menzies Lyth's study of nursing, Social Systems as a Defence against Anxiety (1960), showed that organisational structures can function to protect their members from anxiety, and that the protection is often distributed unequally, with the most anxiety-producing work pushed to the least powerful. Distributing risk without distributing protection moves anxiety down the gradient.

The Instruments and the Theories

Instrument / practice Power mechanism Theory What the instrument does not represent
Signifier set Agenda-setting; preference-shaping Lukes 2 and 3 What could not be signified; how signifiers become categories
Dashboard Asymmetric visibility; diffusion to centre Foucault; Boisot Who sees first; what was done with what was seen
Cynefin categorisation Symbolic power; naming Bourdieu That the category is imposed, not discovered
Mapping session Status effects; first confident voice Berger and others; Freeman Who placed, who challenged, whose placements stood
Distributed risk Exposure moves down the gradient Menzies Lyth Who bears the consequences of distributed decisions
Pattern feedback Timing and form of return Foucault Whether the population saw what leadership saw, and when

Snowden Against Himself

The thesis can be put in Snowden's own terms. Talk of restoring balance tends to favour whoever defines it. Substitute:

  • Talk of sensing what people are experiencing tends to favour whoever defines the signifiers.
  • Talk of seeing patterns in narrative tends to favour whoever holds the dashboard.
  • Talk of categorising the situation tends to favour whoever supplies the categories.
  • Talk of mapping which constraints can be shifted tends to favour whoever's judgement of what is fixed goes unchallenged.
  • Talk of distributing risk tends to favour whoever can bear its consequences.

Each substitution is an application of the November post's own critique. Each identifies a point at which the instrument, like the norm of balance, is presented as neutral while encoding a position.

CAS With Better Adjectives

The thesis's strongest formulation is that a theory of human complexity without power is CAS with better adjectives. The phrase is deliberately provocative, and its meaning should be stated carefully.

Chapter 8 found that anthro-complexity's differences from CAS are of kind where they are relation-located (reflexivity, the interpretant, contested evaluation, the ethics of intervention), and of degree where they are agent-located (identity, intention, cognition). The human cluster, stated without power, describes humans as agents with richer capacities: multiple identities, open-ended intentions, constraint-changing abstractions. A CAS theorist can model agents with richer capacities, by giving them more rule sets, better anticipation and the ability to change parameters. What she cannot model, without changing the kind of model, is a system in which some agents define the categories through which others perceive, see others without being seen, set the agendas within which others decide, and push risk onto others who cannot refuse it. Those are relations of power. Their effects are not emergent properties of interacting agents with equal standing. They are structured asymmetries in the arrangement of agents, which shape what can emerge.

A human cluster that included power would be different in kind from CAS, not because its agents are richer but because its arrangement is asymmetric, and the asymmetry is reproduced by the instruments that describe it. A human cluster without power describes richer agents in a symmetric arrangement, which is CAS with better adjectives.

The adversary will say that power is not missing from anthro-complexity. It is everywhere in the practice: in disintermediation, ritual dissent, the critique of the systems-change industry's Augustinianism, the concern with entrainment, the warning about whoever defines balance. A theory need not name power in its definition of the human to be power-aware in its practice. Snowden's refusal to make power the master concept is deliberate. Theories that make power central (Foucault, Bourdieu, critical theory) tend to see power everywhere and find nothing else, and their practitioners become critics who cannot act. Anthro-complexity keeps power in view without letting it swallow everything.

The adversary wins half, and it is an important half. Anthro-complexity's practice is power-aware, more so than most management practice, and the danger of a theory that sees only power is real. CRP's Eliasian figurations and CST's emancipatory interest have both been accused of it. But the thesis does not ask that power become the master concept. It asks that power be in the theory of the human, as a dimension of human systems on a par with identity, intention and abstraction, so that it is in the instruments' design and not only in practitioners' vigilance. A theory that keeps power out of its definition of the human will build instruments that keep power out of their representations. The vigilance will then depend on the practitioner, which is the same gap as in Thesis II: the reflexivity of the author, not of the instrument. partial

What Power-Aware Instruments Would Represent

The thesis implies a set of additions to the instruments, parallel to those of Theses I, II and IV.

  • Signifier sets declare their agenda. Each states what it was designed to make sayable and what it leaves out. Narrators can propose additional signifiers, and the proposals are recorded.
  • Dashboards are mutual. The population sees what leadership sees, at the same time. Leadership's responses to the patterns, what it attended to and what it decided, are themselves sensed and shown.
  • Categorisations are recorded as acts. Who categorised, with what categories and with what consequences for whose authority.
  • Mapping sessions record who placed what. The map shows not only constraints but the judgements that placed them, and which judgements were challenged.
  • Risk is tracked along the gradient. For each probe or distributed decision, who bears its consequences, and whose anxiety it creates (the Tidal Loom's risk follows anxiety).

None of these additions makes power disappear. Power cannot be designed away. Each makes power visible in the record, so that its effects can be seen, discussed and, where possible, countered by the people subject to it. That is the minimum a theory of human complexity owes the humans it describes.

Verdict

Thesis V: the missing term is power. holds

The human cluster defines human distinctiveness by capacities and is silent on the asymmetric relations through which capacities are exercised. Snowden's critique of eurhythmia, that whoever defines balance wins, is a power critique of the first rank, and it applies to signifier sets (Lukes's agenda and preference-shaping), dashboards (Foucault's asymmetric visibility), categorisation (Bourdieu's symbolic power), mapping sessions (status characteristics, the first confident voice) and distributed risk (the gradient). Anthro-complexity's practice is power-aware, and the adversary's warning against power as master concept is apt. But power belongs in the theory of the human as a dimension, so that it is represented in the instruments' design and does not depend on the practitioner's vigilance alone.

WHOEVER DEFINES BALANCE —

WHOEVER DESIGNS THE TRIAD —

WHOEVER HOLDS THE DASHBOARD — WINS.

Addition: Power in a Narrative Study, Step by Step

The chapter applied four theories of power to anthro-complexity's instruments in general terms. This addition follows a single invented narrative study through its life, from commissioning to action, and marks at each step where power operates, by which mechanism, and what a power-aware design would add. The study is illustrative and corresponds to no real deployment.

The study

A regional health board commissions a narrative study of staff experience across its hospitals, after a run of critical inspection reports. The board's chief executive wants to know "what is really going on". A consultancy is engaged. Five thousand staff are invited to share stories.

Step 1: Commissioning

What happens. The chief executive and the board's director of workforce agree the study's purpose with the consultancy: "to understand staff experience and identify cultural risks."

Power. The purpose is set by the commissioners, and its wording frames what will count as a finding. "Cultural risks" are risks to the board, things that might produce further critical reports. They are not risks to staff, such as exhaustion, bullying or moral distress. Lukes's second dimension is at work: the agenda is set before any story is told.

Power-aware addition. Ulrich's boundary questions are asked at commissioning (addition to Chapter 12). Whose interests is the study for? Who is affected but not involved? Staff representatives, including the unions, are invited to co-author the purpose.

Step 2: Designing the signifiers

What happens. The consultancy designs a signifier set with the director of workforce. Triads include "compliance / initiative / avoidance" and "support from manager / support from peers / no support". Dyads run from "I feel safe to speak up" to "I do not".

Power. The signifiers encode the commissioners' interest in compliance and speaking up, the issues raised by the inspections. There is no signifier for workload, for staffing levels or for pay, which staff might name as the cause of the problems. Lukes's third dimension is at work: over repeated use, staff will learn to sort their experience into "compliance", "initiative" and "avoidance", and the categories will shape what they notice (Hacking's loop, Chapter 33).

Power-aware addition. The declination table is completed and published (Chapter 45). The fourth vertex is added, so staff can name what is missing. Staff representatives review the signifier set before deployment.

Step 3: Prompting

What happens. Staff receive an email from the chief executive inviting them to "share a story about a time you felt proud, or a time you felt worried, about the care in your hospital". The email states that responses are anonymous.

Power. The email comes from the chief executive. It makes the employee identity salient (self-categorisation, Chapter 13) and the chief executive the addressee (Bakhtin, Chapter 19). Staff who distrust the chief executive will shape their stories accordingly, and some will not respond. Those most at risk, junior staff, agency staff and staff from minority groups, may be least likely to trust the anonymity. Status characteristics (Chapter 23) predict whose stories will be told.

Power-aware addition. The prompt comes jointly from the board and staff representatives. It states who will see the stories, in what order, and what will be done with them. The address question is added: "to whom would you usually tell this story?"

Step 4: Collection

What happens. Two thousand stories are collected. Response rates vary. Senior nurses and consultants respond in high numbers, and porters, cleaners and agency staff in low numbers.

Power. The margin (Midgley, addition to Chapter 12) is forming. The least powerful groups are under-represented, and their absence will be read by the commissioners as noise. Their stories, the ones not told, are the missing mass (Chapter 24).

Power-aware addition. Response rates by group are reported as a finding, not a limitation. A Good–Turing estimate of the missing mass is made. Targeted collection through anecdote circles, conducted in person by trusted facilitators, is used for under-represented groups (addition to Chapter 6).

Step 5: Analysis

What happens. The consultancy's analysts produce patterns. Stories of "avoidance" cluster in two hospitals. Stories of "no support" cluster among night staff. The report highlights these as cultural risks.

Power. The analysts' reading is the first interpretant to be reported (Chapter 32). It is shaped by the signifiers, which encode the commissioners' interests, and by the analysts' relationship with the commissioner, who pays them. The patterns are read as facts about groups ("night staff lack support") rather than as traces of conditions ("night staffing levels are inadequate"). That is the established-and-outsiders risk (Chapter 28): the pattern describes the less powerful group's experience in terms that locate the problem in the group.

Power-aware addition. The established-and-outsiders lens is applied (Chapter 45). Before reporting, the analysts ask whether the patterns reflect power ratios. Night staff and staff in the two hospitals are shown the patterns first and asked what they mean.

Step 6: Reporting

What happens. The report goes to the chief executive and the board. A summary goes to staff two months later.

Power. The asymmetry of visibility (Foucault, Chapter 23) is complete. The board saw the full report first, and staff see a summary later, after decisions have been made. Staff do not know what the board concluded.

Power-aware addition. Mutual dashboards are used (addition to Chapter 39). Staff see the patterns at the same time as the board, and the board's reading and decisions are visible to staff, who can respond.

Step 7: Action

What happens. The board launches a "speaking up" programme in the two hospitals with clusters of avoidance stories, and a "manager support" training programme for night-shift managers.

Power. The actions address the commissioners' framing (speaking up, management) and not the conditions staff might have named (workload, staffing) had the signifiers allowed. Risk flows down the gradient (Chapter 16). Night-shift managers are sent on training, and the implicit message is that they were the problem. The night staffing levels are unchanged.

Power-aware addition. Return before action (Chapter 45): the proposed actions are put to the staff whose stories prompted them, who are asked whether the actions address what they described. Risk is tracked: who bears the consequences of each action, and whose anxiety does it create?

Step 8: The next round

What happens. A year later the study is repeated. Stories of "avoidance" in the two hospitals have fallen. The board concludes that the speaking-up programme has worked.

Power. The wake reader (Chapter 45) would ask whether the fall reflects change in experience or change in what staff will say, now that "avoidance" has been named as a problem and a programme has been run to fix it. Staff in the two hospitals know they were singled out, and some may now tell stories that show they speak up, whatever their experience. Thesis VIII's prediction 4, reflexive content, is the test.

Power-aware addition. Reflexive content is coded and reported separately. Signifiers are rotated, so that "avoidance" is not asked the same way twice. Night staff are asked directly whether the manager training changed anything.

Summary

Step Power mechanism Theory Power-aware addition
1 Commissioning Agenda set by purpose Lukes 2 Boundary questions; staff co-author purpose
2 Signifiers Categories shape noticing Lukes 3; Hacking Declination table; fourth vertex; staff review
3 Prompting Addressee and salience Bakhtin; self-categorisation; status Joint prompt; address question; stated use
4 Collection Margin forms Midgley; Good–Turing Report response by group; estimate missing mass; anecdote circles for the margin
5 Analysis Pattern locates problem in group Elias Established-and-outsiders lens; return first to groups
6 Reporting Asymmetric visibility Foucault Mutual dashboards
7 Action Risk down the gradient Menzies Lyth Return before action; track risk
8 Next round Reflexive content read as change Hacking; MacKenzie Wake reader; rotate signifiers

The study, as first described, is not badly designed by the standards of current practice. Every step is conventional, and many are done carefully. Power operates at every step anyway, by mechanisms that no one in the study intends, and the cumulative effect is that the study confirms the commissioners' framing and moves the burden of change down the gradient. The power-aware additions do not remove power. They make it visible at each step, so that those subject to it can see and contest it.

POWER AT THE COMMISSIONING — IN THE SIGNIFIERS — IN THE PROMPT —

IN THE MARGIN — IN THE READING — IN THE REPORT — IN THE ACTION — IN THE NEXT ROUND —

MAKE IT VISIBLE AT EVERY STEP.

Addition: Thesis V After Book Nine, and Its Argument in Cards

Power is now named by Snowden: the power to filter, its abuse, and epistemic injustice. It is not yet mapped. The Zen Garden study shows the gap as a missing axis: the Stacey matrix recorded agreement, and Cynefin records only knowability. Jaeggi adds that an abstaining practitioner hands the judgement of value to the sponsor. See Chapters 52, 53 and 54.

The three cards below set out the thesis in full: its claim and premises, the evidence with the strongest objection and the reply, and the consequence with a test and the verdict as it now stands.

Chapter 24

Thesis VI: An Anti-Computational Ontology with a Computational Practice

3,546 words, about 15 minutes of reading

Thesis VI: an anti-computational ontology with a computational practice. Anthro-complexity's ontology says the human brain does not run a programme. Its practice increasingly converts human noticing into machine-readable signal. The harvest of narrative into data is where the contradiction bites. Recursive learning on harvested narrative erases the tails of the distribution, which is where weak signals live, so a weak-signal instrument that feeds a model can starve itself. (Theses 15 and 16.)

Two Commitments

Anthro-complexity holds two commitments that are each well founded and are, in current practice, in tension.

The first is ontological, stated in Element 15 and examined in Chapter 15. Human cognition is not the execution of a specifiable programme. It is embodied, situated, pattern-based and socially distributed. Meaning is made between people, in participatory sense-making, and cannot be extracted without loss. The tacit–explicit–entangled trialectic names the loss. Entangled knowledge exists only in the interaction of people, artefacts and context, and disappears when the interaction is broken up.

The second is practical. Anthro-complexity's instruments work at scale, and work at scale requires computation. SenseMaker is software. Narratives are collected digitally, signified as coordinates on triads and dyads, stored in databases, aggregated statistically and displayed as patterns. More recent instruments, as The Compass and the Loom documents, go further, towards a human/AI hybrid in which human noticing is gathered as input for automated systems that detect patterns, flag anomalies and suggest responses.

The two commitments are not contradictory in principle. A theory can hold that humans are not computers and still use computers to help humans see patterns in their own experience. The question this thesis asks is what happens to the human, as the ontology describes her, when her noticing passes through the practice. The answer, it argues, is that the relational, entangled and tacit character of human meaning, which the ontology insists on, is exactly what the computational practice strips away.

Legibility

James C. Scott, whose mētis Snowden adds to his definition of a complex system, gives the first instrument for analysing the tension. Scott's Seeing Like a State (1998) is a study of legibility: the state's project of making its territory and population readable to a central administration. Pre-modern states could see very little of what they governed. Local measures varied, land tenure was customary and complex, names were not standardised, and the population was not counted. Modern states imposed legibility through cadastral surveys, standard weights and measures, permanent surnames, censuses, planned cities and scientific forestry. Legibility made administration possible. It also simplified, and the simplification destroyed what it could not see.

Scott's most vivid example is scientific forestry in eighteenth-century Prussia and Saxony. Foresters, seeking to make the forest legible as a source of timber revenue, replaced diverse, old-growth forest with monocultures of Norway spruce planted in straight rows. The first generation was spectacularly productive. The second generation failed: the monoculture had destroyed the soil ecology, the undergrowth, the insect and bird populations and the fungal networks that the old forest depended on, none of which had been visible on the foresters' tables. Waldsterben, forest death, followed. The legible forest was a dead forest.

Scott generalised the lesson. Every scheme of legibility selects certain features of a complex system, those that serve the administrator's purposes, and represents them in a simplified form. The scheme then acts on the system through the simplified representation, and the features that were not selected are damaged, often fatally, because the scheme does not know they exist. Mētis, the practical local knowledge that sustained the system, is the first casualty, because mētis is precisely what cannot be made legible.

Narrative Made Legible

SenseMaker is not high-modernist legibility. It does not impose a single standard measure. It keeps the narrator's own words and her own signification, and it was designed in explicit opposition to the survey instruments that reduce human experience to ratings. But it is a scheme of legibility nonetheless. It makes a population's experience readable to an analyst, through a representation that selects certain features and simplifies them.

What does it select? The signified features: where on each triad the narrator placed her story, which multiple-choice options she picked, which demographic categories she belongs to. These become the data points that are aggregated. The narrative text is kept, and can be read, but at scale it is the signified coordinates that form the patterns, because coordinates can be aggregated and texts, without further processing, cannot.

What does it leave out? What Chapters 15 and 19 identified:

  • The address. To whom the story was told, and what response it anticipated (Bakhtin).
  • The relations. Who else was in the story, and what their part in it meant to them (Elias, Mead).
  • The participatory sense-making. How the experience was made sense of in the interaction where it happened, before it was told (De Jaegher and Di Paolo).
  • The entangled knowledge. What was known in the interaction of people, artefacts and context, which cannot be extracted into a narrative at all (Snowden's own trialectic).
  • The tacit. What the narrator knows but cannot tell, which by definition does not appear in the narrative (Polanyi).
  • The mētis. The practical, local, experiential knowledge that the narrator uses but does not reflect on, which a prompt for a story about "a time when..." may not elicit (Scott).

The analogy with scientific forestry is not exact, and should not be pressed too hard. SenseMaker does not plant monocultures. But the structure is the same. A representation selects legible features. The representation is acted on, and the features it did not select (here the relational, entangled, tacit and mētic character of human meaning) are not available to the action. If the action changes the conditions in which narratives are made, through interventions, probes or feedback, it changes them on the basis of the legible features alone. Scott's lesson is that this is how complex systems are damaged by those who meant to help them.

The Harvest

The second stage of the tension is the move from legibility to harvest: from making human noticing readable to an analyst, to making it usable as input for automated systems.

The Compass and the Loom traced what it called the harvest loop in the instruments it examined. The deliverables turn human noticing into training data for automation, so that human "superpowers" are captured in order to be automated. The brochure's language of human-AI collaboration, it argued, conceals a one-directional flow. Humans notice and the system learns from their noticing, so that it can notice without them.

Behavioural surplus

Shoshana Zuboff's The Age of Surveillance Capitalism (2019) gives the harvest loop a political-economic analysis. Zuboff argued that the large technology platforms discovered that the data generated by users' behaviour (their searches, clicks, movements and pauses) contained a surplus beyond what was needed to improve the service. The surplus could be used to predict and shape future behaviour, and the predictions could be sold. She called the surplus behavioural surplus, and the economic logic built on it surveillance capitalism. Its key features are that the raw material, human experience, is taken without meaningful consent; that the products, predictions, are sold to third parties and not to the people whose experience produced them; and that the ultimate aim is not only prediction but modification, shaping behaviour so that it becomes more predictable.

The harvest loop in narrative instruments is not surveillance capitalism. The scale, the business model and the intent are different. But the structure has points of contact that a theory of human complexity cannot ignore. Human experience, in the form of narratives and noticings, is the raw material. It produces a surplus beyond what the narrator intended: patterns, signals, training data. The surplus flows to those who commissioned the instrument and, in the hybrid case, to the systems they deploy. And questions of consent, ownership and benefit are, as The Compass and the Loom found, absent from the instruments' brochure vocabulary.

Chapter 16 raised the consent question. Consent to share a story is not consent to what is done with it. The harvest loop sharpens the question. A narrator who shares a story in 2026 cannot know whether her story will become training data for an automated system in 2028, what that system will be used for, or whether its use will benefit or harm her. Consent given at the moment of sharing cannot cover uses that were not foreseen. Chapter 19 added a further complication. A story is relational, involving others who did not consent to its sharing, so individual consent cannot cover the relational meaning the story carries.

The Tidal Loom's principle consent to the harvest (no narrative becomes training data without informed consent and a share in the outcome) is a minimum. It requires that harvest be a separate consent from sharing, and that the narrator benefit from what her narrative produces. A fuller principle would add relational consent, at least to the extent of considering the interests of others in the story, and revocable consent, letting a narrator withdraw her story from training data when its use changes.

The Tails

The third stage of the tension is the most technical, and the most damaging for a practice built on weak signals.

Weak signals live in the tails

Anthro-complexity's interest in narrative is, in large part, an interest in weak signals: early indications of change that are not yet visible in aggregate indicators. A weak signal is, by definition, atypical. It is the story that does not fit the pattern, the outlier, the narrative in the corner of the triad where few others are. Snowden has long argued that weak signals are where the important information in complex systems lies, because they show what is beginning to emerge before it becomes dominant. Fat-tailed distributions, in which rare events have disproportionate consequences, make them especially important. The Compass and the Loom noted Snowden's own point that fat-tailed power laws make "most likely to change" a misleading guide to where the important changes lie.

Model collapse erases the tails

A computational system that learns from narratives learns their distribution. It learns what is typical, what patterns are common and what signification goes with what kind of story. If the system's outputs are then used to shape what is collected, attended to or acted on, for instance by flagging typical patterns for attention, suggesting signifiers, or generating synthetic narratives for training, the distribution it learned feeds back into the distribution it learns from next.

Research on large language models has documented what happens when models are trained recursively on data generated by earlier models. Shumailov and colleagues, in work published in Nature in 2024, called the phenomenon model collapse. With each generation of recursive training, the tails of the original distribution are progressively lost. Rare events become rarer, then disappear. The model converges on a narrower and narrower version of the typical. The mechanism is statistical. A finite sample from a distribution under-represents its tails, a model trained on the sample learns a distribution with thinner tails, a sample from that model under-represents the tails further, and so on.

The Compass and the Loom drew the consequence. Recursive training erases the tails, which are where weak signals appear.

Why tails are lost. Consider a distribution with a rare category, occurring with probability p, small. A sample of size n will contain, on average, np instances, and with probability (1−p)n it will contain none. A model trained on a sample that happens to contain none will assign the category probability near zero. A model trained on a sample with a few instances will estimate p with large error, and errors in small probabilities tend towards underestimation, because the estimate cannot go below zero but can easily go below the truth. If the model's outputs then become (part of) the next sample, the rare category is further under-represented, and the process compounds. The Good–Turing estimator, devised by Alan Turing and I. J. Good during the Second World War for estimating the probability of unseen events, gives a principled correction: the probability mass of unseen categories is estimated from the frequency of categories seen exactly once. The Compass and the Loom's Pyxis game includes "Turing's Glass", a Good–Turing estimate of the missing mass, as a discipline for exactly this reason. An instrument that wants to see weak signals must estimate what it is not seeing, and must protect its tails from the model's pull towards the typical.

The weak-signal instrument that starves itself

The consequence for anthro-complexity is direct. A narrative instrument whose purpose is to detect weak signals, if coupled to a computational system that learns from narratives and shapes what is collected or attended to, will over time see fewer weak signals. This is not because the system has fewer of them, but because the instrument's coupling to the model has thinned the tails it was meant to watch. The anti-computational ontology says the human is not a computer, and that the important information is in the atypical. The computational practice, if it feeds a model's expectations back into what humans are asked to notice, makes the human more like the model's picture of the human, more typical and less surprising, and erases the atypical it was built to find.

This is the sharpest point of the thesis. It is not only that the computational practice loses what the ontology values, the relational, entangled and tacit. It is that the practice, under recursive coupling, actively reduces the thing the instruments exist to detect.

Hutchins and the Arrangement

The thesis does not conclude that anthro-complexity should abandon computation. That would be impractical, since work at scale requires it, and it would be wrong, since computation can genuinely help humans see patterns in their own experience. The conclusion is that the arrangement of humans and machines matters, and that the current arrangement is one-directional in a way the ontology does not support.

Edwin Hutchins's distributed cognition, examined in Chapter 31, gives the right framing, and The Compass and the Loom adopts it. Cognition is distributed across arrangements of people and tools, so the question is how to arrange what passes between them. On the navigation bridge Hutchins studied, the instruments (the alidade, the chart, the plotting tools, the bearing record) did not replace the navigators' cognition. They were part of it, arranged so that each person's contribution was visible to the others, errors could be caught, and the whole team's knowledge exceeded any individual's. Crucially, the arrangement was shared. Everyone on the bridge could see the chart. The instruments served the team's distributed cognition and did not extract it for use elsewhere.

The narrative instrument's arrangement differs from the bridge's on exactly this point. The narrators' contributions flow to a centre, where they are aggregated and seen first, often only, by those who commissioned the instrument. In the hybrid case they flow further, to automated systems that learn from them. The arrangement is not shared. It is extractive, in the plain sense that what is contributed at the periphery is used at the centre.

The Grating

The Compass and the Loom proposes a different arrangement, through Kipling's Mariner. In the Just So Story "How the Whale Got His Throat", a whale who eats every fish in the sea swallows a shipwrecked Mariner, who, being a man "of infinite-resource-and-sagacity", builds a grating out of his raft inside the whale's throat. The grating lets small fish through and stops anything large. The whale can never again swallow a man, or eat everything in the sea.

The treatise reads the grating as a design principle for human-machine arrangements. A "grating" is a checkpoint where a human must sense before a machine decides. Its "suspenders" are trust, informal networks and people who notice. Mesh size matters. Too wide and the system swallows everything, too fine and it starves, which echoes the Zeno effect. Frictionless adoption removes these gratings for efficiency, and the treatise's warning is that this lets the whale eat everything in the sea.

The grating answers the thesis's tension directly. It does not refuse the machine. The whale keeps eating small fish. But it places a human checkpoint at the point where the machine's appetite would otherwise extend to everything, and it sets the mesh size so that what passes is what the humans have sensed and judged. In narrative instruments, a grating would be placed at several points:

  • Between narrative and data. Before a narrative becomes data for a model, the narrator, or a body representing narrators, consents to the use. The Tidal Loom's consent to the harvest.
  • Between pattern and action. Before a pattern detected by a model prompts action, a human who knows the context senses whether the pattern means what the model says. Return as design requirement: narrators receive patterns before anything else.
  • Between model and collection. Before a model's expectations shape what is collected (which signifiers are offered, which patterns are flagged), a human checks that the tails are being protected. A Good–Turing discipline: estimate what is not being seen.
  • Between typical and atypical. Atypical narratives, the potential weak signals, are routed around the model, to humans who read them as stories and not as outliers to be discounted.

The mesh size at each grating is a design choice with consequences. Too fine, with every narrative reviewed by a human before any computation, and the instrument cannot work at scale. Too wide, with computation unchecked, and the whale eats the sea. Snowden's own mētis principle applies: the mesh size is a matter of judgement, local to each deployment, and cannot be specified in advance.

Arrangements Compared

Feature Extractive arrangement (harvest loop) Distributed arrangement (Hutchins's bridge) Grated arrangement (Mariner)
Direction of flow Periphery → centre → model Shared among participants Periphery → grating → centre, with return
Who sees the patterns Commissioners first; model All participants Narrators first (return), then commissioners
Consent To share Implicit in shared work Separate consent to harvest; revocable
Tails Thinned by recursive coupling Visible to all; errors caught Protected; atypical routed to humans
Human role Sensor; source of training data Participant in distributed cognition Checkpoint; judge of what passes
Fits the anti-computational ontology? No Yes Yes

The adversary will say this thesis attacks a straw man. SenseMaker is not surveillance capitalism. Narratives are not sold to third parties, and the Cynefin Company's practices around consent and data protection are, by any industry standard, careful. The harvest loop The Compass and the Loom found in one brochure is a feature of one product's marketing, not of anthro-complexity's practice as a whole. And model collapse is a phenomenon of large language models trained on their own outputs, not of narrative instruments that aggregate human signification.

The adversary is right on each factual point. SenseMaker is not surveillance capitalism, and its data practices are careful. The thesis does not claim otherwise. It claims that the structure of narrative-to-data conversion, as instruments move towards human/AI hybrids, has points of contact with surveillance capitalism's structure (raw material taken from human experience, surplus flowing to the centre, consent covering sharing but not use) that a theory of human complexity must address in its design and not only in its data-protection policy. Model collapse in its strict form is a phenomenon of recursive training. But the general mechanism, models that learn the typical and feed their expectations back into what is collected thin the tails, applies to any coupled system of humans and models, and anthro-complexity's turn towards hybrids makes it relevant. The adversary wins on the facts of current practice. The thesis concerns the direction of travel. partial

Verdict

Thesis VI: an anti-computational ontology with a computational practice. holds

The ontology holds that human meaning is relational, entangled, tacit and mētic. The practice makes it legible (Scott) by selecting signified features and leaving out the address, the relations, the participatory sense-making, the entangled, the tacit and the mētic. As instruments move towards human/AI hybrids, legibility becomes harvest, with points of contact with Zuboff's behavioural surplus that a theory of the human must design against. Recursive coupling of narrative instruments to models thins the tails where weak signals live, so the instrument starves itself. The remedy is not to refuse computation but to change the arrangement, from extractive to distributed and grated: Hutchins's shared bridge and the Mariner's grating, with consent to the harvest, return before action, protection of the tails and a mesh size set by judgement.

THE HUMAN IS NOT A COMPUTER —

BUT THE PRACTICE CAN MAKE HER LOOK LIKE ONE —

FIT A GRATING IN THE WHALE'S THROAT.

Addition: Thesis VI After Book Nine, and Its Argument in Cards

The post on Luhmann grounds self-signification in a right and promises a record that later readings must answer to. Chapter 52 asks who can consult the record. The trialectic's own word for what the instruments yield is legibility. Flinn's large group is the mirror image: the tellers are present, and few. See Chapters 50, 52 and 57.

The three cards below set out the thesis in full: its claim and premises, the evidence with the strongest objection and the reply, and the consequence with a test and the verdict as it now stands.

Chapter 25

Thesis VII: The Triad Is a Simplex

4,453 words, about 19 minutes of reading

Thesis VII: the triad is a simplex. The barycentric triad of narrative instruments builds a zero-sum trade-off into its geometry. It cannot tell strong-in-all from weak-in-all, and in that respect it dichotomises more than independent scales do. Peirce's thirdness, which the trialectic claims to honour, is the interpretant. In narrative sensing the interpretant role is split between narrator and analyst, and the mediation the trialectic celebrates sits with the designer. (Theses 17 and 18.)

The Argument for Three

Chapter 5 set out the argument for the trialectic. Give people a pair and they will turn it into a dichotomy: a dialectic hardens into an either/or, with sides to be taken and a winner expected. That is habit, not misunderstanding, and better explanation will not fix it. So change the structure people think with. Three elements in tension cannot be reduced to a pair of sides. Each pair is mediated by the third, so the tension does not resolve into oscillation, balance or synthesis, and the system stays open and generative. The debt is to Peirce, for whom thirdness, the category of mediation, habit and law, is irreducible.

The argument is attractive, and its behavioural premise, that people dichotomise pairs, is well supported. Snowden holds the dyad-and-triad point "most loosely", and invites the question whether the Chinese sources have a way of handling mediation he is missing. Chapter 38 takes up that invitation. This chapter examines the triad as instrument: the triangle on which SenseMaker narrators place their stories. It argues that the instrument's geometry has properties the concept does not, and that some of those properties work against the concept's purpose.

The Triad as Instrument

A SenseMaker triad is a triangle with a label at each vertex. A typical example might be a story's relation to a decision, with vertices labelled following the rules, doing what is right and doing what works. The narrator places a dot inside the triangle to show how far her story reflects each of the three. A dot at a vertex means the story is entirely about that one. A dot at the centre means it reflects all three equally. A dot on an edge means it reflects two and not the third.

Mathematically, the dot's position is given in barycentric coordinates: three non-negative numbers, one for each vertex, that sum to one. A dot at the centroid has coordinates (⅓, ⅓, ⅓). A dot at the first vertex has coordinates (1, 0, 0). A dot halfway along the first edge has coordinates (½, ½, 0). The set of all possible positions is a 2-simplex, the simplest two-dimensional figure in which three quantities that sum to a constant can be represented.

The triad's designers chose this geometry for good reasons. A triad, they argue, is harder to game than a scale. On a five-point scale there is usually an obvious "right" answer (agree that the organisation is innovative), and narrators who want to please or protect themselves can give it. On a triad, with three positive or three neutral labels, there is no obvious right answer, so narrators must think about where their story actually belongs. Triads also elicit more nuanced signification than binary choices, and they resist the dichotomising habit by offering three directions, not two.

The Geometry of the Simplex

The simplex has properties that follow from its geometry and do not depend on the designers' intentions. Three bear on the thesis.

Property one: closure

Barycentric coordinates sum to one. This is the closure constraint, and it means the three values are not independent. If a narrator moves her dot towards one vertex, she must move it away from at least one of the others. More of A means less of B or C, or both. The geometry builds in a trade-off. The three qualities are represented as shares of a whole, competing for a fixed total.

Statisticians have studied closed data, data whose components sum to a constant, for over a century, and they have found it treacherous. Karl Pearson warned in 1897 of spurious correlation arising when quantities are expressed as proportions of a common total. If three quantities are independent, their shares of the total will be correlated, typically negatively, simply because of the closure. John Aitchison's The Statistical Analysis of Compositional Data (1986) founded a field devoted to the problem. It showed that standard statistical methods applied to compositional data produce misleading results, and that the data must be transformed, by log-ratio methods, before analysis.

For narrative triads, closure means that patterns of negative association between vertices may be artefacts of the geometry, not facts about the stories. If many narrators place their stories towards following the rules, they will by geometric necessity place them away from doing what is right and doing what works. An analyst who reads this as "stories of rule-following are stories of not doing what is right" has read the closure, not the narratives. The trade-off was built into the instrument.

Property two: the centre conflates

A dot at the centroid, (⅓, ⅓, ⅓), means equal shares of the three qualities. But equal shares of what total? The simplex has no way to represent magnitude, only proportion. A story that strongly reflects all three qualities, a rich story in which rules, rightness and pragmatism are all deeply engaged, and a story that weakly reflects all three, a thin story in which none is much engaged, both plot at the centre. The Compass and the Loom observed exactly this: on a barycentric (sum-to-one) SenseMaker triad, strong-in-all and weak-in-all plot at the same centre.

The conflation is not trivial. The centre of a triad is where many narrators place stories they find hard to signify, and also where stories that engage all three qualities are placed. An analyst seeing a cluster at the centre cannot tell which it is. Rich and empty, engaged and indifferent, look the same. The Compass and the Loom drew the consequence for Snowden's acumen framework: acumen is a conjunction and needs independent scales or a weakest-link reading. A quality that requires all three elements to be strong cannot be represented on an instrument that represents only their proportions.

Property three: the edges and corners exclude

Every position on a triad's edges and corners excludes something. A dot on the edge between A and B has zero C. A dot at vertex A has zero B and zero C. The simplex has no way to represent "A strongly, B strongly, and C not at all relevant to this story", as distinct from "A and B, and the opposite of C". Absence and opposition are the same position.

Does the Simplex Dichotomise?

The thesis claims that the triad, in respect of closure, dichotomises more than independent scales. The claim is strong, and must be stated carefully.

Consider the alternative: three independent scales, one for each quality, each running from "not at all" to "strongly". A narrator can mark her story strongly on all three, weakly on all three, strongly on one and weakly on others, or any combination. The scales are independent: marking one high does not force another low. There is no built-in trade-off.

The triad, by contrast, forces a trade-off. Placing a story towards one vertex requires placing it away from others. The narrator is asked, implicitly: of these three, which does your story reflect more, at the expense of the others? That is a comparative judgement, and comparative judgements among a fixed total are a form of either/or. Not a binary either/or between two poles, but a competitive allocation among three. The narrator cannot say "all of them, strongly". She can only say "all of them, equally", which, as Property two shows, is also what "none of them, much" looks like.

In respect of closure, then, the triad imposes a structure of competition among its three elements that independent scales do not. The trialectic concept says that three elements in tension cannot be reduced to sides, and that each pair is mediated by the third, keeping the system open. The triad instrument says the three elements compete for a fixed share, and that more of one means less of the others. The instrument does not reduce the three to two sides. But it does reduce their relation to a zero-sum allocation, which is a different kind of closure than the trialectic intends.

Proposition granted: Three elements in tension cannot be reduced to a pair of sides; each pair is mediated by the third, keeping the system open and generative.

Then it must be granted that an instrument honouring the proposition should let all three elements be strongly present at once, since tension among strongly present elements is what keeps the system generative. — Granted.

Then it must be granted that a barycentric triad cannot represent all three elements as strongly present, only as equally proportioned. — Granted. That is the geometry.

Then it must be granted that the barycentric triad, in this respect, represents the three elements as competing shares and not as mediated tensions. — Granted.

Then the respondent may reply that the triad is not meant to measure magnitudes, only the narrator's sense of where her story belongs among three possibilities, and that this relative placement is what elicits nuanced signification. — Granted, as a reply. But then the triad measures relative emphasis, and the concept it embodies, the trialectic, is a concept of mediated tension. Relative emphasis and mediated tension are different things, and the instrument measures the first while the concept describes the second.

Where Ternary Plots Belong

The simplex is not a bad instrument. It is the right one for a particular kind of data, and seeing where it belongs clarifies where it does not.

Ternary diagrams are standard in geology, soil science, metallurgy and chemistry, wherever a material is composed of three components that together make up the whole. The classic example is the soil texture triangle, which classifies soils by their proportions of sand, silt and clay. These three really are shares of a whole: a soil sample is 100 per cent sand, silt and clay together, and more of one does mean less of the others. The closure constraint describes the material exactly. The same holds for the composition of alloys, the mineral content of rocks and the phase diagrams of three-component mixtures.

The question for narrative triads is whether the qualities at the vertices are components of a whole in this sense. Is a story made of following-the-rules, doing-what-is-right and doing-what-works, as a soil is made of sand, silt and clay, so that more of one means less of the others? For some triads the answer may be yes, roughly. A story's focus may be divided among three concerns, and more attention to one may mean less to the others. For many triads the answer is no. A story can be deeply about rules and deeply about rightness and deeply about pragmatism. Indeed, the richest stories of moral and practical life are those in which all three are in play at full strength, in tension. Such stories are the generative tensions the trialectic describes, and they are the stories the simplex represents worst.

Simmel: Three Is Not Neutral

The sociological literature on triads adds a further consideration. Georg Simmel, in his essays on the quantitative determination of groups (1908), showed that the move from two to three transforms social relations, and not only in the direction the trialectic hopes.

In a dyad each party faces the other directly, and the relation exists only as long as both sustain it. The dyad is intimate and fragile. In a triad a third party appears, and with it three new possibilities that Simmel identified:

  • The mediator or arbitrator, who stands between the other two and helps them reach agreement. This is the trialectic's hope: the third mediates.
  • Tertius gaudens, "the third who enjoys", who benefits from the conflict between the other two, by playing them off against each other or by gaining when they are weakened by their dispute.
  • Divide et impera, divide and rule, in which the third deliberately creates or sustains conflict between the other two in order to dominate both.

Theodore Caplow's Two Against One (1968) developed Simmel's insight into a theory of coalitions in triads. In any triad, the most likely structure is a coalition of two against one, and which coalition forms depends on the distribution of power among the three. Triads are not stable mediating structures. They are arenas of coalition.

The Compass and the Loom summarised it: threes create coalitions (the third who benefits, divide et impera). The trialectic hopes that three keeps the system open, because each pair is mediated by the third. Simmel and Caplow show that three can also close the system, by coalition of two against one, or by a third who profits from or engineers the conflict of the others. The number three is not inherently mediating. It is a structure within which mediation is one possibility among several, and which possibility is realised depends on power.

This matters for the triad as an instrument of signification because the narrator is not alone with the three vertices. There is a fourth party, the designer, who chose the vertices, and a fifth, the analyst, who reads the dots. The narrator places her story among three qualities that someone else chose, and her placement is read by someone else for purposes she does not know. In Simmel's terms, the designer and analyst stand in the position of the third to the narrator's relation with her own experience. They mediate it, or they enjoy it, or they rule it, depending on what they do with what she signifies.

Peirce: Where Is the Interpretant?

The deepest point of the thesis concerns Peirce, from whom the trialectic takes its authority. Peirce's thirdness is not simply "three things". It is the category of mediation: of law, habit and generality, of the sign relation in which a representamen stands for an object to an interpretant. The irreducibly triadic relation in Peirce is the sign. A sign cannot be reduced to two-place relations because the interpretant, the effect of the sign on an interpreter, is essential to it. Without an interpretant there is no sign, only a thing and another thing.

Where, in the narrative triad, is the interpretant?

Map the triad onto Peirce's sign relation. The narrator's story is the object, the experience being signified. The narrator's dot on the triad is the representamen, the sign-vehicle that stands for the story's meaning. The interpretant is the effect of the dot on an interpreter, the meaning it comes to have for someone who reads it.

There are two candidate interpreters. The narrator interprets her own story in placing the dot. Her placement is her interpretant of her experience. The analyst interprets the dot, along with thousands of others, in reading the pattern. The pattern is the analyst's interpretant of the population's significations. Peirce's semiotics is endless: each interpretant is itself a sign that calls out a further interpretant. The narrator's dot calls out the analyst's pattern, and the analyst's pattern calls out leadership's decision. The question is which interpretant governs action.

In SenseMaker as typically deployed, the narrator's interpretant is captured in the dot, and the dot is then a sign for the analyst. The analyst's interpretant, the pattern, is what is reported, discussed and acted on. The narrator's interpretant is the raw material. The analyst's is the product. Thirdness, the mediating interpretation that makes the sign relation complete, is located with the analyst.

The trialectic celebrates thirdness as what keeps systems open: mediation, habit and the irreducible third that prevents collapse into dichotomy. In the narrative instrument, the third that mediates is the analyst's reading, and behind it the designer's choice of vertices. The narrator supplies the second, the reaction, the brute placement of a dot in response to a prompt. The mediation the trialectic celebrates sits with the designer and analyst, not with the narrator. The triadic structure of the sign has been preserved, but its third term has been placed at the centre, not with the person whose experience it is.

A Peircean Triad

What would a triad look like that honoured Peirce, and kept the interpretant with the narrator?

  • Independent scales alongside the triad. Let narrators mark each quality's strength independently as well as placing the dot. Strong-in-all and weak-in-all are then distinguishable, and the trade-off is no longer forced.
  • A weakest-link reading for conjunctions. Where the quality of interest requires all three elements (as acumen does, on The Compass and the Loom's analysis), read the minimum of the three independent scales, not the triad position.
  • "None of these" and "all strongly". Let narrators mark explicitly that their story is about none of the vertices, or about all of them strongly. Then the centre no longer conflates.
  • A fourth vertex proposed. Let narrators name what is missing, the quality their story is about that no vertex names. Record the proposals as data about the signifier set's agenda (Thesis V).
  • The narrator's interpretant returned. Show narrators the pattern their dots contributed to before anyone else sees it, and ask them what it means. Their reading of the pattern is a second-order interpretant, and it becomes part of the record, alongside the analyst's.
  • Dialogic signification. Where possible, let stories be signified in conversation, by the narrator with others in the story or with a group, so that the interpretant is formed in participatory sense-making and not by the individual alone (Thesis I).

Each proposal is feasible, and several are close to practices anthro-complexity's more careful deployments already use. Together they would move the third term of the sign relation, the interpretant, back towards the narrator, and so honour Peirce's thirdness in the instrument as well as in the concept.

The adversary has a strong case. The triad was not designed to measure magnitudes, and the closure constraint is known to its designers. It elicits relative placement, and relative placement is valuable precisely because it forces the narrator to think about where her story belongs instead of rating everything highly. The conflation at the centre is a known limitation, addressed in practice by other signifiers (dyads, multiple-choice questions, free text) that sit alongside the triads. Pearson's and Aitchison's warnings apply to statistical analysis of compositional data, and SenseMaker's patterns are read visually, by people exploring the data, not by regression. And the analyst's interpretant is not privileged: SenseMaker's principle is that patterns are returned to the population, who make their own sense of them.

The adversary wins much of this. The triad's designers know its geometry, other signifiers do compensate, and visual exploration is less vulnerable to spurious correlation than regression. The return of patterns to populations is a real principle. But the thesis's claims stand at their core. The geometry builds in a trade-off that the trialectic concept does not intend. The centre conflates rich and empty. Simmel shows three is not inherently mediating. And the interpretant, in typical deployment, is formed and acted on at the centre before (if ever) it is returned. The adversary shows the instrument is used carefully. The thesis shows that careful use is compensating for a geometry that pulls against the concept. partial

Verdict

Thesis VII: the triad is a simplex. partial

The geometric claims hold. The barycentric triad is a 2-simplex with closure, which builds in a zero-sum trade-off, conflates strong-in-all with weak-in-all at the centre and equates absence with opposition at the edges. These are properties of the instrument, not of the trialectic concept, and they pull against the concept's aim of mediated, generative tension. Simmel and Caplow show that triads are arenas of coalition as well as mediation. Peirce's thirdness, the interpretant, is located in typical deployment with analyst and designer, not narrator. The stronger claim, that the triad dichotomises more than independent scales, holds only in respect of closure. The adversary rightly shows that other signifiers compensate and that careful practice mitigates the geometry. The remedy is a Peircean triad: independent scales alongside, weakest-link readings, explicit none and all, a proposed fourth vertex, and the narrator's interpretant returned first.

SAND, SILT AND CLAY SUM TO ONE —

RULE, RIGHT AND WHAT WORKS NEED NOT —

AND THE THIRD THAT MEDIATES SHOULD BELONG TO THE ONE WHO SPOKE.

Addition: Three Stories on One Triangle

The geometric argument of Thesis VII can be made concrete with three invented stories and a single triad. The example uses no real data. It is a thought experiment designed to show what the simplex does to signification.

The triad

Suppose a triad asks narrators about a decision in their story, with three vertices:

  • A: following the rules
  • B: doing what is right
  • C: doing what works

Three stories

Story 1, the rich story. A ward nurse describes a night when a patient's condition deteriorated and the protocol required her to call a doctor who was known to be slow to respond. She called him, and she called the rapid-response team, and she began the treatment she knew was needed while waiting. The protocol was followed, the patient's welfare was served and the outcome was good. In her own experience, the story is strongly about all three: rules, rightness and pragmatism, all fully engaged and in tension.

Story 2, the thin story. An administrator describes completing a routine form at the end of a quiet shift. The form was required (rules), it was harmless (rightness, barely engaged) and it got done (pragmatism, barely engaged). The story is weakly about all three. None was much in play.

Story 3, the conflicted story. A junior doctor describes being told by a senior colleague to discharge a patient early to free a bed. The rules allowed it, it seemed wrong to her, and it worked for the hospital. In her experience, the story is strongly about rules and strongly about rightness, which pulled against each other, and strongly about what works, which sided with the rules.

How the triad represents them

On a barycentric triad, each narrator can only show proportions. The nurse, feeling all three equally and strongly, places her dot at the centre: (⅓, ⅓, ⅓). The administrator, feeling all three equally and weakly, also places his dot at the centre: (⅓, ⅓, ⅓). The junior doctor, feeling all three strongly but in conflict, has to choose a placement that shows proportions. Perhaps she places her dot nearer A and B than C, say at (0.4, 0.4, 0.2), even though in her experience pragmatism was as strongly present as the others. The geometry forces her to reduce C to show A and B.

The result:

Story Narrator's experience Triad placement What is lost
1. Rich (nurse) All three strong Centre (⅓, ⅓, ⅓) Magnitude: indistinguishable from story 2
2. Thin (administrator) All three weak Centre (⅓, ⅓, ⅓) Magnitude: indistinguishable from story 1
3. Conflicted (junior doctor) All three strong, A and B in conflict, C allied with A (0.4, 0.4, 0.2) C's strength; the conflict between A and B; C's alliance with A

Stories 1 and 2, which could hardly be more different, are geometrically identical. Story 3's most important feature, the conflict between rules and rightness and pragmatism's alliance with rules, is not represented at all. The triad shows proportions, and conflict and alliance are relations between the elements, not proportions of them.

How independent scales would represent them

Now suppose the narrators also mark each vertex on an independent scale from 0 (not at all) to 1 (strongly):

Story A (rules) B (right) C (works)
1. Rich 0.9 0.9 0.9
2. Thin 0.2 0.1 0.2
3. Conflicted 0.9 0.9 0.8

Stories 1 and 2 are now clearly distinguished. Story 3 is closer to story 1 than the triad suggested, with all three strongly present. But the conflict and the alliance are still not represented, because independent scales, like the triad, measure elements, not relations.

How a relational signifier would represent them

To represent conflict and alliance, the instrument must ask about relations between the elements, not about the elements themselves. A simple addition would be a question for each pair: in your story, did these work together or pull against each other?

Story A–B A–C B–C
1. Rich Together Together Together
2. Thin (not applicable) (n/a) (n/a)
3. Conflicted Against Together Against

Now story 3's structure is visible. Rules and rightness pulled against each other, pragmatism sided with rules, and rightness stood alone. That is the tension the trialectic concept describes, and it is what makes story 3 important. A pattern of such stories across a hospital would show a systematic conflict between rules and rightness in which pragmatism takes the side of rules, which is exactly the kind of weak signal of moral distress that a narrative instrument exists to detect.

The lesson

The example shows the three levels at which a signifier can work.

Level Instrument What it represents What it misses
Proportion Barycentric triad Relative emphasis Magnitude; relations
Magnitude Independent scales Strength of each element Relations
Relation Pairwise questions How elements work together or against (Least, of the three)

The trialectic concept, three elements in tension with each pair mediated by the third, is a concept at the level of relation. The triad instrument works at the level of proportion. The gap between them is the gap Thesis VII identified. The remedy is not to abandon the triad, whose relative-placement exercise has value, but to supplement it at the other two levels, so that the instrument can represent what the concept describes.

There is a final observation. The pairwise question asks the narrator about relations in her own story. It places the interpretant, the meaning of the tension, with her (Chapter 32), not with an analyst who must infer conflict from clusters of proportions. The narrator knows whether rules and rightness pulled against each other in her experience. The analyst, looking at a dot at (0.4, 0.4, 0.2), cannot know.

THE RICH STORY AND THE THIN STORY SIT AT THE SAME CENTRE —

THE CONFLICT IS A RELATION, NOT A PROPORTION —

ASK THE NARRATOR WHETHER THE THREE PULLED TOGETHER OR APART.

Addition: Thesis VII After Book Nine, and Its Argument in Cards

Snowden's triquetra test requires that a third term be irreducible to the other two, neither a midpoint nor a dissolution. Three corners whose shares must sum to a whole do not meet it. See Chapter 48.

The three cards below set out the thesis in full: its claim and premises, the evidence with the strongest objection and the reply, and the consequence with a test and the verdict as it now stands.

Chapter 26

Thesis VIII: Reflexivity Predicts the Degradation of the Instruments

4,116 words, about 18 minutes of reading

Thesis VIII: reflexivity predicts the degradation of the instruments. Starlings do not learn the camera. Humans learn the signifier set. The more true anthro-complexity's human distinction is, the faster stable narrative instruments become counter-performative, through order effects, the social Zeno effect, decoherence into cliché and Goodhart's law. A theory of reflexive systems that offers a stable sensing technology has not yet applied its theory to its own technology. (Thesis 19.)

The Starling and the Camera

The STARFLAG researchers who reconstructed starling murmurations in three dimensions (Chapter 9) placed stereo cameras on the roof of the Palazzo Massimo in Rome and filmed flocks over the city at dusk. The cameras did not change the flocks. The starlings did not know they were being filmed, and if they had known, they could not have adjusted their topological interaction rule, their six or seven nearest neighbours, to the researchers' interest. The measurement left the measured unchanged. That is what makes the murmuration a measured physical system, and the findings reliable. The same cameras, used next year, would film the same kind of flock.

Now imagine the cameras turned on a human crowd, and the crowd told it is being filmed for a study of collective behaviour. The crowd changes. Some members perform for the camera and some avoid it. Some wonder what the study is for and adjust their behaviour to serve or frustrate it. Next year, if the study's findings have been published, the crowd has read them, and behaves differently again: confirming them, defying them, or doing something the findings did not anticipate. The camera, used next year, films a different kind of crowd, partly made by the last year's filming.

This is the difference anthro-complexity names when it says humans are not ants, or birds. Chapter 14 called it descriptive reflexivity: human systems are changed by public descriptions of themselves. This chapter draws a consequence the November post does not draw. If human systems are descriptively reflexive, then any stable instrument for sensing them will degrade. Its readings will drift from what they were designed to measure, because the measured will learn the instrument and respond to it. The more reflexive the system, the faster the drift. And anthro-complexity, which holds that human systems are deeply reflexive, offers stable instruments.

Six Mechanisms of Degradation

The social sciences have documented the mechanisms by which measurement of human systems degrades. Six are directly relevant.

1. Goodhart's law

Charles Goodhart, an economist at the Bank of England, observed in 1975 that any observed statistical regularity will tend to collapse once pressure is placed on it for control purposes. The popular formulation, due to Marilyn Strathern, is: when a measure becomes a target, it ceases to be a good measure. Goodhart's example was monetary aggregates. Once the Bank began targeting a particular measure of the money supply, the relationship between that measure and the economic outcomes it had predicted broke down, because financial institutions changed their behaviour in response to the targeting.

2. Campbell's law

Donald Campbell, the methodologist, stated a parallel principle in 1976: the more heavily a quantitative social indicator is relied on for social decision-making, the more it is exposed to pressures that corrupt it, and the more likely it becomes to distort and corrupt the very social processes it was meant to monitor. Campbell's examples were educational testing and crime statistics. When test scores are used to evaluate schools, teaching narrows to the test. When crime clearance rates are used to evaluate police, recording practices change.

3. The Lucas critique

Robert Lucas argued in 1976 that macroeconomic models estimated on historical data cannot be used to predict the effects of policy changes, because the parameters of the models (consumption functions, investment equations) are not invariant to policy. They reflect agents' expectations about policy, and agents change their behaviour when policy changes. A model that predicted well under one policy regime will predict badly under another, because the regime change changes the behaviour the model was fitted to. The Compass and the Loom noted the risk for steering: conquest risks the Goodhart and Lucas failure, because the steered system learns the steering.

4. Reactivity of rankings

Wendy Espeland and Michael Sauder's study of US law school rankings, "Rankings and Reactivity" (2007) and Engines of Anxiety (2016), documented in detail how a measure reshapes what it measures. When US News & World Report began ranking law schools, the schools began to manage their rankings: adjusting admissions to raise median test scores, reclassifying graduates' employment to improve placement statistics, and redirecting resources towards ranked factors and away from unranked ones. The rankings did not merely measure the schools. They changed them, and the change was towards the measure. Espeland and Sauder called this reactivity, and identified two mechanisms: self-fulfilling prophecy, as rankings shaped applicants' choices and so shaped the schools' quality, and commensuration, as rankings made unlike schools comparable on a single scale, so that differences the scale did not capture became invisible.

5. Question-order effects

Chapter 40 examines in detail the work of Zheng Wang and colleagues (2014), who modelled survey questions as non-commuting measurements, in the manner of quantum observables. Asking question A before question B produces different answers from asking B before A, because the first question changes the respondent's state, the context in which she answers the second. The Compass and the Loom draws the design consequence: signifier sets are sequences of interventions that change the respondent. A signifier set is not a neutral array of measures applied to a fixed state. Each signifier changes the state the next one meets.

6. Demand characteristics and social desirability

Martin Orne's work on demand characteristics (1962) showed that research participants form hypotheses about what a study is for and adjust their behaviour to fit, typically to be "good subjects" who confirm what they think the researcher wants. The social desirability literature shows that respondents shade their answers towards what they believe is socially approved. Both are forms of reactivity at the level of the individual respondent. The measured person reads the measurement and responds to it.

Mechanism Source What learns How the instrument degrades
Goodhart's law Goodhart (1975); Strathern Agents targeted by a measure Measure decouples from what it indicated
Campbell's law Campbell (1976) Organisations evaluated by indicators Indicator corrupts the process it monitors
Lucas critique Lucas (1976) Agents forming expectations of policy Parameters shift when the regime changes
Reactivity Espeland & Sauder (2007, 2016) Organisations ranked Organisations reshape themselves towards the measure
Question order Wang and others (2014) Respondents' states Each question changes the state the next meets
Demand characteristics Orne (1962) Individual participants Participants confirm what they think the study wants

The Mechanisms Applied to Narrative Instruments

Anthro-complexity's instruments were designed with some of these mechanisms in mind, and the design shows it. SenseMaker's triads, with three positive or three neutral vertices, have no obvious "right" answer, which reduces social-desirability bias and makes gaming harder. Micro-narratives are short and fragmented, which makes it harder for narrators to construct a coherent case. Self-signification means narrators interpret their own stories, which reduces the analyst's imposition. Anonymity reduces the stakes. These are real defences against degradation, and they make narrative instruments more robust than surveys, rankings and key performance indicators.

But defences against gaming are not defences against reflexivity. Gaming is deliberate: a respondent works out what the instrument rewards and supplies it. Reflexivity is broader. A respondent's sense-making is shaped by the instrument whether or not she means to game it. The six mechanisms above include deliberate gaming (Goodhart, Campbell and the law schools all involve strategic adjustment) but also non-strategic reactivity (question-order effects, demand characteristics, the looping of categories into self-understanding). Narrative instruments' defences address the first and not the second.

Consider how a SenseMaker deployment that is repeated, as many are in organisations that use the instrument for ongoing monitoring, would degrade through non-strategic reflexivity.

Round one. The population encounters the instrument for the first time. Its prompts are novel, its triads unfamiliar. Narrators tell stories that come to mind and signify them as best they can. The patterns that emerge are as close to "what is there" as a narrative instrument can get, though even here the prompt, the signifiers and the perceived audience shape what is told (Theses I, V and VII).

Feedback. The patterns are returned to the population, as anthro-complexity's principles require. People see that stories of "compliance" cluster in one area, that stories of "frustration with systems" are common, and that a few outlier stories point to something new.

Round two. The population now knows the instrument. It knows the triads, and it has seen what patterns emerged. Some narrators, without any intent to game, tell stories that respond to the patterns: confirming them ("yes, it's still frustrating"), contesting them ("we're not compliant, we're careful"), or shifting towards what they now see as the interesting categories. The signifiers have become categories through which narrators understand their experience, which is Hacking's looping effect. A narrator who learned in round one that her organisation talks about "trust, compliance, fear" now notices her experiences in those terms. Experiences that do not fit are less noticed, less told and less signified.

Round three and beyond. The looping compounds. The population's narratives converge on the signifiers' categories, because the categories have become the population's self-description. The patterns become more stable, which looks like reliability, but the stability is partly the instrument's own reflection. Outlier stories, the weak signals, become rarer, not because there are fewer weak signals but because the population has learned to tell stories that fit. The instrument has decohered the population's narration into its own categories.

This is not a failure of the instrument's designers or users. It is a consequence of the instrument's being stable, the same prompts and signifiers round after round, in a reflexive system. A stable instrument in a reflexive system is a constant description, and a constant description is learned.

The Social Zeno Effect

The Compass and the Loom's Uncertain Compass, the compass that measures its own uncertainty, offers four laws of the uncertain needle, drawn by careful analogy from quantum measurement. Chapter 40 examines them. Two are directly relevant here.

The first is the Zeno law. In quantum mechanics, the quantum Zeno effect is the phenomenon whereby frequent measurement of a quantum system inhibits its evolution. A system that would otherwise decay or transition is "frozen" by being measured often enough, because each measurement collapses it back to its initial state. The treatise's social analogue: over-frequent monitoring freezes change, so monitor less often than the change you hope to see. A population that is sensed continuously is kept, by the sensing, in the state the sensing expects. Change, which needs room to develop before it is visible, is interrupted by measurement before it can take shape. Organisations that run pulse surveys weekly, monitor sentiment daily, or deploy continuous narrative capture may be freezing exactly the emergent changes they hope to detect.

The second is the decoherence law. In quantum mechanics, decoherence is the process by which a quantum system's interaction with its environment destroys the interference effects that make it quantum, selecting a small set of robust "pointer states" that survive the interaction (Zurek's einselection). The social analogue: saturating expectation selects safe, reportable clichés and destroys interference. A population surrounded by sensing instruments, each expecting certain kinds of report, converges on reports that survive the expectation: safe, recognisable, expected. The interference of perspectives that produces novelty is suppressed. The Pyxis's remedy is to protect pockets of low monitoring, places where sensing is light enough that novelty can form.

The analogical discipline. The Compass and the Loom insists that a quantum image may cross into social life only when the borrowed structure is real in the source domain and the inference does not depend on the source's physics. Does the Zeno analogy pass? The borrowed structure is this: frequent measurement, by resetting a system to the measured state, inhibits transitions that need time to develop. In quantum mechanics this structure is real and well tested. In social systems an analogous structure exists without quantum physics. A frequently asked question resets the respondent's attention to the question's frame, and transitions in attention, the noticing of something new, need time free of the frame to develop. The inference, that frequent sensing inhibits emergent noticing, does not depend on quantum physics. It depends on attention, framing and the time novelty takes to form. The analogy passes. "Decoherence into cliché" passes on the same grounds: interaction with an expecting environment selects robust, expected reports, and that is a fact about attention and social desirability, not about quantum states.

Counter-Performativity

The most dramatic form of degradation is counter-performativity, introduced in Chapter 14 through MacKenzie's study of the Black–Scholes model. A model that once fitted the market because it was used began to misfit it after 1987, because traders' use of it, in conditions it did not anticipate, produced outcomes that contradicted it. The volatility smile was the trace of the model's own effect.

Narrative instruments can be counter-performative in an analogous way. Suppose an organisation's SenseMaker patterns show a cluster of stories about a toxic manager, and the organisation acts on the pattern: the manager is moved, a culture programme is launched. In the next round, stories about the issue change. They now concern the culture programme, the manager's replacement, and the sense that "they only act when the numbers look bad". The instrument's patterns now reflect the organisation's response to the previous patterns, and a reading of them as "what people experience" misses that what people experience is, in part, the instrument's effect. If the organisation reads the new patterns as fresh information about its culture, rather than as responses to its own previous reading, it will be steering by its own wake.

Snowden has a precise formulation for one version of this risk. He describes his acumen framework (the salience, sapience and sentience conditions of collective judgement) as safe as a sensitising concept and unsafe as a score. The distinction comes from Herbert Blumer's 1954 essay "What Is Wrong with Social Theory?", which contrasted definitive concepts, which specify their referents precisely, with sensitising concepts, which give a sense of reference and direct attention without specifying it. A sensitising concept is hard to game because it does not define a target. A score is easy to game because it does. Snowden is right, and the point generalises. Every stable instrument tends, over repeated use, to drift from sensitising towards definitive, from a way of looking towards a target. The drift is driven by the system's reflexive learning of the instrument.

What the Thesis Predicts

If the thesis is right, specific patterns should be observable in repeated deployments of narrative instruments:

  1. Convergence of signification. Over successive rounds, the distribution of significations on stable triads narrows, and narrators converge on positions the previous rounds made salient.
  2. Thinning of outliers. The proportion of stories in the extreme corners of triads, and of stories coded as novel or anomalous, declines.
  3. Vocabulary uptake. The instrument's signifier labels appear increasingly in narrators' free text, as self-description and not only as response.
  4. Reflexive content. An increasing share of stories concerns the organisation's responses to previous rounds, not experience independent of the instrument.
  5. Frame-shift recovery. When signifiers are changed between rounds, the outlier proportion and the distribution's spread temporarily recover.

Each prediction is testable with data anthro-complexity's practitioners hold. The thesis does not claim to know the results. It claims that a theory of descriptive reflexivity predicts them, and that a practice faithful to the theory should test for them.

Designing Against Degradation

The remedy is not to stop sensing. It is to design instruments that expect their own degradation and correct for it. The Tidal Loom and the Uncertain Compass offer several principles, and the thesis adds others.

Principle Source What it does
Frames, not seers Tidal Loom Rotate the search template between rounds, so no single signifier set loops into self-description
Monitor less often than the change you hope to see Pyxis (Zeno law) Leave time for emergent noticing to form before sensing again
Protect pockets of low monitoring Pyxis (decoherence law) Keep some populations, times or topics lightly sensed, so novelty can form outside expectation
Weak reading Pyxis Sense lightly, in ways that inform without freezing, the leader who measures weakly and learns without freezing
Turing's glass Pyxis Estimate the missing mass, the unseen categories, from the frequency of singletons, and watch whether it shrinks
Noticing, not mapping Tidal Loom Measure the latency between something new becoming actual and someone recognising it; rising latency signals decoherence
Dated instruments This book Give each signifier set an expiry; retire it before it becomes the population's self-description
Reflexive content tracked This book Code separately stories about the instrument's own previous effects, and read them as wake, not water

Verdict

Thesis VIII: reflexivity predicts the degradation of the instruments. holds

Six documented mechanisms (Goodhart, Campbell, Lucas, reactivity, question order and demand characteristics) show how measurement of human systems degrades. Narrative instruments defend well against deliberate gaming and poorly against non-strategic reflexivity. Signifiers loop into self-description, frequent sensing freezes emergent change (the social Zeno effect), saturating expectation selects cliché (decoherence), and acting on patterns produces patterns of response to action (counter-performativity). Snowden's own distinction between sensitising concept and score names the drift. The thesis predicts five observable patterns in repeated deployments, testable with data practitioners hold, and proposes design principles that expect degradation and correct for it. A theory of reflexive systems must build instruments that are themselves reflexive about their own effects. The starling does not learn the camera. The organisation does.

THE STARLING DOES NOT LEARN THE CAMERA —

THE WORKFORCE LEARNS THE TRIAD —

MONITOR LESS OFTEN THAN THE CHANGE YOU HOPE TO SEE.

Addition: A Research Design for Testing Thesis VIII

Thesis VIII made five predictions about repeated deployments of narrative instruments, and the afterword to the Coda named them as the disputation's most testable claim. This addition sketches a research design by which practitioners holding the data could test them. It is offered as an invitation, not a protocol, and it would need refinement by people who know the instruments' data structures.

The data needed

The test requires repeated deployments of a narrative instrument in the same population, with at least three, and preferably five or more, rounds. For each round, the following are needed:

  • the signifier set used, with any changes from the previous round;
  • every narrator's signification on every signifier, with the order in which signifiers were presented if known;
  • the free-text narratives;
  • the date of each round, and the dates of any feedback of patterns to the population or to leadership between rounds;
  • a record of any interventions made on the basis of previous rounds' patterns.

Many organisations that use narrative instruments for ongoing monitoring will hold data of this kind. Anonymised data across many deployments, held by the instrument's provider, would allow a stronger test.

The five predictions, operationalised

Prediction 1: convergence of signification. On signifiers that are kept stable across rounds, the dispersion of significations should narrow over rounds, and the mean position should move towards positions made salient by previous feedback.

Measure: for each stable triad, compute a dispersion statistic for each round's significations. Because the data are compositional, use Aitchison's methods, transforming barycentric coordinates by log-ratio before computing variance (Chapter 25). Test for a downward trend across rounds. For movement towards salient positions, identify the cluster centres reported in feedback after each round and test whether subsequent rounds' significations move towards them.

Prediction 2: thinning of outliers. The proportion of significations in the extreme regions of triads (within a set distance of a vertex) and of narratives coded as novel or anomalous should decline across rounds.

Measure: proportion of significations in extreme regions per round; proportion of narratives classified as singletons (themes appearing once) per round; a Good–Turing estimate of the missing mass per round (Chapter 24). Test for downward trends.

Prediction 3: vocabulary uptake. The labels of signifiers should appear with increasing frequency in narrators' free-text narratives, as self-description and not only in response to the signifiers.

Measure: frequency of signifier vocabulary in free text per round, normalised by narrative length. A control is the frequency of semantically similar terms not used as signifiers. Test whether signifier vocabulary rises faster than the control.

Prediction 4: reflexive content. The proportion of narratives that refer to the organisation's responses to previous rounds (changes made, programmes launched, communications about the results) should increase.

Measure: code narratives for reference to previous interventions or to the study itself, using a coding frame developed from the intervention record. Proportion per round. Test for upward trend.

Prediction 5: frame-shift recovery. When signifiers are changed between rounds, the effects in predictions 1 and 2 should partially reverse: dispersion widens, and outliers increase, on the new signifiers.

Measure: compare dispersion and outlier proportion on new signifiers in their first round with the same statistics on stable signifiers in the same round. Prediction 5 holds if new signifiers show greater dispersion and more outliers.

Controls and confounds

Several confounds would need controlling.

Real change. Signification might converge because the organisation really became more uniform, not because narrators learned the instrument. To distinguish the two, compare convergence on signifiers that were fed back with convergence on signifiers that were collected but not fed back. If learning drives convergence, fed-back signifiers should converge faster.

Population turnover. If the population changes between rounds, convergence might reflect selection, with those who disagreed leaving. Track individual narrators across rounds where anonymity permits, or restrict analysis to stable sub-populations.

Intervention effects. If interventions were made between rounds, changes in signification might reflect the interventions' real effects. This is not a confound for prediction 4, which is about such effects, but it is for predictions 1 and 2. The intervention record allows rounds with and without major interventions to be compared.

Instrument changes. Changes to prompts, platforms or deployment methods between rounds could produce artefacts. Record all changes and test their effects.

What the results would mean

Result Interpretation
All five predictions confirmed Thesis VIII holds; descriptive reflexivity degrades stable instruments as predicted; design principles of Chapter 26 indicated
Predictions 1–3 confirmed, 4–5 not Instruments loop into self-description, but organisations do not respond reflexively to feedback, or frame changes do not recover; partial support
Predictions 4 confirmed, 1–3 not Organisations respond to feedback, but narrators do not learn signifiers; the instrument's categories do not loop
None confirmed Thesis VIII fails; narrative instruments are robust to repeated use in the ways tested; the disputation's account of descriptive reflexivity overstates its effect on instruments
Mixed, varying by context Degradation depends on conditions (frequency, feedback mode, participants dimension); the most likely result, and the most useful

The last row is the most likely outcome and the most informative. Degradation probably depends on how often the instrument is used, how patterns are fed back, how much the population trusts the instrument and how coercive the context is. A study that measured these moderators would tell practitioners not only whether their instruments degrade but when, and so how often to rotate frames, how to space rounds and where to protect pockets of low monitoring.

An invitation

The author does not hold the data and cannot run the test. The Cynefin Company and its network hold data from many deployments over many years. A study of this kind, published openly, would be a contribution to the science of narrative sensing. It would test a hypothesis that follows from anthro-complexity's own human distinction, and it would show, whatever its results, that the practice is reflexive about its instruments in the structural sense this disputation has asked for. It would be the rendezvous with the data.

FIVE PREDICTIONS — FIVE MEASURES — FOUR CONFOUNDS —

THE DATA ARE INSIDE —

THE TEST WOULD BE THE PROOF OF STRUCTURAL REFLEXIVITY, WHATEVER IT FOUND.

Addition: Thesis VIII After Book Nine, and Its Argument in Cards

Three later arguments support the thesis. The alibi test of the post on Luhmann requires stated conditions of failure, and the four predictions here are offered as such. The Zen Garden study holds that produced agreement costs work and that a pattern cannot tell improvement from learned response. Chapter 56 gives the general law: compression from inside makes its object more regular.

The three cards below set out the thesis in full: its claim and premises, the evidence with the strongest objection and the reply, and the consequence with a test and the verdict as it now stands.

Chapter 27

Mead: The Conversation of Gestures

3,834 words, about 17 minutes of reading

Why Mead

Book Four advanced eight theses, and in almost every one George Herbert Mead appeared as a resource. He was there in the possessive fallacy (the social self), in the asymmetry principle (the gesture whose meaning is completed by response), in the estuaries (withholding is also a gesture) and in the ethical reason (address and influence). Book Five brings the corpus to bear more fully, taking in turn the thinkers whose work most directly supplies what anthro-complexity's human cluster lacks. Mead comes first. His account of the social origin of mind and self is the foundation on which the relational account of the human rests, and it comes from the same pragmatist tradition that Snowden claims through Peirce.

This chapter sets out Mead's account in more detail than the earlier references allowed, under five headings: the act, the gesture, the significant symbol, the self, and the generalised other. It then examines what the account implies for anthro-complexity, in particular for its asymmetry principle, its instruments and its practitioner. It ends with Mead's theory of time, the philosophy of the present, which bears directly on anthro-complexity's retrospective coherence and its kairos.

Mead and the Metaphysical Club

Mead (1863–1931) is sometimes placed at the margins of the pragmatist tradition, behind Peirce, James and Dewey. That placement understates him. He was Dewey's closest colleague at Michigan and then at Chicago, where he taught from 1894 until his death, and Dewey regarded him as the most original mind in American philosophy of his generation. He published little in book form. Mind, Self, and Society (1934), The Philosophy of the Present (1932), Movements of Thought in the Nineteenth Century (1936) and The Philosophy of the Act (1938) were all assembled after his death from lecture notes and manuscripts. His influence came through his students, who founded symbolic interactionism (Herbert Blumer coined the term), and through the Chicago school of sociology.

For this disputation, Mead's place in the pragmatist tradition matters because Snowden grounds anthro-complexity in Peirce and distinguishes Peircean pragmatism from Ing's Jamesian and Singerian lineages. Mead offers a fourth lineage, which Snowden does not discuss: a social pragmatism in which the Peircean insights (meaning as practical consequence, inquiry as communal, signs as triadic) are developed into an account of how mind and self arise in social interaction. Mead read Peirce, and his account of the significant symbol is in many ways a social-psychological development of Peirce's semiotics. If anthro-complexity's Peircean pragmatism is to have an account of the human, and not only of signs and inquiry, Mead is where it would naturally find one.

The Act

Mead's starting point is the act: the unit of conduct, from the impulse that initiates it to the consummation that completes it. Mead distinguished four phases. Impulse is the organism's readiness to respond to a stimulus. Perception is the selection and organisation of the environment in relation to the impulse. Manipulation is contact with the object. Consummation is the completion of the act. The act is not a response to a pre-given environment. The environment is constituted in the act, as the organism selects what is relevant to its impulse. An object, for Mead, is not a thing in itself. It is a "collapsed act", the set of possible responses it affords.

This is Gibson's affordance theory, avant la lettre, and it shows that Mead is already an ecological thinker in the sense Ing values. But Mead's distinctive contribution concerns social acts: acts that involve more than one organism, so that one organism's act is a stimulus for another's, and the completed act is the joint product of both.

The Gesture

In a social act, the early phases of one organism's act serve as stimuli to another. Mead's famous example is the dog-fight. Dog A begins to bare its teeth, the early phase of an attack. Dog B responds by crouching, bristling or retreating. Dog A's subsequent act is modified by B's response. A gets ready to spring, or backs off, or circles. B's response is modified in turn. The dog-fight is a conversation of gestures: each organism's act is a gesture to the other, calling out a response that in turn becomes a gesture.

Mead's crucial point about the conversation of gestures is that the meaning of a gesture is not in the gesturer. It is in the relation between the gesture and the response it calls out. Dog A's baring of teeth means "attack" not because A intends attack, though it may, but because it calls out a defensive response in B, and that response completes the social act of which the baring was the first phase. Meaning is triadic. It involves the gesture of one organism, the response of another, and the completion of the social act that the gesture initiates. That is Peirce's sign relation, socialised. The interpretant is the other's response.

The Significant Symbol

The dog-fight is a conversation of gestures without significant symbols. Neither dog anticipates the other's response to its own gesture, or takes the other's attitude towards itself. Human beings can, and this is the step that makes the human difference.

A significant symbol is a gesture that calls out in the gesturer the same response, or the tendency to the same response, that it calls out in the other. When I say "fire!", the word calls out in me the tendency to the response it calls out in you, alarm and readiness to flee, and I can therefore anticipate your response, and adjust my gesture in light of the anticipation. The vocal gesture is especially suited to becoming a significant symbol because the gesturer hears it as the other hears it. When I speak, I am affected by my words as you are.

Significant symbols make possible what Mead called taking the attitude of the other. In using a significant symbol, I take your attitude towards my own gesture. I respond to it, implicitly, as you will respond to it. This is the foundation of mind. Mind, for Mead, is not a substance or a faculty located in the brain. It is the internalised conversation of gestures: the process by which an individual, using significant symbols, carries on with herself the conversation she has learned to carry on with others. Thinking is talking to oneself, taking the attitudes of others towards one's own proposed acts, trying them out in imagination against the imagined responses of others, and revising them.

Mind is therefore social in origin and in structure. It arises from social interaction, and it remains a form of social interaction, internalised. There is no mind prior to the social process, and no thinking that is not, at bottom, a conversation.

The Self: "I" and "Me"

Mead's account of the self follows from his account of mind. The self is the individual as an object to herself, something she can think about, evaluate and act on. Only a being who can take the attitude of others towards herself can be an object to herself. The self arises when the individual takes towards herself the attitudes others take towards her, and so sees herself as they see her.

Mead distinguished two phases of the self. The me is the organised set of attitudes of others that one takes towards oneself: one's sense of how one is seen, what is expected of one, what one's roles and responsibilities are. The me is conventional, social and structured. It is the community in the individual. The I is the response of the organism to the attitudes of others, the spontaneous, novel and unpredictable element in conduct. The I is never fully known in advance. It is known only after it has acted, when its action becomes part of the me. Mead wrote that the I of this moment is present in the me of the next.

The self is the relation between the I and the me: a process, not a substance. The me supplies the social structure within which the I responds, and the I's responses, once made, become part of the me and so change the structure. The self is at once social, constituted by the attitudes of others, and individual, the source of novel response. It is neither a container of properties nor a dissolution into relation. It is a social process with a novel edge.

Multiple selves

Mead's account explains multiple identity directly. One has as many me's as there are groups whose attitudes one takes towards oneself. Mead's predecessor William James had already written that a person has as many social selves as there are individuals or groups who recognise him and whose opinion he cares about. Mead gave this a mechanism. Each group's attitudes, taken towards oneself, constitute a me, and which me is active depends on which group's attitudes are called out by the situation. A person at work takes her colleagues' attitudes towards herself and is her work-self. At home she takes her family's attitudes and is her family-self. The switching is not a choice among stored identities. It is the situation calling out a particular organisation of others' attitudes. That is self-categorisation theory's comparative context (Chapter 13), seventy years earlier.

The Generalised Other

Mead's most important concept for social theory is the generalised other: the organised attitudes of the whole community, taken by the individual towards herself. Mead's illustration is the difference between play and the game. In play, a child takes one role at a time. She plays at being a mother, then a shopkeeper, then a police officer, taking the attitude of each in turn. In a game, such as baseball, a child must take the attitudes of all the other players at once, and the rules that organise them. To play shortstop she must know what the pitcher, the catcher, the first baseman and the runner will do, and how the rules relate their actions. The game requires the child to take the attitude of the team as a whole, organised by its rules. That organised attitude is the generalised other.

The generalised other is how the community enters the individual. In taking the attitude of the generalised other towards herself, the individual becomes a full member of the community. She governs her conduct by its organised attitudes, its norms, values and rules. Social control, for Mead, is not imposed from outside. It is self-criticism from the standpoint of the generalised other. And because the generalised other is internalised, the community is present in every member's thinking, as the audience of her inner conversation.

The generalised other and constitutive constraint

The generalised other is Mead's account of what Chapter 22 called constitutive constraint. The constraints that make a community what it is are carried in its members as the generalised other: the organised attitudes each member takes towards herself. They are reproduced in every act in which a member governs her conduct by the generalised other's standpoint. They are changed when the I's novel responses accumulate in the me and alter the organisation of attitudes. Constitutive constraint is reproduced in compliance because compliance is the taking of the generalised other's attitude, and it is changed through the I's responses because the I is never fully governed by the me.

Mead and the Asymmetry Principle

Mead's account transforms the asymmetry principle. The November post states it as a capacity humans have and rivers lack: only humans can change the story they tell about the other. On Mead's account, the capacity to change a story is the capacity to use significant symbols, and significant symbols are social. A story is a significant symbol, a gesture whose meaning is shared because each party can take the attitude of the other towards it. A story changes constraints only when it is taken up, when others respond to it as the teller anticipated, and their responses complete the social act it initiated.

So the asymmetry principle, read through Mead, is not about what humans can do alone. A planner alone cannot change the story the city tells about the river. She can propose a new story, and whether it becomes the city's story depends on the responses it calls out: from other planners, from councillors, from residents, from the press. The capacity to change the story belongs to the community in which stories circulate, and the individual's share of it depends on her place in the community: whose gestures are responded to, whose proposals are taken up, whose attitudes others take towards themselves. That is power (Thesis V), and it is relational (Thesis I).

Mead's account also makes the asymmetry principle apply to itself (Thesis II). Every description of a human system is a gesture in that system's conversation. Cynefin, SenseMaker and estuarine mapping are gestures. They call out responses, and their meaning is completed by the responses. The practitioner who offers a Cynefin categorisation has made a gesture, and its meaning, what the categorisation does, depends on how the organisation responds: whether it takes up the category, acts on its licences, or resists. On Mead's account there is no description that is not a gesture, and no gesture whose meaning is fixed by the describer.

Mead and the Instruments

Mead's account bears on each of anthro-complexity's instruments.

The micro-narrative is a gesture. A story told to an instrument calls out a response, or anticipates one. The narrator, using significant symbols, takes the attitude of her imagined audience towards her own story, and shapes it accordingly. The generalised other is present in every micro-narrative as the audience the narrator imagines. What the instrument collects is not experience but gestures addressed to an imagined other, and the imagined other is shaped by the narrator's sense of what the instrument is for (Bakhtin's addressivity, Chapter 19).

Self-signification is taking the attitude of the other towards one's own story. When a narrator places her story on a triad, she takes the attitude of the instrument's designer, whose categories the triad embodies, towards her own experience. She sees her story as the designer's categories ask her to see it. That is a me: the instrument's generalised other, taken towards herself. The I, the novel response, appears where the story does not fit the categories, in the hesitation, the placement at the centre or the free-text comment. Mead's account predicts that what is most novel in a narrator's experience will be least visible in her signification, because signification is the me's work.

The pattern is a gesture to the population. When patterns are fed back, they call out responses that complete their meaning. Whether a pattern showing "compliance" in one department means "this department is cautious", "this department is oppressed" or "this department is being blamed" depends on how the department responds. Feedback is not information transferred. It is a gesture in a conversation.

The practitioner is a participant. On Mead's account the practitioner has a me constituted by her clients', colleagues' and profession's attitudes, and an I that responds. Her readings of the organisation are internalised conversations with her generalised other, which includes the organisation's leadership, her profession's norms and her instruments' categories. She does not read the organisation from outside. She reads it from the standpoint of a generalised other that is partly the organisation's own.

The Philosophy of the Present

Mead's last major work, The Philosophy of the Present (1932), the Carus Lectures delivered shortly before his death, developed a theory of time that bears directly on two of anthro-complexity's elements: retrospective coherence (Element 8) and kairos (Element 10).

Mead's central claim is that reality exists in the present. The past and the future are not realms that exist alongside the present. They are dimensions of the present: the past as the conditions the present arises from, the future as the possibilities it opens. But the present is not a knife-edge between them. It is an event with temporal spread, in which something emerges, something new that was not determined by the past.

The emergence of the new has a consequence for the past. When something new emerges, the past must be reconstructed to account for it. The past that led to the emergent event is not the same past as before the event, because the event reveals connections, tendencies and possibilities in the past that were not visible, or did not exist as such, before. Mead's example is the emergence of life. Once life has emerged, the history of the earth becomes the history of the conditions for life, a history that could not have been written before life, because the category of "conditions for life" did not yet apply. Every emergent present rewrites its past.

Retrospective coherence, Meadian

This is retrospective coherence with a twist anthro-complexity has not taken. Cynefin says that in the complex domain cause and effect are coherent only in retrospect. After the event, a coherent story can be told about why things happened as they did, and the same story could not have been told before. Mead agrees, and adds that the retrospective story is not a discovery of what the past was. It is a reconstruction of the past from the standpoint of the present, and different presents reconstruct different pasts. The coherence is real, but it belongs to the present that finds it, and it will change when a new present emerges.

The implication for narrative instruments is significant. A micro-narrative is a retrospective coherence: a story told now about what happened then. On Mead's account the story is a reconstruction of the past from the narrator's present, and the narrator's present includes the instrument. The prompt, the frame and the anticipated audience are part of the present from which the past is reconstructed. Change the present, by changing the prompt or the audience, and the reconstructed past changes. The narrative instrument does not collect the past. It collects pasts reconstructed from a present the instrument helps to constitute.

Kairos, Meadian

Mead's theory also illuminates kairos. The opportune moment, the turn of the tide when what was impossible becomes easy, is on Mead's account an emergent present: a moment in which something new appears that reorganises the possibilities. But the emergent present is not something that happens to the practitioner while she waits. She is part of it. Her waiting, her readiness and her visible attention are among the conditions from which the present emerges. And when it emerges, it reconstructs the past, including the past of her waiting, which becomes, in retrospect, "the patience that made the moment possible" or "the hesitation that nearly lost it". The practitioner's place in the emergent present is not that of a sailor watching the tide. It is that of a participant whose conduct helps constitute the present in which the moment emerges, and whose conduct is reinterpreted by it.

Mead and Anthro-Complexity: An Assessment

Anthro-complexity element Mead's contribution Effect
Element 2: coherence found, not presumed The self is a process, found in social interaction, not a presumed part Extends Element 2 to persons (Thesis I)
Element 8: retrospective coherence Every emergent present reconstructs its past Retrospect belongs to a present that includes the instrument
Element 10: kairos The emergent present includes the practitioner's conduct Waiting is part of what makes the moment (Thesis III)
Element 12: multiple identities Multiple me's called out by situations Identity relational; switching is the situation calling out an organisation of attitudes
Element 12: intentionality The internalised conversation; acts rehearsed against imagined responses Intention social before private
Element 12: abstraction The significant symbol; shared because each takes the other's attitude Abstraction relational; changes constraints only when taken up
Element 13: asymmetry principle Capacity to change stories belongs to the community; individual's share depends on position Asymmetry re-enters among humans (Theses II and V)
Element 16: Peircean pragmatism Social development of Peirce's semiotics; interpretant as the other's response A social Peirceanism available to anthro-complexity
Element 19: distributed sensing Self-signification as taking the instrument's attitude; I visible where categories fail Instruments should attend to the I, the novel response, not only the me

Why Anthro-Complexity Has Not Taken Mead

If Mead offers so much, why has anthro-complexity not taken him up? Several reasons can be suggested.

Mead belongs to CRP. Stacey, Griffin and Shaw made Mead central to complex responsive processes, and Mead is now associated with that school. Taking him up would mean entering CRP's territory, which the positional reason (Chapter 17) discourages.

Mead is a social psychologist, not a natural scientist. Anthro-complexity grounds itself in complexity science's natural-science origins. Mead offers an account of the human grounded in social psychology and philosophy, which does not carry the same authority in the debate with systems thinking.

Mead lacks a theory of constraint. Mead's account of the generalised other is an account of constitutive constraint, but it is not stated in the vocabulary of constraint, energy and timescale that anthro-complexity has developed with Juarrero. Mead has nothing to say about the energy cost of changing the generalised other, or about which parts of it are rigid and which elastic.

Mead's account is hard to instrument. The conversation of gestures, the I and the me, and the generalised other are processes, not variables. They are hard to measure at scale, and an instrument-building practice will find little in Mead to build with.

Each reason is real, and none is decisive. CRP does not own Mead. Social psychology is a science, and its findings (self-categorisation, shared intentionality) confirm Mead's account. Mead's lack of a theory of constraint is what Juarrero supplies, and the combination of Mead's generalised other with Juarrero's constraint energetics would be stronger than either. Mead's account is hard to instrument at scale, but the I, the novel response where the me's categories fail, is exactly what weak-signal instruments are trying to detect. An instrument designed with Mead in mind would look for the I in the places where signification hesitates.

The adversary will say that anthro-complexity does not need Mead, because Peirce already supplies everything Mead offers. The interpretant is the response, semiosis is social, and the community of inquirers is the generalised other. Mead is Peirce socialised, and anthro-complexity already has Peirce. Adding Mead would add a vocabulary, not a substance.

The adversary is partly right. Mead does build on Peirce, and much of what he offers is implicit in Peirce's semiotics and his theory of inquiry. But Peirce's focus is logic, signs and the method of science. He does not give an account of the self, of how a person is formed in social interaction and becomes a bearer of identities, intentions and abstractions. That is exactly what anthro-complexity's human cluster needs, because the human cluster is about persons. Peirce gives anthro-complexity a semiotics. Mead gives it a social psychology of the person, and without one the human cluster falls back into the possessive grammar of Thesis I. partial

THE MEANING OF THE GESTURE IS IN THE RESPONSE —

THE SELF IS THE COMMUNITY WITH A NOVEL EDGE —

EVERY PRESENT REWRITES ITS PAST.

Chapter 28

Elias: Figuration and the Closed Person

3,907 words, about 17 minutes of reading

The Sociologist of the Long Term

Norbert Elias (1897–1990) is the second thinker of Book Five. His work supplies what Mead's does not: an account of power as a dimension of every human relation, of long-term processes in which human figurations change over generations, and of the observer's place within the figuration she observes. These are the three things the November post's human cluster most needs and most lacks. Power is missing (Thesis V). The long term is reduced to kairos, the opportune moment. And the observer is outside (Theses II and III).

Elias's career was one of the strangest in twentieth-century scholarship. Born in Breslau, he studied medicine and philosophy and then sociology with Karl Mannheim. In exile he wrote The Civilizing Process (1939), whose first volume, published in Switzerland on the eve of war, was almost entirely ignored. He spent the following decades in obscurity in England, teaching adult education classes and then at the University of Leicester, and only in the 1970s, when The Civilizing Process was republished and translated, did he become recognised as one of the century's major sociologists. His late works, What Is Sociology? (1970), The Court Society (1969), Involvement and Detachment (1987) and The Society of Individuals (1987), developed the figurational sociology for which he is now known.

This chapter takes four of Elias's contributions in turn: the figuration and its power ratios; the game models; the established and the outsiders; and involvement and detachment. Each bears on anthro-complexity. The last, Elias's account of how detachment arises within involvement, offers an image of the practitioner that answers the estuary of Thesis III directly.

The Figuration

A figuration, in Elias's sense, is a web of interdependent people. The concept was designed to escape the dualism of "individual" and "society", which Elias regarded as a product of the closed-person self-image criticised in Chapter 19. There is no society apart from individuals, and no individual apart from her interdependencies. There are people bound to one another in figurations of varying size, density and structure: families, workplaces, cities, states, the global economy. The figuration is not a thing above the individuals, nor a sum of them. It is the pattern of their interdependence, which changes as they act and which shapes how they can act.

Power as balance

The most important feature of figurations, for this disputation, is that they are always shot through with power. But power, for Elias, is not a substance that some people possess and others lack. It is a structural characteristic of every relationship of interdependence. Wherever people depend on one another, each has some power over the other, to the extent that the other needs what she can give or withhold. A baby has power over its parents, because they need it to thrive. A worker has power over her employer, because the employer needs her labour. A ruler has power over subjects, and subjects over the ruler. Power is a balance, or in Elias's preferred term a power ratio: a relation in which the dependencies run both ways, unequally.

The power ratio changes as dependencies change. Elias's analysis of the civilising process traced how, in European history, lengthening chains of interdependence, through the division of labour, the monetisation of the economy and the growth of states, shifted power ratios in ways no one intended. Kings became more dependent on bureaucrats, nobles on courts, and merchants on markets. The shifts changed people's conduct and, over generations, their very personality structure. The increasing self-restraint and the internalisation of social controls that Elias documented in manners, from table etiquette to the management of bodily functions to the control of aggression, were responses to figurations in which people depended on more and more others, so that impulsive conduct became more dangerous and foresight more necessary.

What the figuration lends anthro-complexity

The figuration supplies anthro-complexity with three things.

A relational account of the person. Persons are nexuses in figurations, and their character, capacities and identities are formed by their position in the web and its history. This completes Mead's social self with a sociology of the webs in which selves are formed.

A theory of power as dimension. Power is not an add-on to human relations. It is a characteristic of every relation of interdependence. A theory of human complexity that includes interdependence, as anthro-complexity's does, must include power, because interdependence is the ground of power ratios.

A theory of unintended long-term change. Figurations change in directions no one intends, through the accumulated effects of interdependent actions. That is emergence, in complexity terms, and Elias's account of it is one of the richest in the social sciences. It is historically grounded and attentive to the slow transformation of personality and institution over centuries.

The Game Models

In What Is Sociology? Elias developed a series of game models to show how power ratios and figurations work. The models are thought experiments of increasing complexity.

The primal contest. Two groups compete for scarce resources without any rules. Each group's actions are determined by its expectations of the other's actions. Neither controls the outcome, which emerges from the interplay of their moves. Even here, with no rules, the outcome cannot be explained by either group's intentions alone.

Two-person games. Two players play a game with rules. If one player is much stronger, he largely controls the game's course, and his intentions explain much of what happens. As the players become more equal in strength, neither controls the game, and its course increasingly emerges from the interplay of moves. The game becomes, as Elias put it, less and less explicable by the plan of either player and more and more explicable by the game process itself.

Multi-person games at one level. Many players play against one another. As their number increases, it becomes harder for any player to see the whole game, to anticipate the moves of all the others, or to control its course. The game's process takes on its own dynamics, opaque to the players.

Multi-tier games. Players are organised in tiers. A small group at the top plays on behalf of larger groups below, who play on behalf of still larger groups. The top players seem to control the game, but they depend on the lower tiers, who can withdraw support, press demands and play their own games. As the figuration grows, the top players' control diminishes even as their apparent power increases. They become representatives of a process they cannot master.

The lesson Elias drew is the one The Compass and the Loom summarised: in long chains of interdependence, no player controls the game, though powerful players feel they stand above it. The feeling of standing above the game is an illusion produced by position. The top player sees more of the game than others and has more influence over it, so she feels in control. But her control is limited by her dependence on players she cannot see, and the game's process outruns her plans.

The game models and anthro-complexity

The game models offer a precise account of the steersman that The Compass and the Loom found in the instruments it examined. The steersman stands outside the system, reads it and chooses whether to let emergence happen or to stimulate it. Elias would say the steersman is a top player in a multi-tier game, who feels she stands above the game because of her position, and who is in fact a representative of a process she cannot master. Her readings of the system are moves in the game, read by other players. Her choice to "let emergence happen" is a move, a withholding that others interpret. Her choice to "stimulate emergence" is another move, which others respond to in ways she cannot foresee.

The game models also explain why the steersman's feeling persists. It is produced by position, not by error. A top player really does see more and have more influence than others. The illusion is not that she has influence, but that her influence amounts to control, and that she stands outside the game. Instruments that serve top players will tend to reproduce the illusion, because they give the top player more to see, and seeing more feels like standing above.

The Established and the Outsiders

The Established and the Outsiders (1965), written with John Scotson, is Elias's most concrete study, and the one most directly relevant to anthro-complexity's narrative instruments. It examined a small community on the edge of Leicester, called "Winston Parva" in the study, consisting of three neighbourhoods. Two were working-class, and their residents were similar in income, occupation, education and ethnicity. The only significant difference between them was length of residence. One neighbourhood, the "village", was settled by families who had lived there for generations. The other, the "estate", was settled by newer arrivals.

Despite the absence of any material difference, the established residents of the village regarded the newcomers of the estate as inferior: rougher, dirtier, less respectable, more prone to delinquency. The newcomers, remarkably, largely accepted the judgement and regarded themselves as inferior. The established excluded the outsiders from local associations, clubs and positions of influence, and the outsiders accepted their exclusion.

Elias and Scotson's explanation turned on cohesion and gossip. The established were more cohesive. Their families had known one another for generations, they shared networks of kinship and friendship, and they could coordinate their attitudes and actions. Their cohesion gave them power. They could control the local associations, maintain a shared view of who was respectable and enforce it. Gossip was the mechanism. The established circulated praise gossip about themselves, which attributed the virtues of their best members to the whole group, and blame gossip about the outsiders, which attributed the vices of their worst members to the whole group. Elias called this the attribution of group charisma to the established and group disgrace to the outsiders. The outsiders, less cohesive and lacking networks through which counter-gossip could circulate, could not resist the stigmatisation, and internalised it.

Narrative as power

The Established and the Outsiders is a study of narrative as power. Gossip is narrative: stories about who people are, what they do and what they deserve. The established's power lay in their capacity to circulate stories that were taken up (praise of themselves, blame of the outsiders) and to block stories that might have challenged them. The outsiders' powerlessness lay in their incapacity to circulate counter-stories, because they lacked the cohesion through which stories spread.

This is anthro-complexity's third mark, abstraction and narrative that change constraints, with the power dimension restored. The established changed the constraints of Winston Parva's social life (who could join the club, who could hold office, who was trusted) by telling stories that were taken up. The outsiders could not, not because they lacked the human capacity for narrative, but because they lacked the figurational position from which narratives are taken up. The capacity to change the story is a human capacity. Its exercise depends on cohesion, network and position: on power ratios.

The established and the outsiders in a SenseMaker study

Imagine a SenseMaker study in an organisation with an established group, long-serving staff in a core function, and an outsider group, recent hires, contractors or a function absorbed in a merger. Both groups share micro-narratives and signify them. What patterns emerge?

Elias and Scotson's findings predict several things. The established's narratives will reflect their group charisma: stories of competence, commitment and "how things are done here". The outsiders' narratives may reflect internalised group disgrace, stories of struggling to fit in or being out of their depth, or they may reflect resistance. Either way, the patterns will show two populations with different experiences. An analyst reading the patterns without Elias might conclude that the established are more engaged and competent and the outsiders less so. That conclusion would reproduce the established's praise and blame gossip with the authority of an instrument. The patterns would be read as facts about the groups, when they are traces of a power ratio, the established's capacity to define what counts as competence, which the outsiders have partly internalised.

An analyst reading with Elias would ask different questions. Which group's stories are taken up in the organisation's self-description? Whose vocabulary do the signifiers reflect? Do the outsiders' stories contain counter-gossip that the signifiers cannot capture? Is the instrument's pattern a description of the organisation, or a description of the established's description of it?

Involvement and Detachment

Elias's essay on involvement and detachment (published in English in 1956 and expanded in the 1987 book) addresses the observer's place in the figuration directly. It offers the image that answers Thesis III's estuary.

Elias's starting point is that all human knowledge lies on a continuum between involvement, knowledge shaped by the knower's emotions, interests and immediate stakes, and detachment, knowledge relatively free of them. Neither pole is ever reached. Pure involvement would be knowledge wholly governed by fear and desire, and pure detachment would be a view from nowhere. All knowledge is a blend. The natural sciences have achieved a relatively high degree of detachment about natural processes over centuries, through institutions and methods that discipline involvement. The social sciences have achieved less, because their objects are people, and the people who study them are inside the same figurations and have stakes in them.

Elias's central claim is that the degree of detachment possible depends on the degree of control over the dangers the knowledge concerns. When people are in great danger, they cannot be detached. Fear governs their perception, and their knowledge is fantasy-laden. When they have some control over a danger, they can afford more detachment, and more detached knowledge gives them more control. There is a spiral. Detachment enables control, which enables detachment. Or, in the reverse direction, involvement produces fantasy, which reduces control, which increases danger, which increases involvement.

The fishermen in the maelstrom

Elias illustrated the spiral with Edgar Allan Poe's story "A Descent into the Maelström" (1841). Two brothers, fishermen off the coast of Norway, are caught in a great whirlpool. Their boat is drawn into the vortex, circling down towards the abyss. One brother is paralysed by terror. He clings to a ring-bolt on the boat and goes down with it. The other, after his first terror, begins to observe. He notices that objects in the whirlpool descend at different rates. Large objects go down faster, and small cylindrical ones go down slowest, some of them not reaching the bottom before the whirlpool subsides. He reasons from his observations, lashes himself to a cylindrical water cask, and throws himself into the water. The boat, with his brother, goes down. The cask descends slowly, the whirlpool subsides, and he is saved.

Elias drew the moral. The brother who survived did not escape his involvement. He was inside the maelstrom throughout, at the greatest possible risk. But within his involvement he achieved a degree of detachment, enough to observe, compare and reason, and his detachment gave him a measure of control. He did not stand outside the danger and read it as a sailor reads a tide. He was in it, and he read it from inside, and his reading was part of what saved him. Detachment arose within involvement, not as an escape from it.

The fisherman and the sailor

Set the fisherman in the maelstrom beside the sailor in the estuary.

The sailor (estuary, Thesis III) The fisherman (maelstrom, Elias)
Position On deck, outside the medium In the vortex, inside the danger
Relation to medium Reads it; waits for it to turn Is carried by it; observes from within
Stakes The voyage Survival
Detachment Given by position Achieved within involvement
Knowledge Of the medium's propensity Of the medium's propensity and his own place in it
Action Waits, then moves Observes, reasons, commits himself to the medium differently
What saves him Timing Detachment within involvement

Both images picture a practitioner reading a complex, dangerous medium of water. But the fisherman is inside, and his detachment is an achievement of involvement rather than a privilege of position. He does not read the maelstrom's propensity as if he were not in it. He reads it as someone in it, and his reading includes his own position and the objects he might bind himself to. That is the image of the practitioner a reflexive anthro-complexity needs: not the sailor on the deck, but the fisherman in the vortex, whose detachment is an achievement within involvement and whose reading includes his own place.

The Compass and the Loom's Mariner inside the whale is the same figure in a different story. He is inside the system, with existential stakes, and intervenes by changing a constraint from within.

Homo Clausus Revisited

Chapter 19 introduced Elias's critique of the closed person. With the figuration, the game models, the established and outsiders, and involvement and detachment in hand, the critique can be seen in its full force.

The closed person is not merely a philosophical error. It is a self-image produced by a particular figuration, the long civilising process of European societies, in which lengthening chains of interdependence required ever greater self-restraint, and the self-restraint was experienced as an inner wall separating the "true self" inside from the world outside. The self-image is real as experience. People in such societies really do feel that they are sealed containers facing a world. But it is false as description. The person is not sealed. Her self-restraint is the internalisation of figurational pressures, her inner self is formed by her interdependencies, and the wall she feels is made of the social controls she has absorbed.

The self-image has a further consequence that bears on the observer. The closed person experiences herself as standing outside the world she faces, looking at it across the wall. That experience is the experiential root of the outside observer. The scientist who feels she observes nature from outside, the manager who feels she reads the organisation from above, and the consultant who feels she maps the system from a neutral vantage are all expressing the closed person's sense of separation. Elias's critique of the closed person is therefore also a critique of the outside observer. Both are produced by the same figuration, and both are illusions of position.

Anthro-complexity's possessive grammar (Thesis I) and its estuarine images (Thesis III) are, on Elias's account, two expressions of one underlying self-image. The person who holds identities as possessions inside her wall, and the practitioner who reads the system from the deck of her vessel, are both the closed person of Elias's account: the sealed individual facing a world across a boundary. Elias's open people, open people formed in figurations, are at once the relational persons that Thesis I's remedy calls for and the involved observers that Thesis III's remedy calls for.

Elias and Anthro-Complexity: An Assessment

Anthro-complexity element / thesis Elias's contribution Effect
Element 2: coherence found Figurations are patterns of interdependence, neither wholes nor sums Extends Element 2 to persons and to power
Element 5: no normal; norms favour definers Power ratios; established define respectability Gives the eurhythmia critique a sociology
Element 12: abstraction and narrative Praise and blame gossip; group charisma and disgrace Narrative's constraint-changing power depends on cohesion and position
Element 13: asymmetry principle The established can change the story; the outsiders cannot Asymmetry re-enters among humans as power ratio (Thesis V)
Element 19: distributed sensing Patterns may reproduce the established's gossip Instruments must ask whose description the pattern describes
Thesis I: possessive fallacy the closed person as self-image of a figuration The possessive grammar is a historical self-image, not ontology
Thesis III: estuaries The fisherman in the maelstrom; detachment within involvement The practitioner's image should be the fisherman, not the sailor
Thesis V: power Power as balance in every interdependence Power belongs in the theory of the human as dimension
The Compass and the Loom: the steersman Game models: top players feel above the game The steersman's illusion is produced by position, and instruments that serve top players reproduce it

Why Elias Matters More Than Mead for This Disputation

Mead supplies the social self, the gesture and the generalised other: the micro-foundations of a relational account of the human. Elias supplies what Mead does not: power, history and the observer. For this disputation, Elias matters more, because the eight theses of Book Four converge on a single point, that the human distinction re-enters among humans as an asymmetry between describers and described, and Elias's sociology is a sociology of exactly that re-entry. The established describe the outsiders. Top players describe the game. The involved observer describes the maelstrom from inside it. In each case, Elias asks the question anthro-complexity's instruments need to ask: from where is this description made, by whom, with what power, and with what consequence for the described?

Elias also offers something no other thinker in this disputation offers so fully: a sense of the long term. Anthro-complexity's temporal vocabulary is kairos, the opportune moment, the turn of the tide. Elias's is the civilising process, the slow transformation of figurations and personalities over centuries. The two are complementary. Kairos is the moment, and the long term is the process within which moments occur. A practice that attends only to kairos risks treating every situation as a tide to be caught, when some situations are phases of long-term processes that no tide will turn. The shift of power ratios in an organisation over decades, the slow internalisation of a new professional norm, the gradual erosion of trust between established and outsiders: these are not tides. They are civilising and decivilising processes, and they need a different kind of attention.

The adversary will say that Elias's sociology is descriptive and historical, and offers no guidance for intervention. Elias himself was famously reluctant to draw practical conclusions, and he regarded sociology's task as understanding, not reform. A practice needs to act, and Elias offers no method of action. Anthro-complexity's instruments are imperfect, but they are instruments. Elias offers only understanding.

The adversary is right that Elias offers no method, and right that he was reluctant to prescribe. But the fisherman in the maelstrom is a method, or at least the image of one: achieve detachment within involvement, observe your own place, find the object you can bind yourself to that will descend slowly. And The Established and the Outsiders is a diagnostic: before reading a pattern as a fact about groups, ask whether it is a trace of a power ratio. Neither is an instrument. Both are disciplines for the use of instruments, and anthro-complexity's instruments need such disciplines. Elias supplies the understanding without which the instruments reproduce what they should reveal. partial

NO PLAYER CONTROLS THE GAME —

THE ESTABLISHED TELL THE OUTSIDERS WHO THEY ARE —

THE FISHERMAN SURVIVES BY OBSERVING FROM INSIDE THE VORTEX.

Addition: Cross-Lights From Book Nine

Foulkes. Elias helped to found the Group Analytic Society with S. H. Foulkes, whose matrix is the group as a web of communication in which every member, the analyst included, is a node. The Zen Garden study sets Foulkes's web beside Stacey's grid as two matrices with one ancestor. Flinn's practice of the large group, in Chapter 57, descends from the same line through Patrick de Maré.

Outsight. De Maré's word for what a large group can gain is close to Elias's detachment within involvement: a shared view, from inside, of the dominating and silencing habits forming among those present.

Chapter 29

Juarrero: Three Kinds of Constraint

4,143 words, about 18 minutes of reading

The Philosopher in the Engine Room

Alicia Juarrero is the thinker on whom anthro-complexity's ontology most depends, and the one this disputation has praised most. Thesis IV called her theory of constraint anthro-complexity's greatest achievement. The achievement is jointly Juarrero's and Snowden's, worked out in collaboration over many years. This chapter reads Juarrero more closely, and finds that her work offers anthro-complexity something more than its theory of constraint. It offers a dynamical account of intention, which grounds the human cluster in constraint and context, not in possession. If anthro-complexity followed Juarrero all the way, it would arrive at the relational account of the human this disputation has argued for, by its own route and in its own vocabulary.

Dynamics in Action

Juarrero's Dynamics in Action: Intentional Behavior as a Complex System (1999) is a work of philosophy of action. Its question is old. How do intentions cause behaviour? When I intend to raise my arm and my arm rises, what is the relation between the intention and the movement? The tradition offered two unsatisfactory answers. On the first, descended from Descartes, the intention is a mental event that causes a physical event by some mysterious interaction between mind and body. On the second, descended from Hume and developed in analytic philosophy of action by Davidson and others, the intention is a cause in the ordinary sense, an event that precedes and produces the movement. But then it becomes hard to see how the content of the intention, what I intend, makes a difference, since causes in the Humean sense act by their physical properties, not their meanings.

Juarrero's proposal was that both answers rest on a mistaken model of causation, the Newtonian model of efficient causes acting by contact and force. Complex dynamical systems show a different kind of causation. Higher-level patterns constrain lower-level processes. They change the probability distribution of what the lower-level processes do, without pushing them. A whirlpool constrains the movement of the water molecules that compose it. It does not push each molecule. It shapes the space of possible paths, so that molecules move in a vortex. The whirlpool is made of the molecules and also constrains them. Causation runs both ways, upward from molecules to whirlpool and downward from whirlpool to molecules.

Intentions, Juarrero argued, are like this. An intention is a higher-level dynamical pattern in the brain and body, an attractor in the space of neural and behavioural possibilities, which constrains lower-level processes so that behaviour flows in a certain direction. The intention to raise my arm does not push my muscles. It reshapes the landscape of possible neural and muscular activity, so that raising the arm becomes highly probable and other movements improbable. The content of the intention matters, because the content is the shape of the constraint. Intending to raise my arm and intending to wave constrain the landscape differently, and the difference shows in behaviour.

Intention as context-sensitive constraint

The decisive point is that intentions, on Juarrero's account, are context-sensitive constraints. An intention does not specify a fixed behaviour. It specifies a pattern that is realised differently in different contexts. My intention to greet a friend is realised as a wave across the street, a handshake in the office or an embrace at the airport, depending on context. The intention is a constraint that makes the behaviour context-dependent. Context, in turn, includes the social context: the friend, the setting, the conventions of greeting, the expectations of onlookers.

Juarrero also emphasised that intentions are embedded. They arise within a history of interaction with the environment, including the social environment, and they carry that history. An intention is not a self-generated act of will, arising from nowhere inside a sealed self. It is a dynamical pattern that has formed through the person's history of engagement with the world and others, and that is continually reshaped by ongoing engagement. In Dynamics in Action, Juarrero drew on the dynamical systems tradition, on hermeneutics (the history of interpretation within which meanings are formed) and on the embeddedness of action in social practice.

What this means for the second mark

Juarrero's account of intention bears directly on anthro-complexity's second mark of the human: humans act with intentionality. On the possessive grammar of the November post, intentionality is a capacity humans have and act with. On Juarrero's account, intention is a dynamical constraint, formed through history, sensitive to context and embedded in social practice. It is not a possession. It is a pattern in a dynamical system that includes the body, the environment and other people.

That is a relational account of intention, stated in the vocabulary of constraint and dynamics, by the philosopher whose theory of constraint anthro-complexity has adopted. Anthro-complexity does not need to go to Mead, Tomasello or Elias for a relational account of intention, though their accounts agree. It has one in Juarrero, its own source. Followed through, the theory of constraint grounds the human cluster relationally.

Context Changes Everything

Juarrero's later book, Context Changes Everything: How Constraints Create Coherence (2023), generalises the account of Dynamics in Action from intentional action to coherence in complex systems of every kind. Its central argument is that coherence, the existence of stable, integrated wholes that are more than collections of parts, arises through constraints, and specifically through the interplay of different kinds of constraint over time.

The book's distinctions can be stated briefly, in the form in which anthro-complexity uses them.

Context-independent constraints narrow possibilities regardless of context. They are the walls, rules and boundaries of a system. They do not create coherence by themselves, because they do not make events interdependent. They only limit them.

Context-dependent (context-sensitive) constraints make the probability of one event depend on others. They link events together, so that the system's behaviour becomes interdependent and patterns can form. They are the source of coherence.

Enabling constraints open possibilities that did not exist before. A catalyst enables a reaction, a common language enables communication and a protocol enables coordination. Enabling constraints are usually context-sensitive: they link events into new patterns.

Governing constraints regulate and stabilise patterns once they have formed. They keep a coherent system within the range in which it remains coherent. A thermostat governs temperature, and a constitution governs politics.

Juarrero's account of how these interact is subtle. A complex system forms when enabling constraints link events into patterns. The patterns, once formed, impose governing constraints that stabilise them. The governing constraints then become the context within which new enabling constraints can arise, and so on. Coherence is not a state. It is a process of constraint formation and transformation, in which each layer of constraint is the context for the next.

Catalysts and constructors

Context Changes Everything discusses catalysts as enabling constraints, and it is here that Snowden's interest in constructors finds a connection. A catalyst enables a reaction without being consumed. That is close to Deutsch and Marletto's constructor. But Juarrero's catalyst is a constraint: it shapes probabilities. It is not an agent. And Juarrero is careful that catalysis in complex systems is often autocatalytic: the products of a reaction catalyse further reactions, so that the catalysts are themselves produced by the system. In autocatalytic sets (Kauffman's model for the origin of life, Chapter 7), no catalyst is external and unchanged. Each is produced and maintained by others. That is constraint closure (Chapter 22). Juarrero's catalysts in complex systems are not constructors in the strict sense. They are participants in a closure, produced by the system they enable.

The connection undercuts the import of constructor theory examined in Thesis IV. Anthro-complexity's own source on constraints treats catalysts in complex systems as produced by the system, not as external agents that persist unchanged. If rituals are catalysts in Juarrero's sense, they are produced and reproduced by the communities they enable, changed by their performance and part of a closure of constraints. That is Giddens's duality of structure and Turner's performance, in Juarrero's vocabulary. Constructor theory's unchanged persistence is not needed and does not fit.

Coherence and identity

The most important theme of Context Changes Everything for this disputation is identity: what makes a coherent system the same system over time. Juarrero's answer is that identity is not a substance or an essence but a pattern of constraints that persists while its components change. A whirlpool persists while its water molecules are replaced. An organism persists while its cells turn over. A community persists while its members come and go. What persists is the pattern of constraints: the way components are linked, the attractors that shape their behaviour, and the governing constraints that stabilise the whole.

Apply this to persons. A person's identity, on Juarrero's account, is a pattern of constraints that persists while its components change: neural states, bodily cells, beliefs, relationships, roles. The pattern includes constraints that link the person to others. Her relationships, roles and community memberships are not external to her identity. They are among the constraints that constitute it. Her identity is context-dependent. It is realised differently in different contexts, as her intention to greet is realised differently in different settings. And it is historical, formed through a history of constraint formation and transformation.

That is anthro-complexity's first mark, multiple identities, stated in Juarrero's terms. A person has multiple identities because her identity is a pattern of context-sensitive constraints, realised differently in different contexts. The multiplicity is not a collection of stored identities. It is the context-sensitivity of a single dynamical pattern. And because the constraints include relationships with others, identity is relational. It is constituted partly by the constraints that link the person to others, and changes when those constraints change.

The Missing Third: Constitutive Constraint

The Compass and the Loom, in its second book, identified constitutive constraint, the constraint that makes a community what it is, as the missing third kind beside Snowden's governing and enabling constraints. Juarrero's work supplies the materials for the concept, even where its practice vocabulary does not name it.

On Juarrero's account, a coherent system's identity is a pattern of constraints. Some of these constraints constitute the system: they are the ones without which it would not be this system. A whirlpool's constitutive constraints are the circular flow and the pressure gradient. Remove them and there is no whirlpool, only water. A community's constitutive constraints are the patterns of relation, practice and shared meaning without which it would not be this community: its sense of who belongs, its rituals, its shared history and its implicit norms. Remove them and the people remain, but the community does not.

Constitutive constraints differ from enabling and governing constraints in one crucial respect. They are the system's own identity, not something the system has or is subject to. A governing constraint can be imposed on a system from outside, as a regulator imposes a rule on a market. An enabling constraint can be introduced, as a facilitator introduces a protocol. A constitutive constraint cannot be imposed or introduced from outside, because it is what the system is. It can only be reproduced or transformed by the system itself, in its ongoing activity. That is Giddens's duality of structure again: constitutive constraints are reproduced in the practices they constitute.

This is why estuarine mapping (Thesis IV) reproduces constitutive constraints in the act of mapping. The mapping session is a practice of the community, and in it the community's constitutive constraints (who speaks, whose judgement stands, what is taken for granted) are reproduced. The map records governing and enabling constraints, and the session reproduces constitutive ones. A map that does not see this records the community's surface and reproduces its depth.

Juarrero Meets Mead and Elias

Juarrero's account and the social-relational accounts of Mead and Elias converge, and the convergence is significant. They come from different traditions (dynamical systems philosophy, pragmatist social psychology, figurational sociology) and arrive at a common picture.

Question Juarrero Mead Elias
What is a person? A persisting pattern of context-sensitive constraints, including relational constraints A social process: the relation of I and me, formed in the conversation of gestures A nexus in a figuration, formed by interdependence over time
What is intention? A dynamical constraint, formed through history, sensitive to context, embedded in practice An act rehearsed in the internalised conversation, against imagined responses An orientation formed in figurations, shaped by power ratios and long-term processes
What is identity? The pattern that persists while components change; realised differently in contexts Multiple me's, called out by situations Position in figurations; formed by the history of interdependence
How do higher levels act on lower? Downward causation through constraint The generalised other in self-control Figurational pressures internalised as self-restraint
Is the person presumed? No; found as a persisting pattern No; arises in social experience No; formed in figurations (open people)
Is the person dissolved? No; the pattern is real and distinct No; the I is irreducibly individual No; individuals are intensely real

The convergence shows that the relational account of the human is not a parochial preference of CRP, or a borrowing from social theory that anthro-complexity's natural-science grounding could reject. It is what the dynamical systems account of persons, anthro-complexity's own account, implies when followed through. Juarrero, Mead and Elias agree that persons are neither presumed parts nor dissolved relata. They are persisting patterns, formed in relation, realised in context, and changed over time. That is the position anthro-complexity's Element 2 requires and its human cluster, in the possessive grammar, does not hold.

Juarrero and the Theses

Juarrero's work bears on several of the eight theses, and in each case it supports the thesis from within anthro-complexity's own resources.

Thesis I (possessive fallacy). Juarrero's account of intention as a dynamical constraint and identity as a persisting pattern grounds the human marks relationally, in anthro-complexity's own vocabulary. The possessive grammar is not required by the theory of constraint, and the theory of constraint does not support it.

Thesis II (asymmetry applied to itself). Juarrero's account of downward causation through constraint means that a description adopted by a community becomes a constraint on the community's behaviour. It changes the probability distribution of what members do. Anthro-complexity's descriptions, adopted, become constraints. The asymmetry principle applied to itself is, in Juarrero's terms, the observation that the practitioner's descriptions are constraints in the system she describes.

Thesis III (estuaries). Juarrero's whirlpool is the fisherman's maelstrom, a coherent pattern of constraints in water. But Juarrero's interest is in what the whirlpool is made of: molecules constrained by the pattern they compose. The practitioner reading a human system is one of the molecules. Her reading constrains the pattern, as the pattern constrains her reading. The estuary's sailor stands outside the water. Juarrero's account places every component inside the pattern.

Thesis IV (constitutive constraint). Juarrero's account of identity as a pattern of constraints supplies the concept of constitutive constraint, and her treatment of catalysts in complex systems as produced by the system undercuts the import of constructor theory.

Thesis V (power). Juarrero's work does not foreground power, and here it needs Elias. But her account of constraints as changing probability distributions is a natural vocabulary for power. Power is the capacity to impose constraints on others' probability distributions, to make some of their actions more likely and others less. Elias's power ratios are the balance of such capacities in interdependence. A Juarrero-Elias synthesis would state power as the asymmetric capacity to constrain, and would place it within the theory of constraint, where anthro-complexity could use it.

The Bridge

This chapter's conclusion is that Juarrero is the bridge between anthro-complexity as it is and anthro-complexity as this disputation proposes it could be. Anthro-complexity has adopted Juarrero's theory of constraint as its ontological foundation. It has used the theory to ground its account of complex systems: coherence through constraint, downward causation and constraint energetics. It has not used the theory to ground its account of the human. The human cluster is stated in the possessive grammar, as capacities persons have, not in Juarrero's dynamical grammar, as patterns of context-sensitive constraint formed in relation.

Crossing the bridge requires no foreign import. It requires applying Juarrero's theory of constraint to persons, as Juarrero herself did in Dynamics in Action. Intentions are constraints. Identities are persisting patterns of constraint. Abstractions are constraints that, once adopted, change the probability distributions of a community's behaviour. All three are context-sensitive, historically formed and relationally embedded. The three marks of the human, restated in Juarrero's terms, are three aspects of the human as a pattern of constraints in relation. The asymmetry principle, restated, is the observation that some such patterns can form and adopt constraints about the constraints, the descriptions that change the system, and that this second-order constraint formation is what distinguishes human systems in kind.

On the other side of the bridge is a reflexive anthro-complexity, grounded in its own ontology and stated in its own vocabulary, that has absorbed the relational insights of Mead, Elias, Tomasello and Searle without needing to adopt their vocabularies. Book Eight describes it.

The adversary will say that this chapter puts words into Juarrero's mouth. Juarrero's work is about constraints and coherence in complex systems in general, and her collaboration with Snowden has focused on how constraints work in organisations. She has not endorsed a relational account of the human in the sense of Mead or Elias, nor criticised anthro-complexity's human cluster. To claim that her work "implies" the relational account is to enlist her in an argument she has not made.

The adversary is right that Juarrero has not made this argument, and the chapter does not claim she has. It claims that her account of intention in Dynamics in Action (dynamical, context-sensitive, historically formed and embedded in social practice) and her account of identity in Context Changes Everything (a persisting pattern of constraints, including relational ones) are compatible with, and in the chapter's reading support, a relational account of the human. That is a reading of her work, which she may reject. If she does, the argument would need to rest on Mead, Elias and the science of Chapter 13 alone, and they are sufficient. The chapter's claim is that anthro-complexity's own philosopher offers a route to the relational account, and that the route deserves to be taken. partial

INTENTION IS A CONSTRAINT —

IDENTITY IS A PATTERN THAT PERSISTS WHILE ITS PARTS CHANGE —

THE BRIDGE IS ALREADY BUILT; IT ONLY NEEDS CROSSING.

Addition: Constraint and Power, a Synthesis

The chapter noted that Juarrero's work does not foreground power, and suggested that a Juarrero–Elias synthesis would state power as the asymmetric capacity to constrain, and place it within the theory of constraint, where anthro-complexity could use it. This addition sketches the synthesis.

Power as constraint on others' probability distributions

Juarrero defines a constraint as anything that changes the probability distribution of events. Elias defines power as a structural characteristic of relationships of interdependence, a balance in which each party has some capacity to affect what the other can do. Put the two together, and power can be stated in Juarrero's vocabulary:

Power, in the vocabulary of constraint. A has power over B to the extent that A can impose, maintain or remove constraints on B's probability distribution of actions, and B cannot do the same to A to the same extent. Power is asymmetric constraint capacity within interdependence.

The definition has several features.

It is relational. Power is not a property of A. It is a property of the relation between A and B, of their asymmetric capacities to constrain each other. This is Elias's power ratio.

It is graded. Both A and B have some capacity to constrain each other. A baby constrains its parents, and a worker her employer. Power is the asymmetry, and asymmetries come in degrees.

It covers Lukes's three dimensions. First dimension: A constrains B's decisions directly, for instance by forbidding an action. Second dimension: A constrains the agenda, removing options from B's probability distribution before B considers them. Third dimension: A shapes B's preferences, changing the probability distribution of B's wants as well as B's actions. All three are constraints on B's probability distributions, at different levels.

It covers Foucault's visibility. Seeing without being seen is a capacity to constrain. The one who sees can anticipate and respond to the other's actions, constraining them, while the one seen cannot do the same. Asymmetric visibility is asymmetric constraint capacity.

It covers Bourdieu's symbolic power. Supplying the categories through which others perceive is constraining their probability distributions at the level of perception. What is not categorised is less likely to be noticed.

Constitutive constraint and power

The synthesis bears especially on constitutive constraints (Thesis IV). Constitutive constraints make a community what it is, and they are reproduced in compliance. But they are not neutral. They distribute constraint capacity. The established in Winston Parva (Chapter 28) were advantaged by the community's constitutive constraints: the shared view of who was respectable, carried in everyone's habitus, including the outsiders'. The constitutive constraints were the power ratio, reproduced in every act of deference and every piece of gossip.

So mapping constitutive constraints is mapping power. When estuarine mapping fails to see constitutive constraints, it fails to see the distribution of constraint capacity, the power ratios that determine who can shift which constraints and at whose cost. When it records "the budget cycle is fixed" without recording "the senior manager's judgement that it is fixed went unchallenged" (Chapter 22), it records a governing constraint and misses the constitutive constraint, deference to senior managers on finance, that is the power ratio.

Energy cost as power

The synthesis also bears on estuarine mapping's central axis, energy cost. The energy needed to change a constraint is not a property of the constraint alone. Chapter 22 argued that it is a property of the constraint's position in a closure. The synthesis adds that it is also a property of who is trying to change it. A constraint that costs a junior employee enormous energy to change, because she lacks the authority, the networks and the standing, may cost a senior executive almost nothing, because her position gives her constraint capacity over it. Energy cost is relative to the changer's position in the power ratio.

An estuarine map that places constraints on a single energy axis, as if their cost were the same for everyone, has averaged over the power ratio and hidden it. A power-aware map would show energy cost for whom: this constraint is cheap for leadership, costly for middle management and impossible for the frontline. The map would then show not only where change is possible but who can make it. It would also show, by implication, who cannot. That is the information those without power most need, and those with power least wish to see displayed.

The synthesis displayed

Concept Juarrero Elias Synthesis
Constraint Changes probability distributions — —
Power — Balance in interdependence Asymmetric constraint capacity
Constitutive constraint Pattern that makes a system what it is Figuration; habitus of established and outsiders The power ratio carried in everyone's practice
Energy cost Cost to change a constraint — Cost relative to the changer's position
Change Shift constraints Shift power ratios over time Shift constraints, and know whose constraint capacity the shift increases

Why the synthesis matters

The synthesis matters because it lets anthro-complexity include power in its theory without importing a foreign vocabulary. The adversary's Article 5 warned against power as master concept, against theories that see only power. The synthesis does not make power the master concept. It makes power a kind of constraint, asymmetric constraint capacity within interdependence, and so places it within the theory anthro-complexity already holds. Power becomes one more thing the estuarine map can show, one more dimension of the constraint field, not a lens that replaces the field.

This is how Thesis V's remedy can be achieved from inside anthro-complexity. Power belongs in the theory of the human as a dimension. In Juarrero's vocabulary, the dimension is the asymmetry of constraint capacity, and the theory of constraint already has the concepts to represent it.

POWER IS ASYMMETRIC CONSTRAINT CAPACITY —

CONSTITUTIVE CONSTRAINTS CARRY THE POWER RATIO —

ENERGY COST IS COST FOR WHOM.

Addition: Cross-Lights From Book Nine

A caution from Snowden. The post on Luhmann objects to a pairing of Juarrero with Luhmann and calls it misappropriation, on the ground that her account puts the human agent at the centre and his puts the human in the environment. The rule it gives is that a concept moved between fields needs its own justification. This chapter's proposal to state power in Juarrero's vocabulary is a reading she may not accept, as the chapter says. The rule applies to it.

Decision premises. The same post allows that Luhmann's decision premises and her constraints deserve closer comparison. This chapter suggests that the comparison be made by position: whose premise, and at what cost to whom.

Chapter 30

Ingold: Meshwork Against Stratigraphy

3,299 words, about 14 minutes of reading

The Anthropologist of Lines

Tim Ingold appears in the November post twice. He is part of the shared ground ("Gibson, Ingold and the ecological turn"), and Snowden notes that he has written separately in the anthro-complexity series on reading Ingold. Ingold is also one of the three steps by which Ing's programme downplays what is distinctive about humans. Ing follows Ingold in treating humans as one kind of animal. Snowden accepts Ingold's ecological anthropology as a source and rejects its use to flatten the human. In Snowden's reading, Ingold is a neighbour taken with reservations.

This chapter reads Ingold for what he offers the disputation. Two contributions matter. The first is his distinction between meshwork and network, and the related distinction between wayfaring and transport, or the wayfarer and the navigator. These give the sharpest available critique of the stratigraphic, map-from-above mode of representation that The Compass and the Loom found in WRAS's substrate and that this disputation finds in the estuarine map. The second is his concept of correspondence, which offers a model for what narrative instruments could be: practices of answering, not capturing. The chapter also examines Snowden's reservation, that Ingold flattens the human, and argues that Ingold's relational ontology can be combined with anthro-complexity's insistence on human reflexivity, so that each corrects the other.

Ingold's Project

Ingold trained as a social anthropologist and did fieldwork among the Skolt Saami of north-eastern Finland, reindeer herders whose relations with animals, land and weather shaped his whole subsequent thought. Over four decades, in The Perception of the Environment (2000), Lines (2007), Being Alive (2011), Making (2013), The Life of Lines (2015) and Correspondences (2020), he developed an anthropology that refuses the separation of humans from the world they live in. Its starting point is not the mind facing a world, nor culture imposed on nature, but the organism-person growing and moving within an environment that is itself growing and moving. Knowledge is not representation but skilled practice, learned through attention and participation. Making is not the imposition of form on matter but a correspondence between the maker's movements and the material's flows.

Ingold's key move, for this chapter, is to treat life as a matter of lines. Living beings are not bounded objects connected by relations, which is the picture of the network. They are lines of becoming, trails of growth and movement that wind through the world, entangling with other lines. A tree is not a thing in a place. It is a line of growth from seed to canopy, entangled with the lines of fungi, insects, birds, wind and rain. A person is not a container in a space. She is a line of life, a trail of movement and activity, entangled with the lines of others. The world is not a space filled with objects. It is a meshwork of entangled lines.

Meshwork and Network

Ingold takes the term meshwork from Henri Lefebvre, who used it for the patterns of movement and activity that weave through urban space, and contrasts it with the network of actor-network theory and network science.

A network consists of nodes connected by links. The nodes are prior. They exist as distinct points, and the links connect them. A network diagram shows dots and lines between dots. The relations are between things that are already constituted.

A meshwork consists of lines that tangle and knot. There are no nodes prior to the lines. What look like nodes are knots, places where lines entangle and bind. The lines are prior, and the knots are where they meet. A meshwork is like a fabric or a tangle of string, not a diagram of dots and links. The relations are not between things. They are the lines along which things become what they are.

Ingold's contrast bears on anthro-complexity's ontology directly. Anthro-complexity's Element 3 adopts Latour's actant, the node of actor-network theory. Ingold has criticised actor-network theory for exactly its network structure. Actants are nodes, connected by relations, and the theory begins with the actants and traces their connections. A meshwork ontology would begin with the lines: the movements, growths and activities along which actants become what they are. It would see actants as knots in the meshwork, not as prior nodes.

Anthro-complexity's Element 2 (neither wholes nor parts presumed) is closer to the meshwork than Element 3's actants are. If neither wholes nor parts are presumed, then neither are nodes, and the ontology should begin with what Element 2 calls interactions, which Ingold calls lines. Anthro-complexity's ontology is meshwork-like in Element 2 and network-like in Element 3. The tension is another instance of the pattern this disputation has traced. The ontological cluster's best principle, Element 2, is not followed through in the elements that build on it.

Stratigraphy

The Compass and the Loom set Ingold's meshwork against WRAS's substrate, the technological, cultural, procedural and network layers on which strategy is said to be built. The substrate is a stratigraphy. It pictures the organisation as layers, one beneath another, each supporting the one above, with strategy on top. Stratigraphy implies passive layers beneath strategy, whereas Ingold's meshwork and Juarrero's constraints describe relations that are co-constituted with action.

Ingold's critique of stratigraphic thinking runs through his work. In The Perception of the Environment he criticised the "layer-cake" model of the human, in which biology is the bottom layer, psychology the middle and culture the top, each built on the one below. He argued that the model gets human development wrong. A person is not a biological organism with a culture added on top. She is an organism-person whose biological development and cultural formation are a single process of growth within an environment. Skills are not cultural software running on biological hardware. They are embodied capacities, grown in the body through practice, and they are as biological as they are cultural.

The same critique applies to the organisational substrate. The organisation is not a technological layer with a cultural layer on top and a procedural layer above that. Its technologies, cultures and procedures are entangled lines, each shaping and shaped by the others, none beneath or above. Culture is not a layer to be manipulated. It is, in The Compass and the Loom's phrase, the recurring shape of local interaction, so that intervening in it is "more culture". A strategy built "on" a cultural substrate is in fact a line among the culture's lines, and its effects will depend on how it entangles with them, not on how it sits on top.

Wayfaring and Navigation

Ingold's Lines develops a second distinction, which bears directly on Thesis III. It is the distinction between wayfaring and transport, and between the wayfarer and the navigator.

The navigator moves from point to point across a surface that has been mapped in advance. He has a destination, plots a route to it on the map, and follows the route. His knowledge is the knowledge of the map: a representation of the territory from above, in which the territory's features are located as points and his route is a line between them. The navigator's movement is transport. It carries him from one location to another, and what happens between locations does not matter, because the journey is only the means of arrival. Ingold calls the navigator's line a connector, a line drawn between points that exist before it.

The wayfarer moves along a path that she makes as she goes. She has a direction, perhaps, but not a route plotted in advance. She knows the country by moving through it, attending to what she meets, adjusting to the ground, the weather and the paths of others. Her knowledge is not a map but a wayfinding skill, the capacity to find her way through a world she knows from inside. Her movement is not transport but wayfaring. The journey is not a means to arrival. It is the life she leads. Ingold calls her line a trail, a line that grows from its tip as she moves, not a line drawn between pre-existing points.

Ingold's examples include the Inuit hunter who knows the sea-ice by travelling over it, the Aboriginal Australian who knows the country through song-lines walked and sung, and the pedestrian who knows a city by walking it. In each case knowledge is gained along the way, not above it. The wayfarer is inside what she knows. Her knowing is part of her moving, and her moving is part of the country's life.

The sailor, the navigator and the wayfarer

The estuary's sailor of Thesis III is, in Ingold's terms, a navigator. He reads the estuary as a mapped surface of channels, sandbanks and currents, waits for the tide that will carry him along his route, and moves. His knowledge is chart-knowledge, and his position is above the medium. The estuarine map is the navigator's chart. It locates constraints as points on a surface, with their energies and timescales, so that a route through them can be plotted.

But Ingold's distinction also shows that the sailor need not be a navigator. Traditional seafarers, notably the Pacific navigators studied by Ben Finney, David Lewis and others, whose techniques The Compass and the Loom discusses as etak, are wayfarers of the sea. They did not use charts. They knew the ocean by travelling it: through the swells, the stars, the birds and the colour of the water, all read from inside the canoe and in motion. The navigator of the etak system conceives the canoe as stationary and the islands as moving past it. That is a frame of reference inside the journey, not above it. The Compass and the Loom draws the contrast. A magnetic compass delegates orientation to an external field, while the star compass and etak cultivate it through embodied practice. Orientation cannot be bought.

So the problem with the estuary is not that it is maritime. It is that estuarine mapping, as named and practised, draws on the navigator's relation to the sea: the chart, the route, the tide as an external force to be waited for. The wayfarer's relation would be different. The wayfarer knows the estuary from inside, in motion, as a meshwork of currents she is part of, and her knowledge is the skill of finding a way through it, not a chart of its features. A wayfaring anthro-complexity would be one in which practitioners and participants know their organisation as wayfarers know country, from inside and along the way, through attentive participation, and not as navigators know a chart.

Correspondence

Ingold's later work, especially Making and Correspondences, develops the concept of correspondence, which offers a model for what narrative instruments could be.

Correspondence, in Ingold's sense, is not matching (as in "this corresponds to that"), nor communication by letter, though Ingold plays on both senses. It is the process by which lines of life answer to one another over time. When a woodworker planes a board, her movements answer to the grain of the wood, and the wood's resistance answers to her movements. The making is a correspondence between maker and material, each responding to the other in real time, neither imposing on the other. When two people walk together, their steps fall into rhythm, each answering to the other's pace. When a musician plays with others, she listens and responds, and the music is their correspondence.

Correspondence has three features that matter here.

It is temporal. Correspondence happens over time, in an ongoing process of answering. It is not a single exchange but a continuing responsiveness.

It is mutual. Each party answers to the other, and neither is only the receiver. The woodworker answers the wood and the wood answers the woodworker.

It is attentive. Correspondence requires attention to the other, a waiting upon and listening to what the other is doing, so that one's own movement can answer it.

Correspondence is close to Mead's conversation of gestures, though Ingold does not use Mead's vocabulary. The meaning of a gesture is completed by the response, and the response is itself a gesture that calls out a further response. It is also close to Cook Ding's grain of 理 (Chapter 36): the cook's knife answers the grain of the ox, and the grain guides the knife. Correspondence is the relational structure of skilled human engagement with materials and with one another.

Narrative instruments as correspondence

Current narrative instruments are not correspondences. A SenseMaker study sends a prompt, receives narratives, aggregates them and, in good practice, returns patterns. That is a sequence of exchanges, not an ongoing mutual answering. The narrator answers the prompt, but the instrument does not answer the narrator, except, eventually and in aggregate, through the returned patterns. The relation is asymmetric (the instrument asks and the narrator answers), episodic (one round, then another) and, at the moment of narration, inattentive to the particular narrator, whose story is one of many.

An instrument designed as correspondence would differ on each count. It would be ongoing: a continuing exchange in which narrators' stories are answered, by other narrators, by the organisation's responses and by the instrument's own adjustments, and the answers are answered in turn. It would be mutual: the instrument and its users would answer the narrators, and the narrators would answer them, so that the designers' readings, signifiers and decisions are themselves open to the narrators' response (the Tidal Loom's reading of the reader). It would be attentive: designed so that particular stories can be heard as particular, and not only aggregated, so that the outlier, the I of Mead's social self, the weak signal, is answered and not merely counted.

The Tidal Loom is Ingold's correspondence turned into a design. The warp is laid by convenors and the weft woven by everyone. Each thread answers the others, and the cloth is made in the answering. There is no position outside the cloth.

Snowden's Reservation

Snowden's reservation about Ingold, stated in the November post through Ing's use of him, is that Ingold treats humans as one kind of animal and so downplays what is distinctive about humans. The reservation is fair in part. Ingold does refuse the sharp separation of humans from animals that Western thought has often drawn. He argues that the capacities taken to distinguish humans (language, culture, technology, symbolic thought) are grown in human organisms through development within an environment, as other animals' capacities are grown in them, and that there is no point at which "culture" is added to "nature". In The Perception of the Environment he criticises the idea that humans live in a world of symbolic representations while other animals live in a world of direct perception, and argues that humans, like other animals, perceive and act directly, and that symbols are tools within a life of direct engagement, not a separate realm.

But Ingold does not deny that humans differ. He denies that the difference is a separation, a line across which humans stand apart from the rest of life. Humans are distinctive organism-persons, with distinctive capacities grown in distinctive ways. They are distinctive within life, not outside it. That is close to anthro-complexity's own position, that human systems are complex systems that differ in kind from others within the complex domain (Element 6 and Chapter 13), and not a separation of the human from the natural.

The difference between Snowden and Ingold is therefore narrower than the November post suggests. Both hold that humans are distinctive within the natural world. Snowden emphasises the distinctiveness (multiple identities, intention, abstraction), and Ingold emphasises the continuity (organisms-persons growing in environments). This disputation proposes that each needs the other. Snowden needs Ingold's relational ontology, the meshwork, the wayfarer and correspondence, to escape the possessive grammar and the navigator's chart. Ingold needs Snowden's reflexivity, the asymmetry principle and the observation that human descriptions change the constraints they describe, because Ingold's meshwork, for all its richness, does not foreground how descriptions of the meshwork become lines in it. Ingold's wayfarer knows the country from inside, but she does not publish a map of it that others then follow, changing the country by following. Human systems have both wayfarers and cartographers, and the cartographers' maps become trails.

A Synthesis: Reflexive Meshwork

Combine the two and the result is a reflexive meshwork. Human systems are meshworks of entangled lines of life, as Ingold says, with persons as knots, not prior nodes, and knowledge gained along the way, not above it. They are also meshworks in which some lines are descriptions of the meshwork, maps, models, frameworks, patterns and stories, and those lines entangle with the others and change the meshwork's course, as Snowden says. The cartographer is a wayfarer too: her map is a trail she lays, and others follow it. The analyst is a knot in the meshwork, and her patterns are lines that tangle with the narrators' lines. The practitioner is not above the meshwork plotting routes. She is a line within it, and her interventions are movements that others answer.

Feature Network (actants) Meshwork (Ingold) Reflexive meshwork (synthesis)
Basic unit Nodes (actants) and links Lines of life; knots where they tangle Lines of life, some of which are descriptions of the meshwork
Persons Nodes, prior to relations Knots; lines of becoming Knots; lines that also describe and are described
Knowledge Of the network, from a vantage Along the way; wayfaring Along the way, including knowledge of how descriptions change the way
Representation Diagram of nodes and links Trail, not chart Trails and charts, where charts become trails others follow
Practitioner Analyst of the network Wayfarer Wayfarer whose maps are lines in the meshwork
Instruments Capture links and nodes (Ingold offers few) Correspondences that answer and are answered
Human distinctiveness Restored by qualification (Snowden) Within life, not apart (Ingold) Descriptive reflexivity: lines that describe the meshwork and change it

Ingold and the Theses

Thesis I (possessive fallacy). Ingold's person as a knot in a meshwork, a line of becoming and not a container, supports the relational account directly. Ingold's organism-person grows her capacities in relation, and does not hold them.

Thesis III (estuaries). Ingold's wayfarer, against the navigator, gives the sharpest available critique of the estuarine chart. It also shows that the maritime image need not be abandoned. It needs to shift from navigation to wayfaring, from the chart to etak.

Thesis IV (constraints). Ingold's critique of stratigraphy supports The Compass and the Loom's critique of the substrate, and extends it to any layered representation of constraints. Meshwork constraints are entangled lines, not layers.

Thesis VI (computation). Ingold's correspondence offers an alternative to the extractive arrangement of the harvest loop. Instruments that answer and are answered, rather than capture and aggregate.

The adversary will say that Ingold is a poor guide for practice. His writing is lyrical and philosophical, his examples are of craftspeople, walkers and hunters, and he offers nothing that could be deployed in an organisation of ten thousand people. The meshwork is a beautiful image and a useless method. Anthro-complexity's instruments, whatever their limitations, can be deployed. Ingold's correspondence cannot.

The adversary wins half. Ingold does not offer instruments, and his examples are small-scale. But the chapter does not propose Ingold as a source of instruments. It proposes him as a source of design principles for instruments: begin with lines, not nodes; know along the way, not from above; design for correspondence, not capture. These principles can be built into instruments at scale. The Tidal Loom is one attempt, and Chapter 45 proposes others. A principle that cannot itself be deployed can still shape what is deployed, as Ingold's critique of stratigraphy shaped The Compass and the Loom's critique of the substrate. partial

NOT NODES BUT KNOTS —

NOT THE NAVIGATOR'S CHART BUT THE WAYFARER'S TRAIL —

AND THE CARTOGRAPHER'S MAP BECOMES A PATH THAT OTHERS WALK.

Chapter 31

Hutchins: Cognition in the Wild

4,152 words, about 18 minutes of reading

The Ship Where No One Knew Where the Ship Was

Edwin Hutchins's Cognition in the Wild (1995) is a study of navigation aboard a United States Navy amphibious assault ship, called the USS Palau in the book. Hutchins, a cognitive anthropologist who had worked on Micronesian navigation before turning to naval ships, spent months aboard, observing and recording how the ship's navigation team fixed the ship's position as it entered and left harbour. The task is demanding. In confined waters, the ship's position must be determined every few minutes, accurately enough to avoid shoals, other ships and the harbour's edges, and the determination must be fast enough to inform the next manoeuvre.

Hutchins's central finding was that the cognitive work of fixing the ship's position was not done by any individual. It was done by a system: bearing-takers on the ship's wings, sighting landmarks through alidades; a bearing-recorder on the bridge, writing the bearings in a log; a plotter at the chart table, laying the bearings on the chart with a plotting tool, so that the lines intersected at the ship's position; and the navigator, who supervised and projected the ship's track forward. Each person did a part of the task, using tools that embodied centuries of accumulated navigational knowledge: the chart, the alidade, the plotting tool, the log, the procedures. No one person held the whole computation. The ship's position was computed by the arrangement of people, tools and procedures, and in a real sense the system knew where the ship was. No individual in it did.

This chapter takes Hutchins as the fifth thinker of Book Five because his work offers what anthro-complexity's human/AI hybrid most needs: an account of how humans and artefacts can be arranged so that cognition is distributed among them without being extracted from the humans for use elsewhere. Chapter 24 introduced the contrast between Hutchins's shared bridge and the harvest loop's extractive arrangement. This chapter develops it.

Distributed Cognition

Hutchins's theoretical claim was that cognition, at least much of the cognition that matters in human life, is distributed: across people, across artefacts, and across time. The claim was directed against the computational theory of mind (Chapter 15), which located cognition in individual heads as symbol processing. Hutchins argued that the computational theory had in effect taken the properties of socio-technical systems, such as the navigation team computing a position by manipulating representations on a chart, and attributed them to individual minds. The symbol-processing mind was a model of the navigation team, mistaken for a model of the navigator.

Distributed cognition has several features.

Representations propagate across media. A bearing begins as a visual alignment of a landmark in the alidade. It is spoken as a number over the phone circuit to the recorder, written in the log, read by the plotter, laid as a line on the chart, and intersected with other lines to give a position. At each step the representation changes medium (light, sound, ink, graphite, geometry), and the computation is performed by its propagation, not by any one processor.

Artefacts embody knowledge. The chart is not just a picture of the harbour. It embodies centuries of surveying, cartography and navigational practice, and its projection (Mercator) is designed so that bearings can be laid as straight lines. The plotting tool embodies the geometry of the computation. The procedures embody the accumulated lessons of navigational practice, including its errors. The navigators do not need to know all of this. They use the artefacts, and the artefacts do part of the thinking.

Cognition is cultural. The navigation system is the product of a long cultural history, which Hutchins traced from medieval European pilots through the development of the chart and the compass to modern naval practice. The system's cognitive capacities are cultural achievements, accumulated over generations and carried in artefacts and practices.

Individuals are parts, not wholes. Each navigator has individual cognitive capacities, perception, memory, skill, but these capacities are deployed within the system and shaped by it. A bearing-taker's skill is the skill of contributing to the system. Outside it, the skill has little use.

How a cockpit remembers its speeds

Hutchins's companion paper, "How a Cockpit Remembers Its Speeds" (1995), applied the same analysis to commercial aviation. Before landing, pilots must compute and remember several critical speeds (for flap extension, approach and so on) that depend on the aircraft's weight. Where are these speeds remembered? Not in the pilots' heads alone. They are written on a card, set on "speed bugs" (small markers around the airspeed indicator), spoken aloud in a checklist, and cross-checked between the pilots. The cockpit as a whole remembers the speeds, through an arrangement of people, artefacts and procedures that makes the memory robust against any single failure. Hutchins's title is meant literally. The memory is a property of the cockpit system, not of any pilot.

The Horizon of Observation

The feature of Hutchins's analysis that matters most here is what he called the horizon of observation: the portion of the task that each member of the team can see. On the Palau's bridge the horizons of observation overlapped. The plotter could hear the bearings as they were called. The recorder could see the plotter's work. The navigator could see the chart, the log and the plotter. Junior members, working at the edges of the task, could see what senior members were doing, and senior members could see what junior members did.

The overlapping horizons had three consequences.

Error detection. Because each person's work was visible to others, errors could be caught. A bearing wrongly called could be noticed by the plotter when it produced a position inconsistent with the others. A line wrongly laid could be noticed by the navigator. The system was robust because its members could see one another's work.

Learning. Because junior members could see senior members' work, they learned by observing it. Hutchins drew on Jean Lave and Etienne Wenger's Situated Learning (1991), which had developed the concept of legitimate peripheral participation. Newcomers learn a practice by participating in it at the periphery, in simple tasks with low stakes, while observing the more central tasks of experienced members. Over time they move towards the centre. The overlapping horizons of observation on the bridge were the condition for this learning. A navigation team whose members could not see one another's work could not train its newcomers.

Shared situation awareness. Because horizons overlapped, every member had some sense of the whole task, not just her part. When something went wrong, many members could see it and respond.

The Palau's crisis

Hutchins's most famous episode illustrates all three. As the Palau was entering San Diego harbour, it suffered a loss of main steam power, and with it the loss of the gyrocompass repeater that the navigation team used to read bearings in true degrees. The team's computational procedure, which depended on true bearings, was suddenly broken. The ship was moving in confined waters, and its position had to be fixed.

What happened next, which Hutchins reconstructed from recordings, was that the team reorganised its computation on the fly. Without any plan or instruction, over several minutes of trial, error and adjustment, the team's members redistributed the work. The plotter began to convert magnetic bearings to true bearings by arithmetic, a step the gyrocompass had formerly done. The arithmetic was then split between two members, and a new division of labour emerged that kept the fixes coming. No one designed the new procedure. It emerged from the team's interactions, under pressure, through members adjusting to what they could see of one another's work. It was self-organisation in the precise sense: a new pattern of coordination arising from local interaction, without central design.

The reorganisation was possible because of the overlapping horizons of observation. Members could see what others were doing and what was going wrong, and they could adjust their own work to fill the gaps. In a system where each member saw only her own part, the crisis would have been unmanageable. In a system with overlapping horizons, it was resolved by emergence.

Hutchins and Complexity

The Palau episode is a complexity case study of the first rank, and it is notable how closely it fits anthro-complexity's own concepts. The ship's navigation system was a complex system: distributed, emergent and adaptive. The loss of steam power was a shock that moved it from the ordered domain, procedure, to the complex domain, where no procedure applied. The team's response was probe, sense, respond: try a redistribution, see if it works, adjust. The new procedure emerged and was then stabilised as a pattern. Snowden could have used the Palau to illustrate Cynefin's dynamics.

But the Palau also illustrates what this disputation has argued anthro-complexity's instruments lack. The emergence was possible because the team's cognition was distributed among its members, with overlapping horizons of observation. It was not collected by a central instrument and read by someone outside the team. Every member could see enough of the whole to contribute to its reorganisation. The system's intelligence was in its arrangement, and the arrangement was shared.

The Horizon of Observation as Design Principle

Hutchins's horizon of observation gives the clearest available design principle for anthro-complexity's instruments: the horizons of observation should overlap. Each participant should be able to see enough of the whole, including what others see, to contribute to its sense-making, catch its errors and learn from it.

Apply the principle to the narrative instrument.

Feature Current narrative instrument Bridge-like narrative instrument
Narrator's horizon Her own story and signification Her story, others' stories, the emerging patterns
Analyst's horizon All stories and patterns All stories and patterns, and narrators' readings of the patterns
Leadership's horizon Patterns as reported Patterns as reported, and narrators' readings, and leadership's own decisions visible to all
Overlap Analyst and leadership see what narrators do not All see substantially the same; each sees the others' readings
Error detection Analyst's errors in reading patterns rarely caught by narrators Narrators can see and contest the analyst's readings
Learning Narrators learn little from participation Narrators learn from others' stories and patterns (legitimate peripheral participation)
Response to crisis Central reading, central response Distributed reading, emergent response

The bridge-like instrument is a mutual dashboard, which Thesis V called for. Everyone sees what everyone sees, and the readings of each are visible to the others. It is also the condition for the emergent, self-organising response that anthro-complexity values. The Palau's team reorganised itself because its members could see one another. A population that cannot see what its leadership sees, or what the analyst makes of its stories, cannot reorganise itself in response. It can only wait for the centre to respond.

Distributed Cognition and the Human/AI Hybrid

Hutchins's framework bears most directly on the human/AI hybrid that anthro-complexity's recent instruments move towards. The Compass and the Loom adopted Hutchins's framing: cognition is distributed across arrangements of people and tools, so the question is how to arrange what passes between them. This chapter can make the question more precise.

In Hutchins's bridge, artefacts do part of the cognitive work. The chart computes geometry and the plotting tool computes intersections. But the artefacts are in the team's horizon of observation. Everyone can see the chart. Its outputs are visible, checkable and correctable by the humans who use it. The chart does not take bearings from the bearing-takers and compute positions somewhere they cannot see, for purposes they do not know. It is a shared surface on which their work is coordinated.

A human/AI hybrid could be arranged like the bridge or like the harvest loop. In the bridge arrangement, the AI is a shared surface. Its pattern detections, anomaly flags and suggested responses are visible to everyone whose narratives contributed to them, open to their checking and contestation, and used within the team's own sense-making. In the harvest arrangement, the AI is a central processor. It takes human noticing as input, computes outputs that are seen by the centre and used for the centre's purposes, and learns from the noticing so that, in time, the noticing is no longer needed.

Hutchins's analysis says that the bridge arrangement is not only more ethical. It is more intelligent. Distributed cognition with overlapping horizons is robust to error, supports learning and enables emergent response to crisis. A system with a central processor and non-overlapping horizons is fragile. Its errors propagate uncaught, its participants do not learn, and its response to crisis depends on the centre. The harvest arrangement is the cockpit in which only one pilot can see the speed bugs.

Hutchins's warning about automation

Hutchins and colleagues in the cognitive engineering tradition (including Lisanne Bainbridge, whose 1983 paper "Ironies of Automation" is a classic) have documented a specific danger of automation in socio-technical systems. Bainbridge observed that automating the routine parts of a task leaves humans with the non-routine parts, the emergencies, the anomalies, the cases the automation cannot handle, while simultaneously depriving them of the routine practice through which they would have built the skill to handle those parts. The more effective the automation, the less prepared the humans are for its failures. Bainbridge called this the central irony of automation.

The irony applies to narrative instruments that move towards automation. If pattern detection is automated, humans no longer practise reading patterns in narratives. When the automation fails, as it will on the weak signals, the novel and the out-of-distribution, which are exactly what anthro-complexity's instruments exist to catch, the humans who should catch the failure will have lost the skill. The weak signal that the model misses will be missed by the humans too, because they have been trained by the automation to look where the model looks. That is Thesis VI's tail erasure, seen from the human side.

The grating of Chapter 24 is the remedy Hutchins's analysis supports. Keep a human checkpoint at the points where the automation's failures would be most consequential. Keep the human practice of reading narratives alive, so that the skill is there when it is needed. Keep the automation's outputs inside the horizon of observation, where they can be checked.

Hutchins and the Theses

Thesis I (possessive fallacy). Hutchins's individuals are parts of a distributed system, and their cognitive capacities are shaped by and deployed within it. Cognition is not possessed by individuals. It is distributed across arrangements. The possessive grammar, applied to cognition, mistakes the system's properties for the individual's.

Thesis V (power). Hutchins's horizon of observation is an analysis of the distribution of visibility, which Foucault analysed as power. Overlapping horizons distribute visibility, and non-overlapping horizons concentrate it. The mutual dashboard is the bridge's principle and the panopticon's opposite.

Thesis VI (computation). Hutchins's bridge is the alternative to the harvest loop: artefacts inside the horizon of observation, not processors outside it. Bainbridge's irony of automation predicts the loss of the human skill that weak-signal detection needs.

Thesis VIII (degradation). Distributed cognition with overlapping horizons self-corrects, because errors are visible and contested. A centralised instrument cannot self-correct in the same way, because its errors are visible only to the centre, which may not notice them.

What Hutchins Does Not Supply

Hutchins's framework has a limit that should be noted. The Palau's navigation team was a team with a shared purpose, fixing the ship's position, and a shared stake, the ship's safety. Its distributed cognition worked because everyone wanted the same thing. Organisations are not always like this. Their members have different purposes and stakes, their power ratios are unequal (Elias), and their established and outsiders tell different stories (Chapter 28). Overlapping horizons of observation in such organisations do not automatically produce shared sense-making. They may produce contested sense-making, in which different groups see the same patterns and read them differently according to their interests.

That is not an argument against overlapping horizons. A contested sense-making in which all parties can see what the others see is better than a centralised one in which only the centre sees. But it means that Hutchins's bridge cannot be the only model. Human organisations need distributed cognition and the political institutions within which contested readings can be negotiated, which brings the disputation back to CST's participants dimension (Chapter 12) and the quorum with inhibition of the bee swarm (Chapter 9). The bridge supplies the epistemics, and the politics must be supplied elsewhere.

The adversary will say that the analogy between a navy bridge and an organisation is strained. The bridge is a small team, physically co-located, with a well-defined task and strict hierarchy. Organisations are large, dispersed, with ill-defined tasks and contested authority. Overlapping horizons are feasible on a bridge of eight people. They are not feasible in an organisation of eighty thousand, where no one can see everything and the centre must aggregate. The harvest arrangement is not a moral failure. It is a necessity of scale.

The adversary wins half. Overlapping horizons cannot be complete at scale, and some aggregation is necessary. But Hutchins's principle is not that everyone must see everything. It is that horizons should overlap: each participant should see enough of what others see to contribute, catch errors and learn. At scale, this means designing the aggregation so that its outputs are visible to those whose inputs produced them, at the level of aggregation relevant to them. A department sees its own patterns and the organisation's, and leadership's readings of both. Mutual dashboards at scale are a design problem, not an impossibility. The adversary's necessity of scale justifies aggregation. It does not justify concentrating the horizon of observation at the centre. partial

THE SHIP KNEW WHERE THE SHIP WAS —

BECAUSE EVERYONE COULD SEE THE CHART —

DESIGN THE HORIZONS TO OVERLAP.

Addition: A Bridge for Eighty Thousand

The adversary objected that Hutchins's overlapping horizons are feasible on a bridge of eight people and not in an organisation of eighty thousand, where the centre must aggregate. The chapter replied that the principle is overlap, not completeness, and that mutual dashboards at scale are a design problem, not an impossibility. This addition sketches the design, to show that the reply is not merely rhetorical.

The problem of scale

On the Palau's bridge, every member could see the chart. In an organisation of eighty thousand, no one can see everything, and the centre must aggregate. The question is not whether to aggregate but how: whether aggregation concentrates the horizon of observation at the centre, as the panopticon does, or distributes it so that horizons overlap at every level, as the bridge does.

Nested bridges

The design principle is nested bridges. The organisation is not one bridge but many, nested at different scales, each with its own shared chart, and with the charts of each level visible to the levels above and below.

Level 1: the team. A team of eight to twenty people sees its own narratives, signified and patterned, on a shared chart. Every member sees what every other member sees. This is Hutchins's bridge at its natural scale. The team can read its own patterns, catch errors in the analysis, learn from one another's stories and reorganise itself in response, as the Palau's team did when the gyrocompass failed.

Level 2: the unit. A unit of several teams sees its teams' patterns side by side, and its own aggregate, labelled as an aggregate. Every team in the unit sees every other team's chart, within the limits of anonymity. Each team can see how its patterns compare with others', where it diverges and where it agrees. The unit's leadership sees the same charts as the teams, and its reading of them is shared with the teams.

Level 3: the division. A division of several units sees the units' patterns side by side, and its own aggregate. Every unit sees the others. The division's leadership reads, and its reading is visible to the units.

Level 4: the organisation. The organisation sees the divisions' patterns side by side, and its own aggregate. Every division sees the others. The executive reads, and its reading is visible to everyone.

At every level the horizon of observation is shared among those at that level, and the readings of each level are visible to the level below. No level sees what the level below cannot. The executive sees the organisation's aggregate, and so does every team, labelled as an aggregate from nowhere in particular (Chapter 39). The executive's reading of the aggregate is visible to every team, which can respond.

What flows upward and downward

Direction What flows Hutchins's analogue
Upward Patterns; aggregates; teams' readings of their own patterns Bearings called to the recorder
Downward Higher levels' readings; decisions; reasons The chart visible to all; the navigator's projection spoken aloud
Sideways Peer teams' and units' patterns and readings Bearing-takers on both wings hearing each other

In a panoptic arrangement, patterns flow upward and decisions flow downward, without the patterns' readings or the decisions' reasons. In a nested-bridge arrangement, readings and reasons flow in both directions, and patterns flow sideways too.

Error detection at scale

The Palau's team caught errors because each member's work was visible to others. In nested bridges, errors are caught at the level where they are visible. A team catches an error in the analysis of its own patterns, because it knows its own stories. A unit catches an error in the aggregation of its teams, because it can see each team's chart. The executive's misreading of the organisation's aggregate can be caught by any team, because every team sees the executive's reading and can say, that is not what our stories mean.

In a panoptic arrangement, errors at the centre are caught only by the centre. An analyst's misreading of the patterns, or an executive's misinterpretation, propagates unchecked, because no one below can see the reading to contest it.

Learning at scale

Lave and Wenger's legitimate peripheral participation requires that newcomers can see the work of experienced members. In nested bridges, every team can see every other team's patterns and readings. A new team, or a team in difficulty, can learn from teams that have faced similar patterns and found ways to respond. The organisation's learning is distributed. It does not depend on the centre's synthesis of best practice, which is often too abstract to use.

Emergent reorganisation at scale

The Palau's crisis showed a team reorganising its computation without design, because its members could see what was going wrong. In an organisation of eighty thousand, a crisis such as a supply failure, a regulatory shock or a reputational collapse requires reorganisation at many levels at once. In a panoptic arrangement, the reorganisation must come from the centre, which sees the whole and must design the response. In nested bridges, reorganisation can emerge at every level. Each team, seeing its own and its neighbours' charts, can adjust. Each unit, seeing its teams', can coordinate. The centre, seeing the divisions', can attend to what only the centre can do. The centre is not abolished. It is relieved of the impossible task of designing every response.

The cost

Nested bridges cost more than a single central dashboard. Every level needs its chart, its readings must be collected and shared, and anonymity must be protected at small scales, where a team of eight may be identifiable from its patterns. But the cost is mainly in time and attention, not technology. The technology for nested dashboards with role-based views is standard. And the cost should be set against the cost of the panoptic arrangement: errors uncaught, learning lost and crises managed from a centre that cannot see what the teams see.

Nested bridges and the principles

Chapter 44 principle How nested bridges meet it
4. The observer is inside Every level reads its own patterns; no level stands outside
5. The interpretant belongs to the community Readings at every level are recorded and shared
8. Make the horizons overlap The design's defining feature
11. Power as dimension No level sees what the level below cannot; asymmetry of visibility reduced
12. Recognition Teams respond to their own patterns; when change comes, they can say we did it ourselves

NOT ONE BRIDGE BUT MANY, NESTED —

EVERY LEVEL SEES ITS CHART, AND THE CHARTS BESIDE IT —

THE CENTRE'S READING IS VISIBLE TO EVERY TEAM.

Addition: Cross-Lights From Book Nine

Who sees which information. To show what working on conditions means, Snowden's post on Luhmann lists a meeting rhythm, a reporting line, or who sees which information. The third item is the horizon of observation. Changing it is the main proposal of this chapter.

The large group as a bridge. Flinn's large groups, described in Chapter 57, are rooms in which everyone sees and hears what forms among them. Their limit is number. A design in which nested groups of that kind each read their own patterns from an instrument would join Hutchins's principle to Snowden's reach.

Chapter 32

Peirce: Thirdness, the Interpretant, the Community of Inquirers

3,563 words, about 15 minutes of reading

The Philosopher Snowden Claims

Charles Sanders Peirce (1839–1914) is the philosopher anthro-complexity claims as its own. The November post distinguishes Snowden's Peircean pragmatism, a pragmatism of signs, abduction and habit grounded in the natural sciences, from Ing's Jamesian and Singerian lineages. It grounds the trialectic in Peirce's thirdness. It says that for anthro-complexity, Peirce is still live.

This chapter takes Snowden at his word and reads Peirce for what he offers anthro-complexity's human cluster. It finds that Peirce, read in full, supports this disputation's argument at every point. His semiotics places meaning in the interpretant, which is the effect of a sign on an interpreter, and so in relation. His theory of inquiry places truth in the community of inquirers, and abduction is disciplined by that community. And his account of the self is explicitly anti-possessive. A person is not absolutely an individual, Peirce wrote, and the self is a sign. If anthro-complexity is Peircean, and Peirce is still live, then the relational account of the human is not a foreign import. It is anthro-complexity's own inheritance, waiting to be claimed.

The Categories

Peirce's philosophy rests on three categories, the most general modes in which anything can be or be thought. He called them firstness, secondness and thirdness, and described them in many ways over his career.

Firstness is the mode of being of that which is such as it is, positively and without reference to anything else. It is quality, feeling and possibility, the redness of red considered by itself, before it is compared with anything or attributed to anything. Firstness is freshness, spontaneity and originality.

Secondness is the mode of being of that which is such as it is with respect to a second, but regardless of any third. It is reaction, resistance and brute fact: the shock of an unexpected event, the resistance of a door one pushes against, the existence of this particular thing here and now. Secondness is struggle, actuality and the dyadic clash of one thing with another.

Thirdness is the mode of being of that which is such as it is in bringing a second and a third into relation with each other. It is mediation, law, habit, generality and representation. A law of nature is thirdness. It mediates between particular events, making them instances of a general pattern. A habit is thirdness, mediating between situations and responses. A sign is thirdness, mediating between an object and an interpretant.

Peirce held that the three categories are irreducible. Thirdness cannot be built from secondness, nor secondness from firstness. And he held that thirdness is the category of mind in the broadest sense: of meaning, purpose, intelligence and law. Where there is thirdness, there is something like thought, even in nature.

The Sign

Peirce's semiotics is the theory of thirdness in its most developed form. A sign (or representamen) is something that stands to somebody for something in some respect or capacity. The sign relation has three terms:

  • the sign itself, the sign-vehicle: a word, a mark, a gesture, a symptom;
  • the object, that which the sign stands for;
  • the interpretant, the effect of the sign on an interpreter, which is itself a sign of the same object, and which calls out a further interpretant in its turn.

The interpretant is the crucial term. It is not the interpreter, the person who interprets, but the interpretation: the sign's effect, the meaning it comes to have. And it is itself a sign, so semiosis, the action of signs, is in principle endless. Each interpretant calls out a further interpretant, and meaning unfolds in a chain of interpretations that never reaches a final, fixed signification, except, Peirce held, as an ideal limit.

Peirce distinguished several kinds of interpretant. The immediate interpretant is the meaning the sign is fitted to convey, its interpretability. The dynamical interpretant is the actual effect the sign has on a particular interpreter on a particular occasion. The final interpretant is the effect the sign would have on an interpreter who had considered the matter fully, the interpretation that inquiry would converge on in the long run.

He also distinguished three kinds of relation between sign and object. An icon signifies by resemblance: a portrait, a diagram, a map. An index signifies by causal or existential connection: smoke as a sign of fire, a weathervane as a sign of wind, a symptom as a sign of disease. A symbol signifies by convention or habit: a word, a flag, a legal title. Symbols are thirdness in the fullest sense, because their relation to their objects is mediated by a habit of interpretation, a general rule.

The narrative instrument, semiotically

Apply Peirce's semiotics to the narrative instrument with more precision than Chapter 25 allowed.

A micro-narrative is a symbol, a story in words, whose object is the narrator's experience and whose meaning depends on conventions of language and storytelling. It is also partly an index, because it was caused by the experience it reports, and bears traces of it.

The narrator's signification, her dot on a triad, is a sign of the micro-narrative's meaning to her. It is an icon: its position resembles, in a diagrammatic way, the relative weight of the three vertex qualities. Its object is the narrative's meaning, and its dynamical interpretant is the narrator's own understanding of her story, as expressed in the placement.

The pattern, the cluster of dots seen by the analyst, is a sign of the population's significations. It is an icon too, a diagram. Its object is the distribution of meanings across the population, and its dynamical interpretant is the analyst's reading.

The decision leadership takes on the basis of the analyst's report is an interpretant of the report, and through it of the pattern, the significations, the narratives and the experiences.

Each step is a sign, and each sign calls out an interpretant that is itself a sign for the next step. Semiosis runs from experience to narrative to signification to pattern to report to decision, and at each step the interpreter changes: narrator, analyst, leadership. Peirce's question, which interpretant governs?, becomes: at which step, and by whom, is the meaning fixed that leads to action?

In typical deployment the answer is that the meaning that governs action is fixed at the end of the chain, by leadership's interpretation of the analyst's report. The narrator's dynamical interpretant, her own understanding of her story, is the first link. It is captured and then interpreted by others. Peirce's semiotics shows that this is not a neutral transmission of meaning. It is a chain of reinterpretations, each by a new interpreter with her own habits, and the narrator's meaning is transformed at every step.

The Community of Inquirers

Peirce's theory of inquiry, set out in his papers of 1877–78, "The Fixation of Belief" and "How to Make Our Ideas Clear", is social through and through.

"The Fixation of Belief" considered four methods by which people settle their beliefs and escape the irritation of doubt. The method of tenacity is clinging to what one already believes and avoiding anything that might disturb it. The method of authority is believing what an institution, a church or a state prescribes, and suppressing dissent. The in advance method is believing what is "agreeable to reason", what seems plausible on reflection. The method of science* is believing what is determined by an external reality that is independent of anyone's opinion, discovered through experience and reasoning that any inquirer could, in principle, repeat. Peirce argued that only the method of science could produce stable belief in the long run, because only it submits belief to a check that does not depend on the believer.

The check is social. No individual can test her beliefs against reality alone, because individual experience is limited and individual reasoning fallible. Truth is what the community of inquirers would agree on, if inquiry were pursued long enough. Peirce's famous definition, in "How to Make Our Ideas Clear", is that the opinion fated to be ultimately agreed to by all who investigate is what we mean by truth, and the object represented in this opinion is the real. Reality is what the community of inquiry would converge on. The individual inquirer is disciplined by the community: by the possibility that others will check, test and contest her claims.

Peirce's fallibilism followed. No individual belief is certain, because any belief may be overturned by further inquiry. The community's current opinion is not final either. It is the best current approximation to the opinion inquiry would converge on. Inquiry is ongoing, self-correcting and social.

Abduction is social

Peirce's abduction (Chapter 5), inference to the best explanation and the only form of inference that introduces new ideas, is often presented as an individual creative act: the detective's flash of insight, the scientist's bold hypothesis. Peirce held that abduction is the beginning of inquiry, and that inquiry is social. An abduction is a hypothesis offered to the community, which tests it by deduction (deriving consequences) and induction (checking consequences against experience). The abduction's value is not in the individual's insight but in its fruitfulness for the community's inquiry.

The Compass and the Loom draws the consequence for the human/AI question: abduction is social and disciplined by a community of inquirers, so the human/machine split is not clean. An abduction made by a machine and offered to a community of human inquirers is part of the community's inquiry. One made by a machine and acted on by a centre without community testing is not inquiry at all. It is authority.

The four methods and the instruments

Peirce's four methods of fixing belief offer a test for anthro-complexity's instruments. Which method does a given deployment embody?

Method How belief is fixed Instrument deployment that embodies it
Tenacity Clinging to existing belief Reading patterns as confirmation of what leadership already thought
Authority Prescribed by an institution Patterns reported by an authorised analyst and accepted because of the analyst's authority
A priori What seems reasonable Patterns interpreted by what "makes sense", without testing against narrators' own readings
Science Tested against reality by a community Patterns returned to narrators, who test the analyst's reading against their experience; readings contested and revised

Only the last is Peircean. It requires that the community whose experience the patterns describe be part of the community of inquiry that interprets them. The narrators must be inquirers, not just sources. A Peircean anthro-complexity would design its instruments so that narrators test the analyst's readings, contest them where they do not fit, and contribute to the final interpretant. Return before action, the Tidal Loom's principle, is Peirce's method of science applied to narrative.

Peirce on the Self

The most striking support Peirce offers this disputation is his account of the self. It is explicitly anti-possessive, and Snowden, in claiming Peirce, inherits it.

In his early paper "Some Consequences of Four Incapacities" (1868), Peirce argued that we have no power of introspection independent of inference from external facts, no power of intuition independent of previous cognitions, no power of thinking without signs, and no conception of the absolutely incognisable. From these he concluded that the self is not a substance known directly by introspection. It is inferred from signs, and is itself a kind of sign. The man is the thought, Peirce wrote, and the word or sign which man uses is the man himself. A person is a sign, an ongoing process of interpretation, whose identity consists in the consistency of what he thinks and does.

In later work Peirce developed this into an account of the self as dialogical. Thinking is a dialogue with oneself, between the self of the present moment and the self "just coming into life in the flow of time". In "What Pragmatism Is" (1905) he wrote that a person is not absolutely an individual. His thoughts are what he is "saying to himself", that is, saying to that other self that is just coming into life in the flow of time. The self is not a container of thoughts. It is a conversation, carried on through time, between present and future selves, and the conversation is in signs, whose meanings are social.

Peirce went further. In several places he suggested that the boundaries of the self are not fixed by the skin, that a person's identity extends to those with whom she shares thought and feeling, and that the community is in some sense a person. His "Evolutionary Love" (1893) developed a cosmology in which agape, the love that draws beings into communion, is a principle of evolution alongside chance and law. Growth comes, in part, from sympathy and fellow-feeling, from beings giving themselves to one another. These are speculative writings, and they should not be pressed too hard. But they show that Peirce's thought about the self was relational, communal and anti-possessive from beginning to end.

Peirce against the possessive grammar

Set Peirce's account of the self beside the November post's human cluster:

November post Peirce
Humans hold multiple identities A person is a sign; identity is the consistency of an ongoing interpretation
Humans act with intentionality Thought is dialogue with the self coming into being; purpose is a habit of interpretation
Humans create abstractions Abstractions are symbols, whose meaning lies in habits of interpretation shared by a community
The people remain distinct A person is not absolutely an individual

The contrast is stark. Snowden's own philosopher rejects the possessive grammar in which Snowden states the human distinction. If Peirce is still live for anthro-complexity, as the November post says, then the human cluster must be restated in Peirce's terms. The human is a sign, in dialogue with itself and others, whose identity, intentions and abstractions are habits of interpretation formed and sustained in community. That is the relational account of Thesis I, stated in the vocabulary of anthro-complexity's chosen philosopher.

Habit, Synechism, Tychism

Three further Peircean doctrines bear on anthro-complexity's ontology, and each supports it while adding a relational dimension.

Habit. For Peirce, a habit is a general disposition to act in a certain way under certain circumstances. Beliefs are habits of action, and the meaning of a concept is the habits it would produce. In Peirce's cosmology, laws of nature are habits the universe has taken: regularities that emerged from an original chaos of chance and became more fixed over time. Habit is thirdness in its active form, mediation becoming regular. Anthro-complexity's interest in patterns, attractors and stabilised constraints is an interest in habits. Peirce adds that habits are taken, formed through repetition and reinforcement, and can be broken by surprise. Human habits are formed socially, in shared practices, and broken by the surprises that inquiry and experience bring.

Synechism. Peirce's doctrine of continuity holds that reality is continuous, that sharp boundaries are abstractions, and that apparent discontinuities resolve, on closer inquiry, into gradations. Synechism bears on the human distinction. It predicts that the boundary between human and non-human complexity will be a gradation, not a sharp line, which is what Chapter 9's comparative evidence found. The cleaner wrasse's audience effects, the ape's gaze-following and the human's shared intentionality lie on a continuum. Synechism does not rule out differences of kind, since Peirce recognised that continua have regions with distinctive properties. But it counsels against the sharp exceptionalism the institutional job pulls towards (Chapter 18). The difference of kind lies in descriptive reflexivity, a region of the continuum where signs about signs become constraints, and synechism expects it to have a border that shades.

Tychism. Peirce's doctrine of absolute chance holds that the universe contains real spontaneity, events not determined by prior conditions. Tychism supports anthro-complexity's insistence on the unpredictability of complex systems. It also supports Mead's I, the spontaneous, novel element in conduct. And it underlies Peirce's cosmology of evolution, in which chance produces variety, habit stabilises it, and agape draws it into communion. A Peircean anthro-complexity would attend to tychism in human systems: to the genuinely novel, the I that breaks the me's categories, the weak signal that is not yet a pattern.

Where the Interpretant Should Sit

Chapter 25 asked where the interpretant sits in the narrative instrument, and found that it sits, in typical deployment, with the analyst and designer, not the narrator. With Peirce's full semiotics and theory of inquiry in hand, the question can be answered more precisely, and a design proposed.

The dynamical interpretant of a micro-narrative, its actual effect on a particular interpreter, exists for every interpreter who reads it: the narrator in signifying, the analyst in reading the pattern, leadership in deciding. None is privileged by Peirce's semiotics. Each is a real effect of the sign on a real interpreter.

The final interpretant, the interpretation inquiry would converge on, is not any individual's. It is the community's, in the long run. Peirce's theory of inquiry says the final interpretant is approached by a community of inquirers testing interpretations against experience and against one another. If the narrators are not part of the community that tests interpretations, if their dynamical interpretants are captured but their readings of the analyst's interpretants are not sought, then the community of inquiry is truncated. Its final interpretant will be the analyst's and leadership's, not the community's.

A Peircean design would therefore place the narrators inside the community of inquiry that approaches the final interpretant:

  • Narrators' dynamical interpretants are captured (as now, through self-signification).
  • The analyst's dynamical interpretant is returned to narrators (patterns and readings shown to them).
  • Narrators' interpretants of the analyst's interpretant are sought (what do you make of this pattern? does it fit your experience? what is missing?).
  • Contested interpretations are recorded as such (not resolved by the analyst's authority).
  • Leadership's interpretant, the decision, is returned and open to response (this is what we concluded and did; how do you read it?).
  • The chain continues (semiosis is endless; the next round begins from where this one left the interpretations).

This is Peirce's method of science applied to the organisation's self-understanding. It is also Ingold's correspondence (Chapter 30), Hutchins's overlapping horizons (Chapter 31), and the Tidal Loom's return as design requirement and reading of the reader. The convergence is not coincidental. Each of these thinkers, from a different direction, has seen that meaning in human systems is made in a community of interpretation, and that a practice which captures meaning at one point and interprets it at another has broken the community.

Peirce and the Theses

Thesis Peirce's contribution
I: possessive fallacy A person is a sign, not absolutely an individual; the self is dialogue with the self to come
II: asymmetry applied to itself Every description is a sign whose interpretants change the system; the theorist's signs are among them
V: power The method of authority, against the method of science; who belongs to the community of inquiry
VI: computation Abduction is social; machine abductions acted on without community testing are authority, not inquiry
VII: triad as simplex The interpretant, Peirce's third, should sit with the community, not with the analyst alone
VIII: degradation Fallibilism; inquiry self-corrects only if its community can contest; habits formed by instruments can be broken only by surprise

Peirce Is Still Live

The November post says Peirce is still live for anthro-complexity, as the earlier posts on semiotics and the abductive warrant set out. This chapter agrees, and draws the consequence. Peirce's semiotics, theory of inquiry and account of the self together support a relational account of the human. Meaning is in the interpretant, truth in the community of inquiry, and the self a sign in dialogue. If anthro-complexity takes Peirce as live, it must take him whole: not only the triad and abduction, but the interpretant that places meaning in relation, the community that places truth in a shared inquiry, and the self that is not absolutely an individual.

Taken whole, Peirce is the strongest available ally for this disputation's argument, and he is anthro-complexity's own. Followed through, the Peircean pragmatism Snowden distinguishes from Ing's leads to the same place as Mead, Elias, Juarrero, Ingold and Hutchins: a human who is a pattern of relation, a meaning made in community, and an inquiry that includes those it describes.

The adversary will say that Peirce's account of the self is a minor and speculative part of his work, and that anthro-complexity's use of Peirce, his semiotics, abduction and thirdness, is legitimate without it. One may take a philosopher's logic without his metaphysics of personhood, as many take Aristotle's logic without his biology.

The adversary is right that one may take parts of a philosopher selectively, and that Peirce's remarks on the self are scattered. But Peirce's account of the self is not separable from his semiotics. It is his semiotics applied to the person: the self is a sign, and its meaning is its interpretants. And it is not separable from his theory of inquiry, because the community of inquirers is a community of selves-as-signs. To take Peirce's semiotics while keeping the possessive self is to keep the sign relation for everything except the beings who make signs. That is an inconsistency, not a selection. The November post's claim that Peirce is still live is a claim about the whole of a system that is unusually unified. partial

THE MAN IS THE SIGN —

A PERSON IS NOT ABSOLUTELY AN INDIVIDUAL —

THE INTERPRETANT BELONGS TO THE COMMUNITY.

Chapter 33

Bourdieu, Giddens, Hacking: Structure Carried in the Body

3,746 words, about 16 minutes of reading

Three Sociologists of Structure

This chapter brings together three social theorists whose work has appeared throughout the disputation: Pierre Bourdieu, Anthony Giddens and Ian Hacking (a philosopher, but one whose work is sociological in its effects). Each offers an account of how social structure is carried, reproduced and changed by the people who live within it, and each supplies a concept that bears directly on anthro-complexity's human cluster and instruments. From Bourdieu come habitus, hysteresis and the scholastic fallacy. From Giddens come the duality of structure, practical consciousness and institutional reflexivity. From Hacking come making up people and dynamic nominalism. Together they make up the most developed account in the social sciences of how constraint is carried in the body and reproduced in action, which is exactly what Thesis IV found missing from anthro-complexity's theory of constraint.

Bourdieu

Habitus, field and capital

Pierre Bourdieu (1930–2002) developed, over four decades of empirical and theoretical work, a sociology built on three concepts.

Habitus is a system of durable, transposable dispositions: ways of perceiving, thinking, feeling and acting that are acquired through socialisation in a particular position in social space, and that generate practices fitted to that position. Habitus is embodied. It is carried in posture, gesture, taste, accent and the feel for what is appropriate. It is durable, persisting across situations and over time. It is transposable, generating practices in new situations that resemble those of the original. And it is generative. It does not determine specific acts, as rules do, but generates an indefinite variety of acts that share a style. Bourdieu called it structured structures predisposed to function as structuring structures: shaped by the social conditions of its acquisition, and shaping the practices through which those conditions are reproduced.

A field is a structured social space with its own stakes, rules and forms of competition: the academic field, the artistic field, the economic field, the field of power. Agents occupy positions in the field according to the volume and composition of their capital, and they compete for the field's specific stakes.

Capital comes in several forms: economic (money and property), cultural (education, knowledge, taste and credentials), social (networks and connections) and symbolic, the prestige and recognition that accrue when the other forms are perceived as legitimate. Each field values particular forms of capital, and agents' strategies are shaped by the capital they have and the field they are in.

Doxa

Doxa is the domain of what goes without saying in a field: the assumptions so deeply shared that they are not experienced as assumptions at all, but as the natural order of things. Doxa is the field's constitutive constraint, in the sense of Chapter 22, carried in the habitus of its members. What is doxic is not argued about. It is the ground on which arguments take place. Bourdieu distinguished doxa from orthodoxy and heterodoxy, the explicit positions taken when doxa is challenged. When something that went without saying is questioned, it becomes an orthodoxy, defended against a heterodoxy that challenges it. The universe of the undiscussed shrinks, and the universe of discourse grows.

The Compass and the Loom summarised the point: habitus and doxa are the mechanisms by which a club's rules are carried in members, not on paper. The constraints an estuarine map records are, for the most part, those that have become discussable, orthodoxies and heterodoxies. The doxic constraints, those that go without saying, are not recorded, because no one thinks to name them. They are reproduced in the mapping session as in every other practice of the field.

Hysteresis

Bourdieu's concept of hysteresis is directly relevant to anthro-complexity's theory of timing. Hysteresis, in physics, is the lag between a change in a force and the response of a system: a magnetised material retains its magnetisation after the magnetising field is removed. Bourdieu used the term for the lag between a change in a field's conditions and the adjustment of the habitus formed under the old conditions. Habitus is durable. It persists after the conditions that formed it have changed, and it generates practices fitted to the old conditions in the new ones. Bourdieu's early fieldwork in Algeria documented the hysteresis of peasant habitus when colonial capitalism transformed the economic field. Peasants whose dispositions had been formed in a pre-capitalist economy of honour and gift-exchange were thrown into a market economy, and their practices, fitted to the old field, misfired in the new one. Bourdieu also called this the Don Quixote effect, after the knight whose chivalric habitus survived the world of chivalry.

Hysteresis bears on kairos. The opportune moment for shifting a constraint is not only a function of the constraint's energy cost and the state of the "tide". It is also a function of the habitus of the people whose practices reproduce the constraint. A constraint whose formal conditions have changed may persist in habitus, so that the moment that looks opportune on the map is not opportune in the bodies of those who must change. Conversely, a habitus may have adjusted ahead of the formal conditions, so that the opportune moment has already passed by the time the map shows it. Estuarine mapping's energy and time axes measure the constraint. They do not measure the hysteresis of the habitus that carries it.

The scholastic fallacy

The concept of Bourdieu's that matters most for this disputation is the scholastic fallacy, developed in Pascalian Meditations (1997) and The Logic of Practice (1980).

Bourdieu argued that the scholar, the theorist or the analyst occupies a peculiar position, that of skholè, leisure from the urgencies of practice. The scholar observes practice without having to act in it, reflects on it at leisure, and represents it in models, diagrams and theories. This position is a privilege. It is made possible by freedom from economic necessity, by institutional support and by distance from the stakes. And it produces a characteristic illusion: the scholar projects her own relation to practice, contemplative, representational and detached, onto the practitioners she studies. She models their practice as if it were the execution of a plan or the following of a rule or the solution of a problem, because that is how practice looks from the position of skholè. But practitioners do not experience their practice that way. They act in the urgency of the moment, guided by a feel for the game, a practical sense, not by representations. The scholar's model is a model of practice as seen from the scholar's position, mistaken for practice itself.

The Compass and the Loom touched on the same idea in Book Two: going nowhere is a privilege of leisure from necessity (Bourdieu's skholè). The point bears directly on anthro-complexity's estuary (Thesis III). The sailor on the deck, reading the tide and waiting for it to turn, occupies the position of skholè. He has leisure to observe, the tide does not demand immediate action, and he can wait. The fisherman in the maelstrom (Chapter 28) does not. He is in the urgency. A practice whose governing image is the sailor's leisure projects that leisure onto the practitioners it serves, and its maps and models represent practice as it looks from the deck. The people in the organisation, working in the urgency of the moment, guided by their feel for the game, are represented as if they too could wait for the tide. The scholastic fallacy is the estuary's error, named by a sociologist.

Participant objectivation

Bourdieu's remedy for the scholastic fallacy was reflexive sociology, set out with Loïc Wacquant in An Invitation to Reflexive Sociology (1992) and later called participant objectivation. The sociologist must objectify her own position: the field in which she works, the capital she holds, the habitus she brings, and above all the scholastic position from which she observes. She must, in Bourdieu's phrase, objectify the objectifying subject. That is not a confessional exercise of narrating one's biases. It is a scientific exercise of analysing the social conditions of one's own knowledge with the same tools one uses to analyse others. Only by doing so can the sociologist correct for the distortions her position introduces.

Participant objectivation is the sociological form of the reflexivity this disputation has asked of anthro-complexity's instruments: the reading of the reader, the declination before direction, the inclusion of the teller in the tale (Thesis II). Bourdieu's version is more demanding than the Tidal Loom's. It asks not only that convenors signify their own experience, but that they analyse their own position in the field: their capital, their stakes, their relation to the clients who pay them and to the populations they sense. A reflexive anthro-complexity would apply participant objectivation to itself, analysing the field of complexity consultancy, its forms of capital and its doxa, with the same tools it applies to its clients.

Giddens

Structuration

Anthony Giddens's The Constitution of Society (1984) set out structuration theory, his attempt to overcome the dualism of structure and agency that had divided social theory. Structural theories (functionalism, structuralism, some Marxisms) explained action by structure, and saw individuals as bearers of roles and positions. Agency theories (interpretive sociology, phenomenology, rational choice) explained structure by action, and saw society as the outcome of individuals' meaningful or rational choices. Giddens argued that both were one-sided, and that structure and agency are two aspects of a single process.

The key concept is the duality of structure. Structure, the rules and resources that agents draw on in action, is both the medium and the outcome of the practices it organises. When agents act, they draw on structure, and in doing so reproduce it. Structure has no existence apart from the practices in which it is instantiated, and practices have no form apart from the structure they draw on. Language is the paradigm. Speakers draw on the rules of a language in speaking, and their speaking reproduces the rules. A language that no one spoke would not exist, and speech without a language would not be speech.

Chapter 22 used the duality of structure to explain why constitutive constraints are reproduced in compliance and in mapping. Structuration adds two further concepts that bear on anthro-complexity.

Practical and discursive consciousness

Giddens distinguished practical consciousness, what agents know about how to go on in social life without being able to put it into words, from discursive consciousness, what agents can say about their actions and the reasons for them. Most of what agents know is practical. They know how to hold a conversation, conduct a meeting or navigate an organisation's politics, but if asked, they could articulate only a fraction of it. Practical consciousness is close to Polanyi's tacit knowledge and Bourdieu's practical sense, though Giddens emphasised that its contents can, in principle, become discursive through reflection or questioning.

The distinction bears on narrative instruments. A micro-narrative draws on discursive consciousness. It is what the narrator can say about her experience. Practical consciousness, which is most of what she knows, does not appear in it, or appears only in traces: in what the story takes for granted, in the details that seem to the narrator too obvious to mention, in the gaps. An instrument that collects what narrators can say collects the discursive surface of their knowledge. The practical depth is not collected, and practical consciousness is where the constitutive constraints (Thesis IV) and the habitus (Bourdieu) are carried.

Reflexive monitoring and knowledgeability

Giddens insisted that agents are knowledgeable. They know a great deal about the conditions of their action and they continuously monitor what they do and what others do. Reflexive monitoring of action is a constant feature of social life. Agents do not just act. They watch themselves and others act, and adjust. Social life is therefore reflexive at its root. The double hermeneutic (Chapter 14) is an extension of this: social science's concepts enter agents' reflexive monitoring and change their conduct.

The insistence on knowledgeability is a corrective to any practice that treats agents as sources of data to be interpreted by experts. Agents already interpret their own situations, often with great sophistication. The expert's interpretation is one more interpretation, entering a field already thick with them. A narrative instrument that treats narrators' self-signification as raw material for expert pattern-reading underestimates their knowledgeability. Narrators already know a great deal about the patterns the instrument will detect, often more than the analyst. What they know is not captured unless they are asked to read the patterns themselves.

Institutional reflexivity and expert systems

In his later work on modernity, The Consequences of Modernity (1990) and Modernity and Self-Identity (1991), Giddens analysed institutional reflexivity: the routine incorporation of knowledge about social life into the organisation and transformation of social life. Modern institutions constantly monitor themselves, through statistics, audits, surveys, reviews and consultancy, and change themselves in light of what they find. Giddens argued that this reflexivity is a defining feature of modernity, and that it produces a characteristic instability. Knowledge about institutions changes institutions, so the knowledge is always out of date, and the institutions are always being remade.

He also analysed expert systems, the systems of technical accomplishment or professional expertise that organise large areas of modern life, and on which people depend without understanding them. Trust in expert systems is a condition of modern life, and disembedding, the lifting of social relations out of local contexts into abstract systems, is its characteristic process.

Anthro-complexity's instruments are expert systems in Giddens's sense. They are systems of professional expertise on which organisations depend, often without understanding their workings. They are instances of institutional reflexivity: they produce knowledge about organisations that is incorporated into how organisations are run, and they change organisations in doing so. Giddens's analysis predicts two things. First, the knowledge they produce will be continuously outdated by its own incorporation, which is Thesis VIII's degradation. Second, they will disembed the local, contextual knowledge of narrators, lifting it out of its context into the abstract system of patterns and dashboards, which is Thesis VI's legibility. Both are not failures of the instruments but features of their place in a reflexive modernity.

Hacking

Making up people

Ian Hacking's work, introduced in Chapter 14, offers the most precise account of how classifications of people change people. In "Making Up People" (1986), Hacking argued that new ways of classifying people, new kinds of person, create new ways to be a person. Before the late nineteenth century there were people who had sexual relations with people of the same sex, but there were not "homosexuals" as a kind of person, with an identity, a psychology and a way of life. The category, once created, made up a kind of person. People came to understand themselves as homosexuals, organised their lives around the category, formed communities and changed the category by living it. Hacking traced similar histories for multiple personality, child abuse, autism and the "fugue" of nineteenth-century French psychiatry.

Dynamic nominalism

Hacking called his position dynamic nominalism. It is nominalist in that kinds of people are not natural kinds waiting to be discovered. They are made, partly by the classifications applied to them. It is dynamic in that the making is a process. Classifications and the classified interact over time, each changing the other. A category is introduced, people are classified, the classified respond (by adopting, resisting or modifying the category), their responses change what the category means, the category is revised, and so on. This is the looping effect of human kinds.

Hacking and the instruments

Hacking's analysis bears on every classificatory act in anthro-complexity's practice. When a Cynefin categorisation calls a situation "complex", it makes up a kind of situation, with its own legitimate responses, and the people in it come to understand themselves as people in a complex situation, which changes what they do. When a SenseMaker pattern identifies a group as telling stories of "compliance", it makes up a kind of group, and the group responds to the classification. When SwarmCompass, in The Compass and the Loom's reading, sets out to "find the 17%" who see what others miss, it makes up a kind of person, the seer, and the people classified as seers, or as not-seers, respond to the classification. The Tidal Loom's frames, not seers is a refusal of this classification. Different people see different gorillas, so the category "seer" is a feature of the frame, not of the person.

Hacking's dynamic nominalism also says what a responsible practice of classification would do. It would track the loop: record each classification, observe how the classified respond, and revise the classification in light of the response, treating the classified as participants in the classification's history and not as its objects. That is Thesis VIII's design principle (track reflexive content, date the instruments), stated in Hacking's terms.

The Three Together

Bourdieu, Giddens and Hacking differ in vocabulary, emphasis and politics, but on the questions this disputation asks they converge:

Question Bourdieu Giddens Hacking
How is structure carried? In habitus: embodied, durable dispositions In practical consciousness; drawn on in action In classifications people live by
How is structure reproduced? Habitus generates practices fitted to its conditions Duality of structure: medium and outcome of practice People live the categories and so sustain them
How does structure change? When field and habitus diverge (hysteresis); when doxa is challenged Through reflexive monitoring and institutional reflexivity Through the looping of classification and response
What does the observer get wrong? The scholastic fallacy: projecting skholè onto practice Underestimating agents' knowledgeability Treating interactive kinds as indifferent kinds
What is the remedy? Participant objectivation: objectify the objectifying subject Recognise the double hermeneutic; agents already interpret Track the loop; dynamic nominalism

What They Lend Anthro-Complexity

Each thinker lends a concept that a reflexive anthro-complexity would need.

From Bourdieu: hysteresis and participant objectivation. Hysteresis adds a third axis to estuarine mapping. Beside the energy cost and timescale of changing a constraint, there is the lag of the habitus that carries it. A constraint may be formally changeable and still persist in bodies, or formally fixed and already eroded in practice. Participant objectivation turns the reading of the reader into a scientific discipline: practitioners analyse their own position in the field of complexity consultancy, its capital and its doxa, with the tools they apply to clients.

From Giddens: practical consciousness and institutional reflexivity. Practical consciousness names what narrative instruments cannot collect: the tacit, embodied knowledge of how to go on, which carries the constitutive constraints. An instrument aware of it would look for traces, in what stories take for granted, in what narrators find too obvious to say, and would supplement narrative with observation of practice. Institutional reflexivity names the instruments' place in modernity. They are expert systems that disembed local knowledge and outdate themselves by their own incorporation, so a reflexive practice would design for both.

From Hacking: the loop. Every classification makes up people, and every pattern, category and frame is a classification. A reflexive anthro-complexity would track its classifications' loops: how the classified respond, how the response changes the category, and when the category should be retired.

The Scholastic Fallacy and Anthro-Complexity

The scholastic fallacy deserves a closing word, because it names the error this disputation has been tracing more precisely than any other single concept.

Anthro-complexity is a practice, not a scholastic discipline. Snowden has spent his career in organisations, governments and militaries, not in the seminar room, and his hostility to theory detached from practice is well known. It would be odd to accuse him of the scholastic fallacy, and this chapter does not. But the scholastic fallacy is not a vice of scholars only. It is a vice of position: of anyone who observes practice from a position of relative leisure, distance from the stakes and representational tools, and projects that position onto the practitioners observed. The consultant, the analyst and the designer of instruments occupy such a position, relative to the populations they sense. They observe from a distance, with time to reflect, through dashboards and maps, and their representations of practice are shaped by their position.

The estuary is the scholastic fallacy's image: the sailor at leisure on the deck, reading the tide, able to wait. The fisherman in the maelstrom is its correction: the practitioner in the urgency, observing from inside, detached only by achievement. Bourdieu's participant objectivation is the discipline that moves the practitioner from the deck towards the vortex. It is not by abandoning reflection, which the fisherman needs as much as the sailor, but by including the reflector's position in what is reflected on.

The adversary will say that Bourdieu, Giddens and Hacking are structural sociologists whose concepts have never been operationalised into practical instruments, and that anthro-complexity, which must act, cannot wait for sociology to supply methods it has never supplied. Bourdieu's participant objectivation is a counsel for academic sociologists writing books, not for practitioners facilitating workshops on a Tuesday afternoon. Hysteresis cannot be mapped, practical consciousness cannot be collected, and the loop cannot be tracked in real time.

The adversary is partly right. These concepts were not designed as instruments, and some cannot be fully operationalised. But each can be approximated. Hysteresis can be approached by asking, of each constraint on an estuarine map, "is this still how people actually behave, or only how the rules say they should?" Practical consciousness can be approached by observation alongside narrative, and by asking narrators what their stories take for granted. The loop can be tracked across rounds by coding reflexive content (Thesis VIII). Participant objectivation can be approached by the reading of the reader and the declination table. None of these approximations is complete, and all are better than leaving the concepts out. partial

STRUCTURE IS CARRIED IN THE BODY —

REPRODUCED IN THE ACT —

AND MISREAD BY WHOEVER HAS LEISURE TO WAIT FOR THE TIDE.

Addition: Cross-Lights From Book Nine

Luhmann, as Snowden now reads him. This chapter used Luhmann for the blind spot and second-order observation. Snowden's post of 14 November 2026 agrees that a system cannot observe the distinction by which it observes. It rejects Luhmann's placing of persons in the environment of the social system, as this chapter did. And it attacks two uses of his theory by consultants: as an alibi, since any outcome can be redescribed as the system's own processing, and as a licence for the consultant's attributions to displace people's own. Chapter 52 applies both tests to anthro-complexity.

Habitus and formation. A Web of Relations treats habitus as what a practitioner's formation re-forms. See Chapter 51.

Chapter 34

Performativity: When the Model Moves the Market

4,233 words, about 18 minutes of reading

An Engine, Not a Camera

The title of Donald MacKenzie's 2006 study of financial economics, An Engine, Not a Camera, is a thesis in five words. Economic models do not merely photograph markets, recording what is there. They drive markets, shaping what happens. The model is an engine. This chapter, the last of Book Five, takes performativity as the general theory of which the reflexive reason (Chapter 14), the asymmetry principle applied to itself (Thesis II) and the degradation of instruments (Thesis VIII) are special cases. It traces the concept from its origin in J. L. Austin's philosophy of language, through Judith Butler and Barry Barnes, to Michel Callon's and MacKenzie's sociology of markets. It then applies MacKenzie's typology of performativity to anthro-complexity's instruments, and argues that a reflexive anthro-complexity would design its instruments as engines, declaring and tracking their performative effects, rather than presenting them as cameras.

Austin: Doing Things With Words

J. L. Austin's How to Do Things with Words (lectures delivered in 1955, published 1962) began from a simple observation. Not all utterances describe. Some do. "I name this ship the Queen Elizabeth", said while smashing a bottle against the bow, does not describe a naming. It is a naming. "I do", said at the right moment in a wedding, does not report a marriage. It makes one. "I bet you sixpence it will rain tomorrow" does not describe a bet. It places one. Austin called such utterances performatives, in contrast to constatives, which describe and can be true or false.

Performatives cannot be true or false, but they can go wrong. Austin called the ways they can go wrong infelicities, and identified the felicity conditions a performative must meet to succeed:

  • there must be an accepted conventional procedure with a conventional effect;
  • the persons and circumstances must be appropriate for invoking the procedure;
  • the procedure must be executed correctly and completely;
  • the participants must have the requisite thoughts and feelings, and must conduct themselves accordingly.

A naming performed by someone without authority to name the ship is a misfire. A wedding conducted by someone not authorised to marry is void. A bet that the other party does not accept has not been placed. Performatives succeed only within institutions, conventions and relations of authority, and their success depends on uptake, the hearer's recognition of what the utterance does.

Austin's later lectures complicated the distinction. He came to see that constatives also do something (asserting, stating, informing) and that the distinction between performative and constative is not sharp. Every utterance has a locutionary dimension (what it says), an illocutionary dimension (what it does in saying it: warns, promises, asserts, orders) and a perlocutionary dimension (what it brings about by saying it: alarm, persuasion, compliance). Description is itself an act.

Austin and Searle

Searle, Austin's student, systematised speech act theory in Speech Acts (1969), and his later social ontology (Chapter 13) grew from it. Status function declarations, which create institutional facts by representing them as existing, are performatives. "The meeting is adjourned", "you are hired" and "this situation is complex", said by an authorised person in an organisation that accepts the authority, make it so.

Butler: Iterability

Judith Butler, in Gender Trouble (1990) and Bodies That Matter (1993), extended performativity from discrete speech acts to the ongoing constitution of identity. Drawing on Austin and on Jacques Derrida's critique of him, Butler argued that gender is not an inner essence that expresses itself in behaviour, but a performative accomplishment, constituted through the repeated performance of gendered acts, gestures and styles. The repetition is crucial. Derrida had argued that performatives succeed because they cite established forms. A wedding vow works because it repeats the conventional formula. Butler argued that identity is constituted by iterated citation of norms, and that because every citation is a repetition with a difference, there is always the possibility of citing differently, of subversive repetition that shifts the norm.

Butler's contribution matters here because it carries performativity from the act to the identity, from what is done with words to who one becomes by doing. Identities are performed, and performed iteratively. They are not held, but enacted again and again, and each enactment can reproduce or shift them. That is Thesis I's relational account of identity in the vocabulary of performativity. It is also Giddens's duality of structure applied to the self. Identity is the medium and outcome of the performances that constitute it.

Barnes: Self-Referring Knowledge

Barry Barnes, a sociologist of science, proposed in a 1983 paper that much social knowledge is self-referring: knowledge that refers to a state of affairs that exists partly because the knowledge exists. His example was money. A coin is money because people believe it is money and act accordingly, and their belief that it is money is part of what makes it money. Social institutions are, on Barnes's account, systems of self-referring knowledge, of "bootstrapped induction" in which people's expectations about what others will do produce the very regularities the expectations predict. Barnes's account underlies what MacKenzie called Barnesian performativity.

Callon: The Laws of the Markets

Michel Callon's introduction to The Laws of the Markets (1998) made the performativity thesis central to economic sociology. Callon argued that economics, the discipline, does not describe an economy that exists independently of it. It performs the economy. It shapes markets by supplying the concepts, models, tools and devices through which markets are organised. Markets are not natural phenomena that economics discovers. They are constructed, partly by economists, through the "market devices" (pricing formulae, auction designs, accounting standards, trading platforms) that embody economic theory. Callon's formulation was that economics, broadly understood, performs and formats the economy instead of merely observing it.

MacKenzie's Typology

MacKenzie's An Engine, Not a Camera studied the history of financial economics, especially option pricing, and asked how far the theory had performed the markets it described. To answer precisely, MacKenzie distinguished several kinds of performativity, ordered by strength.

Generic performativity. A theory, model, concept or procedure is used by participants in economic processes, by regulators or by others. The theory is part of the world it describes, in the minimal sense that people use it. Most economic theories are generically performative.

Effective performativity. The practical use of the theory has an effect on economic processes. Prices, volumes and practices are different from what they would have been without it. A theory that is used but changes nothing is generically but not effectively performative.

Barnesian performativity. The practical use of the theory makes economic processes more like their depiction by the theory. This is the strongest form, named for Barnes. The theory's use brings reality into line with it. MacKenzie's central case was the Black–Scholes–Merton options pricing model. When first published in 1973, it fitted observed prices poorly. As traders adopted it, using it to price options, identify mispricing and hedge positions, prices converged towards the model's predictions. By the late 1970s and early 1980s the fit was excellent. The model had become accurate because it was used.

Counter-performativity. The practical use of the theory makes economic processes less like their depiction by it. After the stock market crash of October 1987, option prices departed systematically from Black–Scholes in a persistent pattern, the "volatility smile", in which out-of-the-money options were priced higher than the model predicted. MacKenzie argued that the crash itself had been partly produced by "portfolio insurance", a hedging strategy based on option-pricing theory, whose widespread use contributed to the cascade of selling. The theory's use had produced an event that broke the theory's fit.

Type Definition Black–Scholes example
Generic The theory is used Traders, exchanges and regulators use the model
Effective Its use changes outcomes Option pricing and trading practices change
Barnesian Its use makes reality more like the theory Prices converge on model predictions (1970s–1987)
Counter-performative Its use makes reality less like the theory Portfolio insurance contributes to the 1987 crash; the volatility smile persists

The Typology Applied to Anthro-Complexity

Apply MacKenzie's typology to anthro-complexity's instruments and concepts.

Cynefin

Generic. Cynefin is widely used by leaders, consultants, militaries, governments and educators. It is part of the world it describes.

Effective. Its use changes outcomes. Situations categorised as complex are handled differently, by probes instead of plans and by tolerance of failure, from how they would have been handled otherwise. Chapter 13's Searle analysis showed the mechanism: the category, once accepted, creates deontic powers.

Barnesian? Does Cynefin's use make situations more like its depiction of them? There is a plausible case that it does, in at least one respect. An organisation that categorises a situation as complex and responds with probes may produce, through the probes, more emergence and more interaction than a planned response would have. A planned response might have made the situation more ordered, by imposing structure, while the probe-sense-respond response keeps it complex. The categorisation helps produce the kind of situation it describes. This is not a criticism. It may be exactly the right outcome. But it means the categorisation is not a neutral reading. It is partly self-fulfilling.

Counter-performative? Could Cynefin's use make situations less like its depiction? One mechanism is plausible. If "complex" becomes a licence that managers seek, because it relieves them of accountability for plans, situations may be categorised as complex that are not, and probe-sense-respond applied where analysis would have worked. The framework's distinctions would then blur in practice, and its depiction of the world as divided into domains with different appropriate responses would fit less well. Snowden has warned against exactly this misuse. The warning acknowledges the counter-performative risk.

SenseMaker

Generic and effective. SenseMaker is used, and its patterns change decisions.

Barnesian. Thesis VIII's mechanisms predict Barnesian performativity over repeated rounds. Narrators learn the signifiers, adopt them as self-description and tell stories that fit them, so the population's narration converges on the instrument's categories. The instrument's depiction of the population becomes more accurate because the population has been shaped by it.

Counter-performative. Thesis VIII also predicts counter-performativity. When patterns are acted on, subsequent narratives respond to the action, and the instrument's depiction ("this is what people experience") fits less well, because what people now experience is partly the response to the last depiction. If the organisation reads the new patterns as fresh information, it steers by its own wake.

Estuarine mapping

Barnesian. A map that is accepted defines what counts as costly to change, and the organisation leaves the "costly" constraints alone and works on the "cheap" ones. The constraints mapped as fixed stay fixed, partly because they were mapped as fixed. That is Barnesian performativity. The map becomes accurate through being believed.

The slogan

Effective. Humans are not ants has changed how organisations use swarm metaphors and agent-based models. Its performativity is, on this disputation's assessment, beneficial.

This disputation

Generic and effective, if read. This book is a move in the field it describes. If it is read and its arguments taken up, it will change how anthro-complexity's instruments are understood and perhaps designed. That is its intention, declared here. Its performativity, like every description's, is subject to the same analysis. It may be Barnesian (practitioners begin to look for the possessive fallacy and find it), counter-performative (its emphasis on reflexivity produces paralysis), or ineffective. The author cannot know which, and will not be the one who decides.

Austin's Felicity Conditions and the Authority of Frameworks

Austin's felicity conditions add a dimension MacKenzie's typology does not. A performative succeeds only if the persons and circumstances are appropriate: if the person performing it has the authority to do so, within an accepted conventional procedure. A Cynefin categorisation is a performative in Austin's sense. "This situation is complex", said by an authorised person, makes it so in the organisation. Its felicity depends on authority. Who is authorised to categorise? A chief executive's categorisation will be felicitous. A junior employee's may misfire, not because it is less accurate but because she lacks the authority to make it stick.

This is power (Thesis V) in Austin's vocabulary. The performative force of anthro-complexity's frameworks depends on who uses them, and the frameworks' performativity therefore follows the organisation's distribution of authority. A framework that, used by those with authority, creates deontic powers (licences, permissions, exemptions) distributes those powers according to the existing authority structure. It adds to the capacities of those who already have the authority to declare. Snowden's insistence on collective categorisation distributes the locution, so that everyone can say "this is complex". It does not distribute the felicity conditions. Whose saying it makes it so is still a matter of position.

Designing Engines

If anthro-complexity's instruments are engines, not cameras, they should be designed and presented as engines. That means declaring what they are meant to do, tracking what they actually do, and taking responsibility for both. MacKenzie's typology and Austin's felicity conditions suggest a set of design principles.

Declare the intended performativity. Each deployment states what it is meant to do to the system, as well as what it is meant to reveal. A SenseMaker study meant to surface hidden concerns is meant to make them discussable, which changes the organisation's doxa. A Cynefin categorisation meant to license experiment is meant to change who may try what. An estuarine map meant to find room to act is meant to change what people believe is possible. Declaring this is the declination before direction principle extended from expectation to intention.

Track the type. For each deployment, ask over time whether its performativity is generic, effective, Barnesian or counter-performative. Is the instrument changing outcomes? Is it making the system more like its depiction, and is that desirable? Is it making the system less like its depiction, and is it steering by its own wake? This is Thesis VIII's reflexive-content tracking, organised by MacKenzie's typology.

Distribute the felicity conditions. Where performatives create deontic powers, design so that the authority to perform them is distributed, not only the locution. A team should be able to declare its own situation complex, with the licences that follow, without that declaration requiring ratification from above. This is a design for the politics of performativity, and it connects to CST's participants dimension (Chapter 12).

Expect counter-performativity. Every instrument whose use changes the system will, in time, change it in ways that break the instrument's fit. Design for retirement. Date the instruments (Thesis VIII), rotate the frames (the Tidal Loom) and treat a degrading fit as a signal that the instrument has done its work, not as a failure to be patched.

Seek uptake, not capture. Austin's uptake, the hearer's recognition of what the utterance does, is a relational condition of performative success. An instrument that seeks uptake addresses those it describes, so that they recognise what is being done and can accept, contest or modify it. An instrument that seeks capture collects from them without their recognition. The first is communicative action and the second strategic (Chapter 16).

Principle MacKenzie / Austin concept Tidal Loom or earlier thesis
Declare intended performativity Effective and Barnesian Declination before direction
Track the type Generic / effective / Barnesian / counter Thesis VIII: reflexive content; noticing, not mapping
Distribute felicity conditions Austin's appropriate persons Thesis V: power; CST participants dimension
Expect counter-performativity Counter-performativity Dated instruments; frames, not seers
Seek uptake, not capture Austin's uptake Chapter 16: address, not influence; recognition, not alignment

Performativity and the Human Distinction

The chapter closes Book Five by returning performativity to the human distinction. Chapter 14 argued that descriptive reflexivity is the human difference of kind: human systems are changed by public descriptions of themselves. Performativity is the general theory of descriptive reflexivity. Descriptions are acts, they change what they describe, and their success depends on uptake within institutions of authority. The human distinction, stated in the vocabulary of performativity, is this. Human systems are systems in which descriptions perform. Non-human systems are systems in which descriptions, however accurate, only constate.

A camera aimed at a starling flock constates. An engine installed in a human organisation performs. Anthro-complexity's instruments are installed in human organisations. They are engines, and the human distinction that anthro-complexity places at its heart says so. A practice faithful to its own distinction would treat its instruments accordingly.

Book Five Concluded

Book Five has brought eight bodies of thought to bear on anthro-complexity's human cluster: Mead, Elias, Juarrero, Ingold, Hutchins, Peirce, the sociologists of structure, and the theorists of performativity. The convergence is striking.

Thinker The human is… The observer is… The instrument should…
Mead A social process; I and me A participant whose readings are gestures Attend to the I where categories fail
Elias A nexus in a figuration; homo apertus A fisherman in the maelstrom Ask whose description the pattern describes
Juarrero A persisting pattern of context-sensitive constraint A component of the pattern it reads Map constitutive constraints and closure
Ingold A line of becoming; a knot in the meshwork A wayfarer, not a navigator Correspond, not capture
Hutchins A part of a distributed cognitive system One horizon among overlapping horizons Make horizons overlap; mutual dashboards
Peirce A sign; not absolutely an individual A member of the community of inquiry Return the interpretant to the community
Bourdieu, Giddens, Hacking Habitus; knowledgeable agent; made-up person Prone to the scholastic fallacy Objectify the objectifier; track the loop
Performativity Constituted by iterated performance An engine, not a camera Declare and track its performativity

Every row says the same thing in a different vocabulary. The human is relational, the observer is inside, and the instrument is an actor. Book Six now reads the same convergence through the Chinese philosophical triad that runs through all of Principia Humanitatis, 道理勢, and finds it there too.

The adversary will say that performativity has become a fashionable concept that explains everything and therefore nothing. If every description performs, then the concept cannot distinguish good descriptions from bad, accurate models from inaccurate, or useful instruments from harmful ones. Black–Scholes was performative, and it was also a brilliant piece of financial mathematics, whose accuracy before 1987 was not only an effect of its use. To reduce instruments to their performativity is to lose the question of whether they are true.

The adversary wins this half, and it is an important one. Performativity does not replace accuracy. A model can be both accurate and performative, and the question of its accuracy remains. MacKenzie himself insisted that Black–Scholes was not merely performative. It captured real features of options markets that made it an attractive model to adopt. The chapter's claim is not that instruments are only performative. It is that they are also performative, and that a practice which presents them only as accurate cameras, and not also as engines, will be surprised by what they do. Accuracy and performativity are two questions, and a reflexive practice asks both. partial

SOME WORDS DESCRIBE —

SOME WORDS DO —

IN HUMAN SYSTEMS EVERY DESCRIPTION ALSO DOES.

Addition: The Performativity of "Complexity" Itself

The chapter applied MacKenzie's typology to anthro-complexity's instruments and slogans. This addition applies it to the word at the centre of the whole field: complexity. The word has had a career in management over the past thirty years, and the career is a case study in performativity: a concept that, adopted, changed the organisations that adopted it. The career is also a case of Kuhn's fate (addition to Chapter 18), a vocabulary taken up outside its home field and reshaped by its users.

Three phases

The career of "complexity" in management can be sketched in three phases. The dates are approximate.

Phase one: the new science (1990s). Following Waldrop's Complexity (1992), Wheatley's Leadership and the New Science (1992) and the Santa Fe Institute's growing profile, "complexity" entered management discourse as the name of a new science that promised to overturn mechanistic management. Organisations were "complex adaptive systems". Leaders should work "at the edge of chaos". Strategy should be "emergent". In this phase the word carried the authority of physics and biology, and its performativity was mostly generic. Managers used it, but it changed little in practice, because it came without methods.

Phase two: frameworks (2000s–2010s). With Cynefin, the work of Stacey and the CRP school, Plsek and Greenhalgh in health care, and a growing literature on complexity in public policy and development, "complexity" acquired frameworks and methods. Its performativity became effective. Situations categorised as complex were handled differently, probes replaced plans in some organisations, and development agencies adopted "adaptive management". The word now did something. It licensed experiment, relieved managers of the demand for certainty, and justified new kinds of practice.

Phase three: the ubiquitous adjective (2010s–2020s). By the 2020s "complex" had become a ubiquitous adjective in organisational life. Every situation was complex, every environment VUCA (volatile, uncertain, complex, ambiguous), and every challenge called for "navigating complexity". In this phase the word's performativity became, in part, counter-performative. Its ubiquity eroded the distinctions that had given it force. If everything is complex, nothing is distinctively so, and the licence the word conferred became universal and therefore meaningless. Worse, "complexity" became an excuse. Failures were attributed to the complexity of the environment, plans abandoned because the situation was complex, accountability diluted because outcomes could not be predicted. Snowden's warnings against this misuse, his insistence that most situations are not complex and that the complex domain should not become an alibi, are warnings against the word's counter-performativity.

The performativity types, mapped

Phase Performativity What the word did
New science Generic Used; little practical change
Frameworks Effective; partly Barnesian Licensed probes; situations treated as complex became more emergent
Ubiquitous adjective Counter-performative Distinctions eroded; became an excuse; licence universal and meaningless

Barnesian complexity

The Barnesian moment of phase two deserves attention. Chapter 34 suggested that Cynefin's categorisation of a situation as complex may help produce the kind of situation it describes. Probes multiply interactions, licensed experiment produces emergence, and the refusal to impose a plan keeps the situation open. At the level of the field, something similar may have happened. As organisations adopted complexity frameworks, they began to behave more like complex systems, decentralising decisions, running experiments, tolerating emergence and attending to weak signals. The frameworks' descriptions became more accurate because they were used.

That is not a criticism. It may be among the best things complexity thinking has done for organisations. But it means the field's empirical claims about organisational complexity are partly claims about organisations that have been shaped by the field. An organisation that has adopted Cynefin is not a neutral test of Cynefin's claims. It is a site of Cynefin's performativity.

Counter-performative complexity

The counter-performative moment of phase three is more troubling, and it bears directly on anthro-complexity's positional reason (Chapter 17). When a field's central term becomes ubiquitous, the field faces a choice. It can defend the term, insisting on its proper use, distinguishing real complexity from false and policing the boundary. Snowden has done this energetically. Or it can replace the term with one that has not yet been diluted, and "anthro-complexity" may be partly this: a term that marks a border against the diluted "complexity" of management fashion, as well as against CAS and systems thinking.

Either strategy is a move in the field's performativity. Defending the term keeps it in circulation and may slow its dilution. Replacing it starts a new career, which will follow its own phases. "Anthro-complexity" is in phase one of its career. It is used by a community of practitioners, with frameworks and methods attached, and its performativity is beginning to be effective. If it spreads, it will face the same fate as "complexity". It will be taken up, simplified, made ubiquitous, and in time used as an excuse. Element 22, resistance to commodification, will then be tested on the school's own name.

What a reflexive field would do

A field reflexive about its own performativity would do three things.

Track its vocabulary's career. It would watch for the signs of each phase: generic use without practice, effective use with distinctions, and counter-performative ubiquity. It would respond to each, defending distinctions when they are eroded and retiring terms when they have become excuses.

Distinguish its findings from its effects. It would recognise that organisations shaped by its frameworks are not neutral tests of them, and would seek evidence from organisations that have not adopted the frameworks, or would design studies that separate the frameworks' accuracy from their performativity.

Expect its own counter-performativity. It would anticipate that its successful terms will be diluted, and would design for that, by dating its vocabulary (as Thesis VIII dates its instruments), by preferring terms that resist simplification, and by teaching practitioners to recognise when a term has stopped doing work.

FIRST THE NEW SCIENCE — THEN THE FRAMEWORK — THEN THE UBIQUITOUS ADJECTIVE —

THE WORD THAT LICENSED EXPERIMENT BECAME THE ALIBI FOR FAILURE —

DATE YOUR VOCABULARY AS YOU DATE YOUR INSTRUMENTS.

Addition: Cross-Lights From Book Nine

Performativity as altered compressibility. MacKenzie showed a pricing model making a market fit it. Chapter 56 states the general case. In human systems a description is a compression performed from inside, and it makes what it describes more regular in its own terms. On this account the old ethics of modelling, which concerned what a model leaves out, gives way to a more severe one, which concerns what a description makes.

The demon. The Zen Garden study uses Maxwell's demon to the same end. Sorting is work done from inside the box, and it is paid for.

Chapter 35

勢 Shì: Propensity Includes the Strategist

3,681 words, about 16 minutes of reading

The Spine

All the work of Principia Humanitatis runs along a spine drawn from classical Chinese thought: the triad 道理勢, dào, lǐ, shì. In the reading used throughout this series, 道 is the way, the course along which things go, made in the going. 理 is the grain, the pattern inherent in things that a skilled hand follows. 勢 is the propensity, the potential of a configuration that carries things in one direction rather than another. The three are not separate principles but aspects of one situation. The way runs along the grain, carried by the propensity. A practice that grasped one without the others would be partial in a way the triad is designed to expose.

The Compass and the Loom applied the spine to the Cynefin Company's instruments and reached a verdict this Book develops. The instruments grasp one term of the triad, 勢, and lose 理 and 道. Their vocabulary is Jullien-like, but their practice is closer to Western means–end planning. This chapter examines 勢 itself, the term anthro-complexity has grasped, and argues that even 勢 has been grasped partially. Its strategic sense, in which the strategist is part of the configuration, and its political sense, in which 勢 is positional power, have been set aside in favour of a sense closer to the "propensity of things" read from outside. The next two chapters take 理 and 道, and the fourth takes up Snowden's question about dyads, triads and mediation in the Chinese sources.

Sunzi's 勢

The Sunzi bingfa, the Art of War attributed to Sunzi, is the classical locus of 勢 as a strategic concept. Its fifth chapter bears the title 勢 (rendered variously as energy, momentum, potential, strategic advantage or configuration), and it is the chapter from which Jullien draws much of his account.

Sunzi illustrates 勢 with images that have become famous. The onrush of a torrent that tosses boulders about in its course is 勢. The swoop of a falcon that strikes and kills its prey is timing. The skilled commander's 勢 is like a drawn crossbow, and his timing like the release of the trigger. Above all, at the chapter's close, Sunzi offers the image of round stones on a slope. The 勢 of troops skilfully commanded is like that of round stones rolled down a mountain thousands of feet high. Stones on flat ground stay still, while round stones on a steep slope roll. The potential is in the configuration, the roundness of the stones and the steepness of the slope, and the commander's art is to create a configuration in which his forces have that potential.

Sunzi says something else in the same chapter that is less often quoted and more important for this disputation. The skilful commander seeks victory from 勢, and does not demand it of his men. He selects men and relies on 勢 (擇人而任勢). The commander does not exhort his soldiers to be brave. He places them in a configuration where bravery comes naturally, as stones roll on a slope. Soldiers on "death ground", with no retreat, fight with desperate courage, not because they are brave but because the configuration leaves no alternative. This is Snowden's behaviour is the shadow of conditions, stated two and a half thousand years earlier, and with the same ethical ambiguity that Chapter 16 found in the shadow principle. The commander shapes his soldiers' conduct by configuring their situation, not by addressing their will.

形 and 勢

Sunzi distinguishes 勢 from 形 (xíng), the subject of his fourth chapter, usually translated disposition or form. 形 is the visible arrangement of forces, their positions, numbers and deployment, which can be seen and assessed. 勢 is the potential that the arrangement releases when it is set in motion, the dynamic that arises from the configuration. 形 is static and 勢 dynamic. 形 can be counted and 勢 must be felt. Jullien makes much of the distinction. Western strategy, he argues, attends to 形, to forces and positions that can be represented on a map and compared. Chinese strategy attends to 勢, to the potential that arises from the configuration and cannot be read off a map.

地形: terrain is not 勢

There is a further distinction in Sunzi that bears directly on Thesis III. His tenth chapter is titled 地形 (dìxíng), terrain. It classifies kinds of ground (accessible, entangling, temporising, narrow, precipitous, distant) and the dispositions appropriate to each. Terrain is the landscape on which forces are arrayed. It can be surveyed, mapped and classified. Sunzi treats it as an aid to the general, something he must know. But terrain is not 勢. Terrain is the ground, and 勢 is the potential the general creates by how he configures his forces on the ground, in relation to the enemy, at a moment. The same terrain can yield very different 勢 depending on the configuration.

The distinction is exact for this disputation's purposes. An estuarine map is a map of terrain, 地形: it surveys constraints, classifies them by energy cost and timescale, and shows which ground is accessible and which precipitous. It is valuable, as Sunzi's chapter on terrain is valuable. But it is not a map of 勢, because 勢 cannot be mapped. It arises from the configuration of forces on the terrain, including the practitioner's own forces and the other side's reading of them, at a moment. The Compass and the Loom's observation that mapping propensity as "energy gradients" turns it into a landscape is, in Sunzi's own vocabulary, the observation that the estuarine map mistakes 地形 for 勢.

勢 Includes the Strategist

The single most important feature of Sunzi's 勢 for this disputation is that it includes the strategist. The configuration whose potential the general reads is not a landscape he surveys from a hill. It is a configuration of which he is a part: his forces, his position, his reputation, his visible dispositions, and the enemy's reading of all of them.

The point is explicit in Sunzi's doctrine of deception. All warfare is based on deception. When able to attack, seem unable. When using forces, seem inactive. When near, make the enemy believe you are far away. Deception is possible only because the general is read by the enemy, and his being read is part of the configuration. A general who could not be seen could not deceive. A general who did not know he was seen could not use deception. Sunzi's whole strategic art presupposes that the strategist is inside a configuration of mutual reading, that his appearance is a factor in the 勢, and that managing his appearance is part of creating the 勢.

The point is also explicit in Sunzi's sixth chapter, on 虛實 (xū shí), empty and full, weak points and strong. The skilful general makes the enemy's full empty and his own empty full, by creating appearances that lead the enemy to concentrate where the general is not and to disperse where the general will strike. Appear at points the enemy must hasten to defend; march swiftly to places where you are not expected. The empty and full are not properties of the terrain. They are properties of the configuration as the enemy reads it, and the general creates them by shaping the reading.

Jullien draws the consequence in A Treatise on Efficacy. The Chinese strategist does not impose his will on the situation. He conditions it, and the conditioning includes the conditioning of how the other side perceives the situation. Efficacy is the art of making the other side's perception work in one's favour. That art is possible only for a strategist who knows that he is inside the configuration and that his visibility is one of its forces.

The Political 勢: Positional Power

勢 has a second classical sense that anthro-complexity's use leaves out entirely, and it bears on Thesis V. In the political thought of the Warring States, especially in Shen Dao and in the Legalist synthesis of Han Feizi, 勢 is positional power: the power that comes from occupying a position, independent of the virtue or ability of the person who occupies it.

Shen Dao's image, preserved in the Han Feizi, is of the flying dragon that rides the clouds and the soaring serpent that wanders in the mist. When the clouds disperse and the mist clears, the dragon and serpent are no different from earthworms, because they have lost what they rode on. A ruler's power is like this. A worthy man without position cannot command even three people. An unworthy man with position can command the world. Power lies not in the person but in the 勢, the seat of authority. Han Feizi built his theory of government on three elements: fǎ 法 (law, the published rules), shù 術 (technique, the ruler's methods for controlling ministers) and shì 勢 (position, the authority of the throne). The ruler who held 勢 firmly need not be a sage. He needed only to keep the position and use law and technique.

The political 勢 is the power of position in a configuration. It is what Elias called a power ratio (Chapter 28), what Abbott called jurisdiction (Chapter 17) and what Austin's felicity conditions require (Chapter 34): the authority that makes a performative stick. It is the strategist's 勢 seen from the inside of a human institution. The person at the centre of a configuration of dependencies has a potential that others lack, which arises from where she sits.

Jullien discusses the political 勢 in The Propensity of Things, and Law and Lin are alert to it. But anthro-complexity's use of 勢 is almost entirely of the first, strategic-cosmological sense, propensity as the potential of a configuration of things, and not of the political sense, propensity as the potential of a position among people. The omission matters because the political sense is where 勢 meets power. A practitioner who reads the propensity of an organisation without reading the positional 勢 of the people in it, who sits where and whose seat carries what authority, has read the slope without seeing who stands at the top of it with a lever.

Sense of 勢 Classical source What it means Anthro-complexity's use
Strategic Sunzi, ch. 5 Potential of a configuration of forces, including the commander's, read by the enemy Taken as propensity; the commander's inclusion and mutual reading dropped
Cosmological Jullien's Propensity of Things; calligraphy, landscape The dynamism inherent in the disposition of things Taken as the propensity of complex systems
Political Shen Dao; Han Feizi Positional power; authority of the seat, independent of the occupant Absent

勢 Without the Strategist Is a Landscape

The consequence of dropping the strategist's inclusion and the political sense can now be stated in the triad's own terms. A 勢 read from outside, as the propensity of a configuration the reader is not part of, is not 勢. It is 地形, terrain: a landscape of potentials that can be surveyed and mapped, along which forces will tend to move, and across which the reader plots a course. Terrain is real and worth knowing, and Sunzi devotes a chapter to it. But the art of 勢 begins where terrain ends, with the configuration of forces on the terrain, the reader among them, read by others.

Anthro-complexity's surfer reads a wave. The wave is terrain in motion, a moving landscape of potential, and the surfer's art is to read it and join it. The surfer is not part of the wave's configuration, and the wave does not read the surfer. The surfer's art is therefore an art of terrain, not of 勢 in Sunzi's sense. Sunzi's general is part of the configuration, read by the enemy, and his art is to shape the configuration including how he is read. The general's art is the art of 勢.

Human organisations are configurations in which every practitioner is read. A consultant who reads an organisation's propensities is being read by the organisation as she reads, so her reading is a factor in the propensity. A leader who waits for the tide is seen waiting, and her waiting conditions the configuration. A designer of instruments whose signifier set will be learned by narrators (Thesis VIII) is part of the configuration of narration. The 勢 of a human organisation includes all of them. A practice of 勢 in human organisations is a practice of the general, not the surfer. It must include the practitioner in the configuration she reads, attend to how she is read, and take responsibility for the conditioning her presence and actions produce.

Proposition granted: You work with the dispositional state of a complex system and move with its propensity; you cannot engineer it causally. (Snowden, endorsing Ing's surfing image and his own sailing image, via Jullien.)

Then it must be granted that propensity, as Jullien reads it from Sunzi, is 勢. — Granted. That is the source.

Then it must be granted that Sunzi's 勢 includes the commander's own forces and dispositions, and the enemy's reading of them. — Granted. The doctrines of deception and of empty and full presuppose it.

Then it must be granted that, in a human organisation, the practitioner's own presence, readings and dispositions are part of the configuration whose propensity she reads. — Granted.

Then it must be granted that the surfing and sailing images, in which the reader is outside the medium and unread by it, picture terrain (地形) and not 勢. — Granted.

Then the respondent must say whether anthro-complexity's practice of reading propensity includes the practitioner and her being read. — Not structurally. The estuarine map maps constraints, not the mapper. SenseMaker reads narratives, not the reading of the reader. The practice of propensity is a practice of terrain.

Law and Lin's Shi-Inflected Social Science

The November post cites John Law and Wen-yuan Lin's work on a shi-inflected social science as one of the sources of Ing's emphasis on propensity. Their work deserves a closer look, because it shows what a social science informed by 勢 might look like, and it is more alert to the strategist's inclusion than anthro-complexity's use.

Law, one of the founders of actor-network theory, and Lin, a sociologist of science and medicine in Taiwan, have argued in a series of papers that Western social science, built on European categories, systematically fails to see some features of non-Western practices, and that Chinese concepts, 勢 among them, could serve as alternative analytic resources. In their work on Taiwanese practices of Traditional Chinese Medicine and on postcolonial science, they used 勢 to describe the dynamics of situations that resist representation as structures of discrete entities. Practitioners of Chinese medicine, they observed, do not diagnose a disease as an entity. They read the 勢 of the body's configuration, its tendencies and imbalances, and intervene to redirect them.

Crucially, Law and Lin emphasise that reading 勢 is itself a practice within the situation. The practitioner is part of the configuration, and her intervention changes it. They also emphasise the postcolonial stakes. The categories through which a situation is read are not neutral. Reading Chinese practices through Western categories makes them look deficient, and reading them through Chinese categories makes visible what Western categories miss. This is Bourdieu's symbolic power (Thesis V) at the level of civilisations. Whoever supplies the categories shapes what is seen.

Law and Lin thus offer anthro-complexity a 勢 that is reflexive: the reader is in the configuration, and the categories of reading are themselves forces in it. It is a richer 勢 than the surfer's, and closer to Sunzi's.

The Calligrapher's 勢

There is a third domain of 勢 in classical Chinese culture that offers an image of the practitioner inside the configuration, and it may be the best of all for anthro-complexity's purposes: calligraphy.

In Chinese aesthetics, 勢 names the dynamism of the brushstroke: the force that runs through a stroke from its entry to its exit, the tension between strokes, and the movement that animates a character and a whole sheet. Jullien devotes part of The Propensity of Things to it. The calligrapher does not draw a character as a shape to be filled in. She performs it, as a sequence of movements whose 勢 carries from stroke to stroke. The 勢 is not in the finished character, which is only its trace. It is in the performance, and it can be read in the trace by someone who knows how to see the movement in the ink.

Three features of the calligrapher's 勢 bear on anthro-complexity.

The calligrapher is inside the 勢. The force that runs through the stroke runs through the calligrapher's arm, hand and breath. She does not read the 勢 of a brushstroke she is not making. She makes it, and her making is the 勢. There is no position outside.

The 勢 is responsive to the material. The brush, ink and paper have their own propensities: the brush's spring, the ink's flow, the paper's absorbency. The calligrapher's 勢 answers to them, and theirs to hers. This is Ingold's correspondence (Chapter 30) and the grain of 理 (Chapter 36).

The 勢 cannot be planned in detail. A calligrapher may know what character she will write, but she cannot plan every inflection of every stroke, because the stroke's 勢 arises in the performance, from the interaction of intention, hand and material. The finished character is a retrospective coherence (Element 8), legible as a whole only after it has been made.

The calligrapher is the image of the practitioner that 勢, taken whole, recommends. She is not the surfer reading a wave she does not make, nor the sailor waiting for a tide that turns without her. She is a maker inside the dynamism she makes, responsive to the material and unable to plan the detail, whose work is legible only in its trace. A practice of 勢 in human organisations would be a practice like hers. The practitioner's interventions are strokes whose force runs through her and through the organisation, answer to the organisation's own propensities, and can be read only afterwards in the patterns they leave.

勢 and the Theses

Thesis What 勢, taken whole, contributes
II: asymmetry applied to itself Sunzi's general is read; deception presupposes the strategist inside the configuration of mutual reading
III: humans are not estuaries The estuarine map is 地形 (terrain), not 勢; the surfer practises terrain, the general practises 勢
V: power The political 勢 of Shen Dao and Han Feizi is positional power, absent from anthro-complexity's use
Ethical reason (Chapter 16) 擇人而任勢: the commander relies on configuration, not on his men's will, which is the shadow principle with its ambiguity
Thesis VIII: degradation 虛實: the configuration as the other side reads it; the reading is a force, and it changes

What Anthro-Complexity Has Grasped

This chapter should not end without acknowledging what anthro-complexity has grasped of 勢, which is a great deal. It has grasped that efficacy in complex situations comes from working with propensity, not from imposing plans. It has grasped that the practitioner's art is to intervene upstream, where small moves have large effects, before the configuration hardens. It has grasped that the configuration's potential cannot be commanded, only joined, and that the wise practitioner reads before acting. It has grasped the difference between 形 and 勢, between the countable arrangement and the felt potential, and it has made the potential, not the arrangement, the object of practice. These are the core insights of Jullien's reading of Sunzi, and they have done more good in organisational life than almost any other import from Chinese thought.

What it has not grasped, on this chapter's reading, are three things that complete the concept: that the strategist is in the configuration and read by it (Sunzi's deception, empty and full); that 勢 includes position, the political 勢 of the seat (Shen Dao, Han Feizi); and that the practitioner's art is the calligrapher's, a performance within the dynamism, not the surfer's reading of a medium outside her. Each completion is drawn from the same tradition from which anthro-complexity took 勢. None requires a foreign import, and together they restore to 勢 the reflexivity and the power that anthro-complexity's own human distinction requires.

The adversary will say that this chapter is a quarrel about the interpretation of an ancient text, and that anthro-complexity's use of 勢 need not be faithful to Sunzi, Shen Dao or the calligraphers. Jullien's reading is itself a Western philosopher's construction, contested by sinologists. Anthro-complexity takes from it a concept, propensity, that is useful for practice, and the concept's usefulness does not depend on its fidelity to classical sources.

The adversary is right that fidelity is not the only test, and right that Jullien's reading is contested. But the chapter's argument does not rest on fidelity alone. It rests on the observation that the features of 勢 anthro-complexity has dropped (the strategist's inclusion, mutual reading, positional power) are exactly the features its own human distinction requires. Faithful or not, a 勢 without them is a concept of non-human propensity, a landscape. Anthro-complexity's theory says human systems are not landscapes. The classical sources happen to supply what the theory needs, which is a reason to use them, not a mere scholarly scruple. partial

地形 IS THE GROUND —

勢 IS WHAT THE GENERAL MAKES ON IT, WHILE THE ENEMY WATCHES —

AND THE CALLIGRAPHER'S ARM IS INSIDE THE STROKE.

Addition: Cross-Lights From Book Nine

A case of the third sense. Snowden's post on Luhmann gives as an example of a decision premise the unwritten rule that nothing is decided until a named senior person has signed. He proposes to treat it as an actant and to ask how much energy and time it would take to change. That is 勢 in the political sense, the power of the seat, and its cost differs for the one who sits there and for everyone else.

The criterion of the gesture. A Web of Relations takes from these sources a test for a practitioner's act: it must lean into a propensity the field already has, and must not install one. See Chapter 51.

Chapter 36

理 Lǐ: The Grain Is Known by the Knife

4,374 words, about 19 minutes of reading

The Cook and the Ox

The third chapter of the Zhuangzi, "The Secret of Caring for Life" (養生主), contains the most famous story in Chinese thought about skill. Cook Ding (庖丁) is butchering an ox for Lord Wenhui. His hand, shoulder, foot and knee all move together, and the blade sings through the carcass in time, so that the whole work looks like a dance to old court music. The lord marvels at such skill. The cook sets down his blade and answers that what he values is the Way (道), which lies beyond skill.

In the paraphrase this chapter will use: when the cook began, he saw nothing but the whole ox before him. After three years he no longer saw the ox as a whole. Now he encounters it with spirit, not with sight. The senses rest and spirit goes where it will. He follows the ox's natural articulation, its 天理 (tiānlǐ, the heavenly pattern), slipping the blade into the great gaps and along the great hollows, going with the ox as it is made, so that he never strains against ligament or tendon, still less against bone. An ordinary cook must replace his blade monthly, because he chops. A good cook replaces his yearly, because he slices. Cook Ding has used one blade for nineteen years over thousands of oxen, and its edge is still as fresh as on the day it left the whetstone. The joints have gaps, and the blade's edge is all but without thickness. Something without thickness, entering a gap, has room to spare.

Then comes the passage this chapter will return to. When he reaches a tangled place (族), where things knot together, the cook sees the difficulty and takes warning. He fixes his gaze, slows his hand and moves the blade with the lightest touch, until the knot gives way and the flesh falls apart like earth crumbling. Then he stands with the blade in hand, looks about him, content, cleans it and puts it away. The lord says that in hearing the cook he has learned how to nourish life.

理: The Grain

The word at the centre of the story is 理 (lǐ). In its earliest uses, 理 referred to the veins or markings in jade, the pattern along which jade could be worked, and by extension to the working of jade along its veins. It came to mean pattern, texture, grain, the inherent structure of a thing that determines how it can be divided, shaped or understood. In the Zhuangzi passage, 天理 is the ox's "heavenly pattern", its natural articulation, the joints and hollows and openings along which it comes apart.

Later, in Neo-Confucian thought, especially the philosophy of Zhu Xi (1130–1200), 理 became a central metaphysical concept. Principle is the rational order inherent in all things, which the scholar comes to know through the investigation of things (格物 géwù). But the earlier sense, grain and the pattern along which a skilled hand works, is the one The Compass and the Loom drew on, and the one this chapter develops. It is the sense in which 理 is known not by observation from outside but by the knife.

理 Is Known by the Knife

Three features of Cook Ding's knowledge of 理 bear on anthro-complexity.

It is embodied

Cook Ding's knowledge of the ox's grain is not a representation in his head. It is in his hand, his shoulder, his foot, his knee and the swish of his blade. It was acquired over years of practice, and it shows itself in performance. He could not write it down, and if he did, a novice could not use it. The Compass and the Loom put the point compactly: Cook Ding's grain is known through years of contact, and "pattern detection at scale" is the opposite kind of knowing.

This is Polanyi's tacit knowledge, Scott's mētis and Dreyfus's background, the knowledge anthro-complexity's Element 11 honours. Hubert and Stuart Dreyfus's model of skill acquisition, in Mind over Machine (1986), traced five stages. The novice follows context-free rules. The advanced beginner recognises situational features. The competent performer chooses plans and goals. The proficient performer sees situations holistically. The expert acts intuitively, without deliberation, responding to the situation as it shows up. Cook Ding at three years no longer saw the whole ox, which is proficiency. At nineteen years he meets it with spirit, not eyes, which is expertise. The progression is from rules and representations to embodied, intuitive response. And expertise, on the Dreyfus model, cannot be captured in rules, because it has gone beyond them.

It is responsive

Cook Ding does not impose a cutting pattern on the ox. He follows the ox's pattern. His knife goes where the ox's structure allows, into the hollows and openings. His skill is not mastery over the ox but correspondence with it, in Ingold's sense (Chapter 30). The knife answers the grain, and the grain guides the knife. That is why the knife stays sharp. It never fights the ox's structure, so it never meets resistance that would dull it.

This is anthro-complexity's work with the propensity, not against it, applied to material. But note the difference from the surfer. The surfer reads the wave from the board. Cook Ding's knife is in the ox. His knowledge of the grain is knowledge from inside, acquired by contact and shown in the cutting. The cook does not read 理 and then cut. He knows 理 by cutting.

It slows at the tangled place

The passage that matters most for this disputation is the tangled place, 族 (zú), where tendons and bones cluster and the grain is hard to read. There Cook Ding does not proceed by rote, nor does he force his way through. He slows down. He sees the difficulty, takes warning, fixes his gaze, slows his hand and lightens his touch. Then the knot comes apart.

The tangled place is where complexity is greatest. In Cynefin's terms it is the complex domain within the ordered: the place where the cook's practised skill, adequate for the open joints, meets a configuration it cannot simply follow. His response is probe-sense-respond at the scale of the knife. He makes small, careful moves, watches their effect and adjusts. But his response is also attention. He fixes his gaze on the work. At the tangled place, the expert who usually meets the ox with spirit and not eyes returns to close watching. Expertise does not dispense with attention. It concentrates it where it is needed.

The tangled place is also, in Ingold's vocabulary, a knot, a place where lines tangle. Cook Ding's practice at the knot is the wayfarer's at a difficult crossing: slow down, attend and move with the lightest touch. It is not the navigator's: consult the chart and plot a route through.

Wang Yangming and the Bamboo

The Neo-Confucian career of 理 contains an episode that bears directly on the question of how 理 is known, and it is one of the most famous stories in Chinese intellectual history.

Zhu Xi taught that 理 is present in all things, and that the scholar comes to know it through 格物, the investigation of things, studying things one by one until their principle becomes clear and, in time, the principle of all things is grasped. Wang Yangming (1472–1529), as a young man devoted to Zhu Xi's teaching, resolved to put it into practice. He and a friend sat before a stand of bamboo in a courtyard and set themselves to investigate its principle. The friend gave up after three days, exhausted and ill. Wang persisted for seven days and then he too fell ill, having grasped nothing. He later said that the experience convinced him that Zhu Xi's method was mistaken. The principle of things could not be found by staring at them from outside.

Wang's mature philosophy, developed after a period of exile and illumination, held that 理 is not out there in things to be investigated by a detached observer. It is in the heart-mind (心 xīn) as it engages with things. His famous doctrine of the unity of knowing and acting (知行合一 zhī xíng hé yī) held that true knowledge is inseparable from action. To know filial piety is to be filial, and someone who claims to know it but does not act on it does not truly know it. Knowledge is not a representation acquired by observation and then applied. It is shown in action and inseparable from the engaged life of the heart-mind.

The bamboo story is the Chinese tradition's own critique of the outside observer. Zhu Xi's investigation of things, as Wang tried to practise it, is observation from outside: the scholar sits before the bamboo and looks. It fails because 理 is not available to such looking. Wang's alternative is knowledge in engagement, in acting, the knowledge of Cook Ding's knife. The parallel with this disputation's argument is exact. The estuarine map, the dashboard and the pattern read from outside are Zhu Xi's bamboo. They are investigations of things by an observer who sits before them and looks. The relational, participatory and correspondence-based knowledge this disputation has argued for is Wang Yangming's unity of knowing and acting: knowledge of the organisation's grain is acquired by working within it and shown in what one does.

理 in Organisations

What would it mean to know the 理 of an organisation, its grain, the pattern along which it can be worked?

The grain is known by those who work it

The people who know an organisation's grain best are those who work it daily: the frontline staff who know where the procedures break down, the middle managers who know which approvals can be expedited and which will stall, the long-serving employees who know the history behind the current arrangements, the informal networks that route around the formal structure. Their knowledge is Cook Ding's. It is embodied, acquired through years of contact and shown in what they do. It is also the knowledge that narrative instruments seek to gather and that the computational practice (Thesis VI) makes legible and so strips of its embodied character.

The grain is not where the map says

The formal organisation chart, the process map and the strategy deck represent an organisation's structure as its designers intended it. The grain often runs elsewhere. The actual pattern along which work flows, decisions are made and change can happen follows informal networks, tacit agreements and historical accidents that no chart shows. A consultant who works from the chart will hack, in Cook Ding's terms. She will cut across the grain, meet resistance and dull her knife. A consultant who works from the grain will cut, finding the hollows and openings where change can pass without resistance. But she can find the grain only by working, by contact, slowly, attending to what she meets.

The tangled place needs attention, not instruments

Every organisation has its tangled places: the knots where interests, histories and constraints tangle and the grain is hard to read. A merger's integration, a contested reorganisation, a long-standing conflict between functions. At the tangled place, Cook Ding's lesson is to slow down, attend and move with the lightest touch. An instrument that sweeps across the whole organisation, gathering narratives at scale, will register the tangled place as a cluster of contested stories, a pattern among patterns. It will not provide the slow, close, attentive engagement the knot requires. Instruments help with the open joints. At the knot, the practitioner must put down the dashboard and pick up the knife: attend, move slowly, fix her gaze on the work.

The knife that stays sharp

Cook Ding's knife stays sharp for nineteen years because it never fights the grain. A practitioner who works with an organisation's grain finds that her interventions do not meet resistance that wears them down. She is not imposing change against the organisation's structure but releasing change along it. A practitioner who works against the grain finds that her interventions dull quickly. Change programmes are resisted and initiatives fade, so she must change her knife every month, launching new programmes to replace the ones that failed. The image is anthro-complexity's own critique of change management. Lewin's three steps, imposed against the grain, are the mediocre cook's hacking.

理 and Constitutive Constraint

The grain of an organisation is, in Juarrero's terms (Chapter 29), its pattern of constitutive constraints: the joints and hollows of its identity, the arrangements without which it would not be this organisation. In Bourdieu's terms (Chapter 33) the grain is carried in the habitus of its members as a practical sense of how things go. In Giddens's terms it is in their practical consciousness, known but not said.

This places 理 at the heart of Thesis IV. The constitutive constraints that estuarine mapping does not see are the grain, and the grain is known by the knife: by those who work it, through contact, shown in practice and not available to a map drawn from discussion. A practice that respected 理 would seek the grain where it is known, in the practical knowledge of those who work the organisation, and it would know that this knowledge cannot be fully extracted into a map or a dataset. It can be supported, consulted and honoured, but not harvested without loss.

理 and the Instruments

What would narrative instruments look like if they were designed to honour 理, the grain known by the knife?

They would ask the knife-holders where the grain runs. Not "tell us a story about your experience" but "where does work actually flow, and where does it stall? where can change pass easily, and where does it meet resistance? what do you know about how things really work here that the chart does not show?" These are questions addressed to practical consciousness, asked of those who hold the knife.

They would accept that the answer is partial. The knife-holder's knowledge is embodied, and what she can say is a fraction of what she knows. The instrument would treat her answers as traces of a deeper knowledge, not as the knowledge itself, and would seek to support that knowledge, for instance by giving her more room to act on what she knows, rather than only to collect it.

They would slow at the tangled places. Where the patterns show a knot (contested stories, conflicting significations, persistent clusters of difficulty), the instrument would not try to resolve the knot statistically. It would flag it for slow, close, attentive engagement by people who can work at the knot with the lightest touch: facilitated conversation, observation, the anecdote circle of Snowden's Stage One.

They would not harvest the grain. Embodied knowledge of the grain, once extracted into a dataset and used to train a model, loses the character that made it knowledge. The instrument would protect it from the harvest (Thesis VI), keep it with those who hold it, and route decisions that depend on it through their judgement, at the grating, not around it.

They would know that the instrument is also a knife. Every intervention, including the instrument's own deployment, cuts the organisation along some line. An instrument deployed with the grain releases knowledge. One deployed against it, imposed on a population that does not trust it, timed badly or framed in categories that cut across the organisation's own articulation, meets resistance and dulls. The designer of instruments must know the grain as well as the cook does, or her instrument will hack.

Cook Ding and the Theses

Thesis What 理 contributes
I: possessive fallacy The cook's knowledge is not a possession but a correspondence with the ox; skill is relational
III: humans are not estuaries The cook's knife is in the ox; knowledge from inside, not from the deck
IV: constitutive constraint The grain is the pattern of constitutive constraints, known by working them
VI: computation Pattern detection at scale is the opposite of knowing the grain; the grain cannot be harvested without loss
Wang Yangming The investigation of bamboo from outside fails; knowing and acting are one

What Anthro-Complexity Has Lost

The Compass and the Loom's verdict was that the Cynefin Company's instruments grasp 勢 and lose 理 and 道. This chapter can say precisely what is lost when 理 is lost.

Anthro-complexity's theory honours 理. Element 11 adds mētis and phronesis to the definition of a complex system, and its whole critique of codification, from Stage One onwards, is a defence of knowledge that cannot be extracted. Snowden's distinction between tacit, explicit and entangled knowledge is a theory of 理. Entangled knowledge is the grain, known only in the interaction of people, artefacts and context.

What is lost is 理 in the instruments. The instruments are designed to read patterns across populations, not to support the knife-holder's knowledge of the grain. They seek patterns at scale, which is the opposite of the cook's slow attention at the knot. They make the grain legible, which is Scott's simplification, and so risk Prussian forestry's fate. They treat the population as a source of signal, not as the holders of the knife. Anthro-complexity's theory knows 理, and its instruments, in the drift traced in Chapter 6, have moved from the knife to the dashboard.

The remedy is not to abandon the dashboard. Patterns at scale are real and worth knowing, and Cook Ding's lord could not carve every ox himself. The remedy is to restore the knife to its place. The dashboard serves the knife-holders, not the other way round. The tangled places are worked by hand, and the grain is honoured as knowledge that lives in those who work it.

The adversary will say that Cook Ding is an image of individual skill, and that organisations are not oxen carved by single cooks. An organisation has thousands of knife-holders, each knowing a small part of the grain, and no one knows the whole. The dashboard exists because no single cook can carve an organisation. Aggregating the knife-holders' partial knowledge is the only way to see the whole grain. To romanticise the individual craftsman is to abandon the problem of scale that anthro-complexity's instruments were built to solve.

The adversary wins half. An organisation is not one ox, and the knife-holders' knowledge is partial and distributed. But the problem of scale is Hutchins's problem (Chapter 31), and his answer was not to aggregate partial knowledge at a centre but to arrange the knife-holders so that their horizons overlap, and each can see enough of what others see to work the grain together. Cook Ding's lesson at scale is not that one cook should carve everything. It is that the carving should be done by those who know the grain, arranged so that they can work together, with the dashboard as a shared chart on the bridge, not a centre's view of the carcass. partial

THE GOOD COOK CUTS — THE MEDIOCRE COOK HACKS —

AT THE COMPLICATED PLACE, SLOW DOWN AND WATCH —

THE GRAIN IS KNOWN BY THE KNIFE.

Addition: The Grain of a Loan Book

Chapter 46 tested the disputation on a receivership. This addition applies the lesson of 理, the grain known by the knife, to a neighbouring field: the management of a high-stakes lending portfolio by a lender. The field is one in which complexity thinking is beginning to be offered as a practical discipline, as tactical problem-solving for credit teams facing portfolios under stress. The lesson of Cook Ding bears on how that offer should be made.

The map and the grain

A lender manages its portfolio through a map. Every loan is recorded with its balance, interest rate, security, loan-to-value ratio, covenants, payment history and risk grade. Portfolios are aggregated into dashboards that show concentrations of risk, arrears trends, sectoral exposures and the distribution of risk grades. Credit committees read the dashboards, set policy and make decisions on individual cases that cross thresholds. The map is essential. No lender could manage thousands of loans without it, and regulators require it.

But the map is not the grain. The grain of a loan book is what the relationship managers, the workout officers and the receivers know: which borrowers will pay if given time and which will not; which security is worth what the valuation says and which is not; which covenant breaches signal real distress and which are technical; which family companies have hidden resources in other properties and which have hidden liabilities; which local markets are turning before the indices show it. This knowledge is Cook Ding's. It is acquired over years of contact, through visits, conversations, site inspections and the accumulated experience of cases that went well and badly. It is shown in judgement, not in the dashboard. It is the mētis of credit.

Where the hacking happens

The lesson of Cook Ding is that a cook who cuts across the grain hacks, and must change his knife every month. A lender that manages its book by the map alone, against the grain, hacks in characteristic ways.

Thresholds trigger the wrong cases. A dashboard flags every loan whose loan-to-value crosses a threshold. Some of those loans are fine: the borrower is strong and the valuation is temporarily depressed. Others below the threshold are in serious trouble that the map does not show. Workout resources go to the flagged cases, the unflagged trouble grows, and the knife dulls.

Policies imposed across the book meet resistance. A policy to accelerate enforcement on all loans in arrears for more than a set period is applied uniformly. It catches borrowers who would have recovered with forbearance, damages relationships that would have produced repayment, and sometimes produces worse recoveries than patience would have. The cut goes across the grain.

The relationship manager's knowledge is overridden. The relationship manager knows that a borrower in technical breach is sound, but the dashboard grades the loan as high risk, the credit committee acts on the grade, and the relationship manager's judgement is recorded as "noted". The knife-holder has been overruled by the map.

Slowing at the knots

Cook Ding slowed at the tangled place, where tendons and bones clustered and the grain was hard to read. A loan book has its tangled places: the cases where borrower, security, market, other lenders, tenants and legal position knot together, and the right course cannot be read off the map. A receivership like Chapter 46's is one. A borrower with several properties charged to several lenders, each lender pursuing its own enforcement, is another. A development loan whose project has stalled, with contractors, planning, buyers and funders entangled, is a third.

At the tangled place, the lesson is to slow down, fix the gaze and move with the lightest touch. In credit terms, this means:

  • Do not apply the policy by rote. At the knot, the policy's thresholds are a starting point, not a decision.
  • Bring the knife-holders in. The relationship manager, the workout officer and the receiver, those who have been in the building and met the borrower, know the grain. Their judgement is consulted, not noted.
  • Small moves first. A short forbearance with conditions, a site visit, a conversation with the other lenders, a request for the borrower's full position. These are probes, in Cynefin's terms, and their responses reveal the grain.
  • Watch the response. Does the borrower engage, provide information and keep the conditions? Do the other lenders cooperate or race? Does the market move? The response tells the cook where the knife can pass.

The dashboard serves the knife

The lesson is not to abandon the dashboard. A lender needs the map as Cook Ding's lord needed his kitchen run. The lesson is that the dashboard should serve the knife-holders, not the other way round.

Map-led credit (hacking) Grain-led credit (cutting)
Thresholds trigger action Thresholds trigger attention; action follows judgement
Policy applied uniformly Policy as warp; cases woven by those who know them
Relationship manager's view noted Relationship manager's view weighed, with reasons recorded if overridden
Credit committee reads the dashboard Credit committee reads the dashboard and hears the knife-holders
Tangled cases processed by the standard route Tangled cases flagged for slow, close, attentive handling
Success measured by speed of resolution Success measured by recovery, relationship and whether the knife stays sharp

Complexity as credit tactics, offered rightly

When complexity thinking is offered to lenders as tactical problem-solving, the lesson of 理 suggests how the offer should be made. It should not be offered as a new map, a better dashboard or a framework that classifies the book's cases by Cynefin domain and prescribes responses. That would be another chart for the navigator, another investigation of bamboo from outside (Wang Yangming, Chapter 36). It should be offered as a discipline that restores the knife-holders' knowledge to its place: a way of recognising the tangled places, slowing at them, bringing in those who know the grain, probing before cutting, and letting the dashboard serve the judgement it cannot replace.

There is also a lesson from this book's other chapters. The lender is not outside its book. Its policies are read by borrowers, who adjust (Thesis VIII). Its enforcement actions are gestures that call out responses in borrowers, other lenders and markets (Chapter 27). Its credit committee's waiting is a gesture too (Chapter 21). A lender that read its book as a reflexive system, one that reads the lender as the lender reads it, would recover more, lose less and keep its knife sharp.

THE DASHBOARD IS THE MAP — THE WORKOUT OFFICER HOLDS THE KNIFE —

AT THE TANGLED CASE, SLOW DOWN AND BRING IN THOSE WHO KNOW THE GRAIN —

THE LENDER IS READ BY ITS BOOK.

Addition: Cross-Lights From Book Nine

The cook and the knot. Snowden's knot and Cook Ding's ox are the same teaching. A tangle is long to describe from outside and short for one who understands it, and the understanding was got by handling. The cook says that he no longer sees the ox whole. Chapter 56 calls this matching complexity, as against abstracted simplification, and uses it against the doctrine that complex things cannot be compressed.

The garden. The dry garden at Ryōan-ji, in Chapter 53, is a companion image. No seat on its veranda shows every stone.

Chapter 37

道 Dào: The Way Is Made by Walking

3,569 words, about 16 minutes of reading

The Way Is Made by Walking

In the second chapter of the Zhuangzi, the "Discussion on Making All Things Equal" (齊物論), there is a short phrase that has become a touchstone for every relational reading of Chinese thought: 道行之而成, the Way is made by walking it. The context is Zhuangzi's critique of fixed distinctions. Things are called "so" because they are called so. Paths are made by people walking on them. What is "this" and what is "that" depend on where one stands. The 道 is not a road laid down in advance, along which one travels. It is the trace of travelling, made in the going.

A Spanish poet arrived at the same image twenty-four centuries later. In Antonio Machado's Campos de Castilla (1912) there is a famous line: caminante, no hay camino, se hace camino al andar. Walker, there is no road. The road is made by walking. The convergence is not influence. It is two traditions, from opposite ends of the world, finding the same truth about how ways come to exist.

This chapter, the third of Book Six, takes 道 as the term that The Compass and the Loom found the Cynefin Company's instruments had lost along with 理. A compass bearing is a way laid down in advance, and that treatise contrasted a path co-made by participants with a course set before the walking. The chapter develops the contrast and then examines two Daoist concepts that bear directly on anthro-complexity: 無為 (wúwéi), action that does not force, which The Compass and the Loom proposed as a third option between natural and stimulated emergence, and 我自然 (wǒ zìrán), "we did it ourselves", which it proposed as the test of recognition in place of alignment. It ends with 經營 (jīngyíng), warp and weft, and the Tidal Loom.

The Compass and the Way

A compass bearing is a direction specified in advance. It is relative to an external field, magnetic north, which is independent of the traveller. The traveller takes a bearing, sets her course and follows it. If the terrain forces her off course, she corrects back to the bearing. The bearing is the way, and the walking serves it.

SwarmCompass, The Compass and the Loom argued, is a compass in this sense, and its name says so. It promises to give organisations "adrift without direction" a direction. That direction is discovered through distributed sensing, but once discovered it is a bearing, a course set before the walking, towards which the organisation is to be aligned. The instrument's governing metaphor is navigation by bearing, the navigator's art in Ingold's sense (Chapter 30).

The 道 of the Zhuangzi is the opposite. It is not specified in advance or relative to an external field, and it is not followed. It is made, by the walking of those who walk it, and it exists only as the trace of their walking. A 道 that has been walked can be seen as a path afterwards, and others can walk it, which is how paths become roads. But the path was not there before it was walked, and it was not walked towards a bearing. It was walked through the country, responsive to the ground.

This is retrospective coherence (Element 8) in Daoist form. The way is coherent in retrospect, as a path that can be seen once it has been made. It could not have been specified in advance, because it did not yet exist. A practice that sets bearings in advance has mistaken the retrospective coherence of 道 for a prospective course.

The way is made by many walkers

The Zhuangzi's phrase has a further implication. Paths are made by people walking, plural. A single traveller crossing a field leaves a trace that the grass soon covers. A path appears when many walk the same way, each following the traces of those before, each deepening the trace for those after. A path is a collective product, a correspondence (Chapter 30) among walkers over time, none of whom designed it.

That is the 道 of an organisation. Its way of going, its practices, routines and culture, is a path made by many walkers over time, each following and deepening the traces of others. No one designed it, and no compass set it. It exists as the trace of collective walking, and it changes when enough walkers begin to walk differently. A practice of 道 in organisations would attend to the paths that exist, the ways people actually go, and to the places where new paths are beginning to form, where a few walkers are leaving the old way and others are starting to follow. It would not set a bearing and align the walkers to it. It would help the walkers see the paths they are making.

無為: Action That Does Not Force

Wúwéi (無為), literally "non-action" or "not-doing", is the central practical concept of the Laozi (the Daodejing) and is important throughout early Chinese thought, Confucian as well as Daoist. It does not mean inaction. It means action that does not force, action so fitted to the situation that it does not register as imposition. The Laozi says that the Way does nothing, and yet nothing is left undone (道常無為而無不為). The sage acts without acting, and things are accomplished.

Edward Slingerland's Effortless Action (2003) and Trying Not to Try (2014) gave the most developed modern account of wúwéi. Slingerland reads it as a state of spontaneous, skilful responsiveness, in which action flows from a cultivated disposition without deliberate effort, as Cook Ding's knife flows through the ox. Wúwéi is the expert's state in the Dreyfus model (Chapter 36), the state in which one acts rightly without trying. Slingerland identified the paradox of wúwéi. How can one try to reach a state of not-trying? If wúwéi is effortless, effort to attain it seems self-defeating. If it cannot be attained by effort, it is unclear how it can be attained at all. Slingerland traced how different early Chinese thinkers responded to the paradox, through ritual training (Confucius), cultivation of innate sprouts (Mencius), return to simplicity (Laozi) and the forgetting of distinctions (Zhuangzi).

無為 as the third option

The Compass and the Loom proposed wúwéi as a third option between the instruments' "natural" and "stimulated" emergence. WRAS's axis of natural versus stimulated emergence presents the practitioner with a choice: leave the system alone and let emergence happen, or intervene and stimulate it. The axis presupposes, the treatise argued, a chooser who can leave a social process alone. Thesis III's estuary and Mead's gesture both show that no such chooser exists in human systems. Leaving alone is a gesture, waiting is an action, and the chooser is in the system.

Wúwéi dissolves the dichotomy. It is neither leaving alone nor stimulating. It is action fitted so closely to the situation that it does not register as forcing: the cook's knife in the gap, the calligrapher's brush following the grain of the paper, the sage's governance that the people barely notice. Wúwéi acknowledges that the practitioner is always acting, since there is no not-acting in a human system, and asks only that her action be fitted, not forced. It is the practice of the general inside the 勢, not the surfer outside it.

Weak measurement as 無為

The Uncertain Compass of The Compass and the Loom found a physical analogue for wúwéi in weak measurement. In quantum mechanics a weak measurement is one that interacts with a system so gently that it barely disturbs it, extracting a little information without collapsing the system's state. Many weak measurements, aggregated, can reveal information that a single strong measurement would destroy. The treatise's social analogue: the leader who measures weakly learns without freezing. A practice of sensing that is wúwéi would sense lightly, in ways that inform without forcing, that do not collapse the population's narration into the instrument's categories, and that do not freeze change by over-monitoring (Thesis VIII's Zeno law). It would be the sensing of a sage who governs so lightly that the people barely know she is there.

The Best Ruler: Laozi 17

The seventeenth chapter of the Laozi gives a ranking of rulers that bears directly on anthro-complexity's question of how a practitioner should stand to the people she serves. In a common paraphrase:

The best ruler is one whose existence the people barely know. Next comes one they love and praise. Next, one they fear. Next, one they despise. When trust is insufficient, there is no trust. Hesitant, the best ruler values her words. When the work is done and the task accomplished, the people all say: we did it ourselves (我自然).

The ranking reverses the usual order of leadership virtues. The ruler who is loved and praised is second-best, because love and praise mean the people are attending to the ruler and not to their own lives. The best ruler is barely noticed. Her governance is so fitted to the people's own propensities that they experience its results as their own achievement. They say we did it ourselves, and they are right, because the ruler's action was wúwéi. It released what was already in them and did not impose what was not.

我自然: recognition, not alignment

The Compass and the Loom took wǒ zìrán as the final principle of the Tidal Loom: recognition, not alignment. The test of a practice is whether people say "we did this ourselves". The principle draws also on Kevin Flinn's work on recognition in leadership (Chapter 11), which treats leadership as a pattern of recognition emerging in interaction, not a property of a leader.

The contrast with alignment is sharp. An aligned organisation is one whose members are oriented towards a direction set by others, a bearing. Alignment can be measured: how far do members' beliefs, priorities and actions match the direction? It is the compass's measure. Recognition cannot be measured in the same way. It is a judgement made by members about the outcome. Do we recognise this as ours? Did we do it? Recognition is the measure of 道. It asks whether the path was made by the walkers, not whether they followed the bearing.

The Compass and the Loom found that SwarmCompass's "authentic alignment" promise describes compliance more than recognition. An organisation aligned through distributed sensing, however authentic the sensing, is aligned to a direction its members did not make. The sensing discovered the direction, and leadership set it as a bearing. The members' narratives were the raw material. The direction was the product. When the work is done, the members do not say "we did it ourselves". They say "we were sensed, and then aligned".

Recognition as a test for anthro-complexity's instruments

Laozi 17 offers anthro-complexity a test for every instrument and intervention: will the people say "we did it ourselves"?

Practice Likely outcome Laozi's rank
Consultant diagnoses, prescribes, implements "They did it to us" Feared or despised
Leadership senses, then aligns the organisation "They listened, then told us" Loved and praised, at best
Patterns returned; population decides what to do "We saw ourselves, and acted" Approaching the best
Population makes the instrument, reads the patterns, acts; convenor barely noticed "We did it ourselves" The best

The test is demanding, and it is not always the right test. Some situations need a ruler who is feared, such as crises in which decisive action must be taken quickly. But it is the right test for anthro-complexity, because anthro-complexity's Pelagian stance (Element 20) says the system has the capacity to heal itself, and that the practitioner's role is to help it use that capacity. A Pelagian practice succeeds when the system recognises its healing as its own. Laozi 17 is the Pelagian test stated as a ranking of rulers.

經營: Warp and Weft

The Compass and the Loom's reconstruction, the Tidal Loom, was built on a Chinese compound: 經營 (jīngyíng), which in modern Chinese means "to manage" or "to operate" a business, and whose characters mean warp (經) and weft-work, or to plan and build (營). The 經 is the warp, the long threads strung on the loom before weaving begins, which give the cloth its length and structure. In classical usage it also means a classic, a canonical text, the warp of a tradition. The 營 is the work of building, planning and constructing, and in the treatise's reading the weft that is woven through the warp.

The treatise's architecture follows the image. Convenors lay a revisable warp: themes, known constraints, dated limits, obligations, probes, a declination table and a return commitment. Everyone weaves, sponsors and consultants included. There is no position outside the cloth. The warp gives the weaving structure, as a governing constraint gives coherence (Juarrero). It does not determine the pattern, which is made by the weft, by everyone's weaving. And the convenors who lay the warp also weave. They are not outside the cloth, setting its pattern. They are threads in it.

The Spirit Tower

The treatise also cites a passage from Mencius that illuminates 經營. Mencius, asked by King Hui of Liang whether worthy men also enjoy pleasures like the king's parks and ponds, recalled the ode about King Wen's building of the Spirit Tower (靈臺). The ode says that King Wen planned and began the Spirit Tower (經始靈臺, the 經 of 經營), and that the people came to work on it like children coming to help their father, so that it was finished in less than a day. King Wen did not drive them. They came of their own accord. Mencius's reading is that the people came because they shared the king's joy. The tower was theirs as much as his, and its parks and ponds were places they too could enjoy. A king who shares his pleasures with the people will find them eager to build. A king who hoards them will find them resentful, and his towers will fall.

The Spirit Tower is 我自然 in architecture. King Wen laid the warp, planning and beginning the tower. The people wove the weft, coming of their own accord to build. When the tower was done they could say we built it, and they would be right. The king's role was to begin, to share and to make the work worth joining, not to command. The Compass and the Loom summarised Mencius's point: people come when they share the king's joy.

道 and the Instruments

What would anthro-complexity's instruments look like if they honoured 道, the way made by walking?

They would map paths, not bearings. Instead of sensing a population in order to discover a direction to align it to, they would sense the paths the population is already making, the ways people actually go and the new ways some are beginning to walk. The question would be not "where should we go?" but "where are we going, and where are new paths forming?"

They would show the walkers their paths. Patterns would be returned to the population as maps of their own walking, so that they could see the paths they are making and decide, together, which to deepen and which to let grow over. This is Hutchins's shared chart (Chapter 31) and the Tidal Loom's return as design requirement.

They would sense lightly. Sensing would be wúwéi: weak measurement, monitoring less often than the change hoped for, leaving pockets unmonitored where new paths can form without being frozen by attention. The Pyxis's laws apply.

They would lay warp, not set course. Convenors would lay the warp of a sensing exercise (the themes, the known constraints, the declination table), and the population would weave the weft: the stories, the readings and the decisions. The convenors would weave too, signifying their own experience alongside everyone else's.

They would be tested by recognition. The success of an instrument would be judged by whether, when the work was done, the population said "we did it ourselves". Not by alignment scores, not by leadership's satisfaction, and not by the clarity of the patterns.

道理勢 Together

Book Six has read anthro-complexity's human cluster and instruments through the three terms of the spine. The results can be gathered in a single table.

Term Classical source What it is What anthro-complexity has grasped What it has lost What restoring it would require
勢 shì Sunzi; Jullien; Shen Dao; calligraphy The propensity of a configuration, including the strategist, read by others; positional power Propensity over causality; work upstream; join, do not command The strategist inside the configuration; mutual reading; political 勢 (position); the calligrapher's performance Include the practitioner in the configuration; map position and power; practise as the general, not the surfer
理 lǐ Zhuangzi (Cook Ding); Wang Yangming The grain, known by the knife; knowing and acting are one Tacit knowledge; mētis; entangled knowledge; the critique of codification The knife in the instruments; slow attention at the knot; Wang against the bamboo Ask the knife-holders; slow at the knots; do not harvest the grain; the dashboard serves the knife
道 dào Zhuangzi; Laozi; Mencius The way made by walking; wúwéi; wǒ zìrán; 經營 Retrospective coherence; Pelagian stance; probe, sense, respond The way as made by many walkers; wúwéi as third option; recognition as test Map paths, not bearings; lay warp, not course; sense lightly; test by recognition

The table shows the verdict of The Compass and the Loom to be right as far as it goes, and in need of one refinement. The instruments grasp 勢 and lose 理 and 道, but they grasp 勢 only partially. They have the cosmological sense of propensity without the strategic sense of mutual reading or the political sense of position. A 勢 without the strategist is terrain (地形), and terrain without grain or way is a landscape seen from above. The three terms, taken together, describe a practice from inside: the general within the configuration, the cook within the ox, the walker on the way. The practice the triad describes is the practice of the fisherman in the maelstrom (Elias), the wayfarer (Ingold), the participant in the conversation of gestures (Mead), the member of the community of inquiry (Peirce) and the navigator on Hutchins's bridge. Book Six has found in Chinese thought the same convergence Book Five found in Western thought, and it is the same point. The human is relational, the observer is inside, and the way is made by walking.

The adversary will say that wúwéi and wǒ zìrán are ideals of an agrarian monarchy, written for sage-kings governing peasants, and that importing them into contemporary organisations is romantic and possibly sinister. The ruler whom the people barely know is also the ruler who cannot be held to account. A practitioner whose influence goes unnoticed is exercising power without visibility, which is exactly what Thesis V condemned. "We did it ourselves", said by people who were in fact guided by an invisible hand, is false consciousness, Lukes's third dimension of power.

The adversary wins an important half, and this reply is the most serious objection to the Daoist ideal in this book. Wúwéi can be the mask of a power that conceals itself, and wǒ zìrán can be the effect of manipulation so skilful that its objects mistake it for their own will. The ideal is not safe in itself. It becomes safe only when joined to the reflexive disciplines of Book Five and the power-awareness of Thesis V. The practitioner's light touch must be declared (declination before direction), visible (the reading of the reader, mutual dashboards) and contestable (return before action, the community of inquiry). Under those conditions, "we did it ourselves" can be true: the people did it, with help they could see and could have refused. Without them, it is the slogan of an invisible ruler. The Daoist ideal needs the Western critique of power, and the Western critique of power needs the Daoist ideal of action that does not force. Neither is complete alone. partial

THE WAY IS MADE BY WALKING —

THE BEST RULER IS BARELY KNOWN, BUT MUST STILL BE SEEN —

AND WHEN THE WORK IS DONE, THE PEOPLE SAY: WE DID IT OURSELVES.

Addition: The Reliance on Cilliers Withdrawn in Part

This chapter drew on Cilliers for the step from incompressibility to provisionality and from there to an ethics of framing. Chapters 56 and 58 withdraw two of those three steps. Incompressibility is found to be a concept of the outside observer. The ethics of framing, as Woermann and Cilliers construct it, is found to be an ethics of the framer's conscience, which asks for awareness and provides no way for those excluded to answer. Provisionality is kept, as the dating of every limit. The demands this book makes, for declared expectations, return and contest, are better described as a replacement for that ethics than as an application of it.

Addition: Cross-Lights From Book Nine

Wayshaping. Snowden's word for the practitioner's work in the capacity mode is wayshaping: arranging conditions so that a way can form. It is close to the line from the Zhuangzi with which this chapter began. The difference lies in who walks. A way shaped by one party for others to walk is a designed path. See Chapter 48.

Cilliers reconsidered. This chapter relies on Cilliers's argument from incompressibility to an ethics of framing. Chapter 56 challenges the argument and keeps the ethics, on a different ground: a description in human affairs acts, and whoever describes answers for what is made.

Chapter 38

The Dyad Revisited: Yin-Yang, Trigram, Triquetra

3,880 words, about 17 minutes of reading

Snowden's Question

The November post's section on dyads and triads ends with a question asked in good faith:

Snowden says he holds the point loosely, and would like to know whether the Chinese sources have a way of handling mediation that he is missing.

The question deserves an answer, and this chapter offers one. The Chinese sources do have ways of handling mediation, several of them, and some are central to the classical tradition. Their structure bears on the dispute between Ing's dyadicism and Snowden's trialectic in an unexpected way. It shows that both have read the Chinese sources partially. Ing reads them as dyadic, yin and yang, and so misses the mediating third that the classical texts insist on. Snowden reads the trigrams as two to the power of three, a combinatorics of binaries, and so misses that the trigram's three lines are positions, and that the middle position is the place of the human. The Chinese sources' answer to Snowden's question is that mediation is real, irreducible and generative, and that its paradigm case is the human, standing between heaven and earth. That answer, followed through, supports the relational account of the human this disputation has argued for.

The Dispute Restated

Ing's programme is dyadic. Yin and yang are inseparable, inter-transforming modes, qi is concentrating and dissipating, things wax and wane, with no fixed superiority of one over the other. Snowden credits the insistence that dyadic is not dualistic, that the poles are context-dependent and complementary, not ranked, but holds that people will dichotomise any pair, and that only a triad, each pair mediated by the third, keeps the system open. Ing brings threes in through the trigrams of the Yijing: three lines, each yin or yang, giving eight combinations, mapped to eight seasons. Snowden's objection is that this is two raised to the power of three, a combinatorics of binaries, not thirdness in Peirce's sense.

Snowden's objection is correct if the trigram's three lines are merely three binary digits, interchangeable positions each holding a yin or a yang. Then the trigram is a three-bit code, and its eight values are 2³ combinations. The classical commentaries, however, do not read the lines that way.

The Three Powers: 三才

The Xici zhuan (繫辭傳), the "Commentary on the Appended Phrases" that forms part of the Ten Wings of the Yijing, gives the classical account of the trigram's structure. Its three lines are not interchangeable. They are positions, and each position corresponds to one of the three powers, 三才 (sāncái): heaven (天 tiān) at the top, earth (地 dì) at the bottom and humanity (人 rén) in the middle. In the hexagram, formed by doubling the trigram, the six lines are paired into three positions, the top two for heaven, the bottom two for earth and the middle two for humanity. The Xici says that the Yijing contains the Way of heaven, the Way of humanity and the Way of earth. It takes these three powers and doubles them, which is why there are six lines.

On this reading, the trigram's threeness is not combinatorial. It is cosmological and positional. The three lines are three different kinds of place in the order of things, and the middle place, the human, is the place of mediation between heaven and earth. A yin or yang line in the middle position has a different meaning from the same line at the top or bottom, because the position changes what the line signifies. The trigram is not 2³ undifferentiated combinations. It is three positioned powers, each of which may be in a yin or yang mode, and whose meaning depends on its position in relation to the others.

Snowden's objection therefore holds against Ing's mapping of trigrams to seasons, if that mapping treats the lines as interchangeable bits. It does not hold against the trigram as the classical commentaries understood it. In the Xici's reading, the trigram already contains thirdness in something close to Peirce's sense: three positions, irreducibly different, with the middle mediating the other two.

The human as mediator

The middle position of the three powers is the most important feature for this disputation. In the classical Chinese cosmos, humanity's place is between heaven and earth. The human mediates. She receives the patterns of heaven (the seasons, the movements of the stars, the mandate) and the resources of earth (soil, water, materials), and through her conduct she brings them into relation. Xunzi, the Confucian philosopher of the third century BCE, put it in a famous passage. Heaven has its seasons, earth has its resources, and humanity has its government (治 zhì), its ordering. That is how humanity can form a triad with heaven and earth. Humanity is the third power because it orders what heaven and earth provide.

The Zhongyong (中庸, "Doctrine of the Mean"), one of the Four Books of Neo-Confucian education, makes the same point in its culminating passage. The person of utmost sincerity can fully realise her own nature, and through it the natures of others and of things, and so assist in the transforming and nourishing of heaven and earth, forming a triad with them.

This is the Chinese answer to Snowden's question, and it is a strong one. The Chinese sources handle mediation through the three powers, and the paradigm of mediation is the human. The human is not one pole of a dyad (human against nature, culture against nature) nor a component in a binary code. The human is the between, the position in which heaven and earth are brought into relation, and the human's distinctiveness lies in that position. The human is distinctive as mediator, not as a bearer of properties.

The Mediating Breath: Laozi 42

A second Chinese answer to Snowden's question is in the forty-second chapter of the Laozi, one of the most commented-on passages in Chinese philosophy:

道生一,一生二,二生三,三生萬物。 萬物負陰而抱陽,沖氣以為和。

In paraphrase: The Way gives rise to one; one gives rise to two; two gives rise to three; three gives rise to the ten thousand things. The ten thousand things carry yin on their backs and embrace yang in their arms, and the blending breath (沖氣 chōngqì) makes their harmony.

The passage is a cosmogony, and its structure answers Snowden's question exactly. Two (yin and yang) does not give rise to the ten thousand things directly. Two gives rise to three, and it is three that gives rise to the ten thousand things. The third is not a further pole. It is the blending breath, 沖氣, the qi that circulates between yin and yang, bringing them into harmony (和). Every thing carries yin and embraces yang, and what makes it a thing, what makes it cohere and live, is the blending breath between them.

On this reading, the Daoist cosmos is not dyadic. Yin and yang are two, but two alone generates nothing. Generation requires the third, the mediating breath. Commentators have long debated what "one", "two" and "three" refer to, and readings vary: some take three as heaven, earth and humanity, others as yin, yang and the blending breath. But on the readings most relevant here, the structure is triadic, with a mediating third that is neither of the two poles and without which the poles are sterile.

This is Peirce's thirdness, in a Daoist cosmogony. Firstness, the undifferentiated One; secondness, the two in reaction, yin and yang; thirdness, the mediating breath that brings them into harmony and makes generation possible. It is also Snowden's trialectic, stated twenty-four centuries earlier: three elements in tension, each pair mediated by the third, keeping the system generative. Ing's dyadicism, if it reads yin and yang as the whole of the Daoist logic, has missed the Laozi's third.

Harmony, Not Sameness: 和 and 同

A third Chinese answer to Snowden's question lies in the distinction between 和 (hé, harmony) and 同 (tóng, sameness or conformity), which runs through classical Confucian thought.

Analects 13.23 gives the distinction its canonical form: 君子和而不同,小人同而不和. The exemplary person harmonises but does not conform; the petty person conforms but does not harmonise. The Zuozhuan (左傳), the great historical commentary on the Spring and Autumn Annals, gives the distinction its most famous illustration, in a conversation attributed to the minister Yan Ying (晏嬰). Harmony, he says, is like making a soup. The cook combines water, fire, vinegar, sauce, salt and plums with fish and meat, and blends them over the fire. He supplements what is lacking and reduces what is excessive. The result is a soup whose flavour is harmonious, a blend in which each ingredient contributes and none is lost. Sameness is like adding water to water. Nothing is gained, and nothing is made. A ruler who wants his ministers to agree with him in everything wants water added to water. A ruler who wants harmony wants his ministers to supply what he lacks and correct what he overdoes, as the cook's ingredients correct one another.

The distinction answers Snowden's concern about dichotomising in a way that neither dyad nor triad alone does. Snowden worries that pairs harden into either/ors, with sides to be taken. The Confucian 和 is a practice of holding differences together without reducing them: the soup in which vinegar and plum, salt and water, each keep their character and together make a flavour none has alone. 和 is not a third element added to two. It is the art of mediation among any number of elements, the cook's art, in which the mediator supplements and reduces until the differences blend. And its opposite, 同, is not dichotomy but conformity, the collapse of difference into sameness.

That reframes the dispute. Snowden's worry is that dyads collapse into dichotomy. The Confucian worry is that any plurality (two, three, many) can collapse into conformity, and that the remedy is not a particular number of elements but the practice of harmonising, which keeps differences alive by blending them. A triad can collapse into conformity as easily as a dyad can harden into dichotomy. Simmel's coalition of two against one (Chapter 25) is exactly such a collapse. The safeguard is the cook, the practice of supplementing and reducing, not the number of ingredients.

和 and the instruments

和 offers a criterion for narrative instruments. Do they harmonise, or do they make the same? An instrument that produces alignment, a population whose significations converge on the instrument's categories (Thesis VIII), produces 同: water added to water. An instrument that produces harmony keeps the differences alive, the outlier stories, the contested readings, the established's and the outsiders' different experiences (Chapter 28), and helps the population hold them together without reducing them. The Tidal Loom's quorum with inhibition, seeking a decision people can live with rather than consensus, is a practice of 和. It does not ask everyone to agree. It asks that the differences be blended into something workable, with stop signals to prevent any one flavour from overwhelming the rest.

The Centre: 中

A fourth Chinese resource is 中 (zhōng), the centre or mean, central to the Zhongyong. In the Zhongyong's opening, zhōng is the state before the feelings of pleasure, anger, sorrow and joy have arisen, and hé is their arising in due measure. Zhōng is the great root of the world and hé its universal path. When centrality and harmony are realised, heaven and earth take their proper places and the ten thousand things flourish.

Roger Ames and David Hall, in Focusing the Familiar (2001), argued that zhōng should not be read as a static midpoint between extremes, the Aristotelian golden mean read as arithmetic average, but as a dynamic focus: the point of balance that is continually found and lost and found again in the flux of experience. Zhōngyōng, which they translated as "focusing the familiar", is the practice of finding the centre in ordinary life, in the familiar relations and activities of everyday conduct, rather than in abstract principles. The centre is not a position one occupies. It is a balance one keeps, like a dancer's or a tightrope walker's, by continual adjustment.

To the Strait-Jacketed, the Other Always Seems Baggy used Aristotle's own version of the point: the golden mean in the Nicomachean Ethics (II.6) is relative to us, not to the thing itself. The mean is not a fixed point between two vices. It is the right response for this person, in this situation, found by practical wisdom. Zhōng and Aristotle's mean converge. Both are dynamic, both are relative, and both are found in practice by an agent inside the situation, not computed from outside.

Answering Snowden

The Chinese sources' answer to Snowden's question can now be stated.

Resource Source How it handles mediation Bearing on dyad vs triad Bearing on the human
三才 the three powers Xici; Xunzi; Zhongyong Heaven, earth and humanity; the middle mediates Trigram lines are positions, not bits; thirdness in the structure The human is the mediator, distinctive by position
沖氣 the blending breath Laozi 42 Two gives rise to three; the third, the blending breath, makes harmony and generation Daoist cosmogony is triadic; Ing's dyadicism misses the third Every thing lives by mediation between its poles
和 harmony vs 同 sameness Analects 13.23; Zuozhuan The cook's art of blending differences Number matters less than practice; any plurality can collapse into conformity Harmonising is a human practice; the minister supplies what the ruler lacks
中 the centre Zhongyong; Ames & Hall Dynamic focus, continually found in practice Not a midpoint between poles but a balance kept by an agent The centre is found by someone inside the situation

So yes, the Chinese sources have ways of handling mediation, and they are central, not marginal. Ing's dyadicism, if it reduces the Chinese logic to yin and yang, reads the sources partially. Snowden's objection to the trigrams, that they are 2³, reads them partially too, missing the positional reading of the Xici. But Snowden's deeper instinct, that mediation is irreducible and that the system stays open only when the third is present, is confirmed by the Laozi and the Xici. And Snowden's concern about dichotomising is reframed by the Confucian 和. The danger is collapse into sameness, whatever the number, and the remedy is the practice of harmonising, not the choice of three over two.

The Human Between

The most important result of this chapter for the disputation as a whole is that the Chinese sources locate mediation paradigmatically in the human, and locate the human's distinctiveness in her position as mediator, not in properties she has.

In the three powers, humanity is the middle. Heaven provides seasons, earth provides resources, and humanity orders them. Humanity's distinctiveness is not that it has properties heaven and earth lack, though in the Confucian account it does have a heart-mind (心) capable of moral cultivation. It is that humanity occupies the between, the position from which heaven and earth are brought into relation. The human is distinctive relationally, by her place among the powers.

Chapter 19 noted that the character 仁 (rén), the Confucian virtue of humaneness, is often glossed as combining 人 (person) and 二 (two): personhood as two-ness, the human as between. The three powers generalise the gloss. The human is between persons (仁) and between heaven and earth (三才). In both cases the human is constituted by a relation, by her place in a configuration. This is the Chinese version of the relational account of the human that Book Five found in Mead, Elias, Juarrero, Ingold, Hutchins and Peirce.

It also answers the possessive fallacy (Thesis I) from inside the tradition from which anthro-complexity takes propensity. The human of the three powers does not hold identities, act with intention or create abstractions as possessions. She mediates. She stands between and brings into relation, and her identity, intention and capacity for abstraction are aspects of her mediating position. A complexity of the human that drew on Chinese thought for 勢 could draw on it for 三才 as well, and would find there a human who is distinctive because she is the third.

The Triquetra Revisited

Snowden's visualisation of the trialectic is the triquetra: three interlaced arcs, a Celtic knot, each pair of arcs crossing within the third. The image is apt for the trialectic as Snowden states it. Three elements, each pair mediated by the third, none reducible to the others. But the Chinese resources suggest three refinements, each of which would bring the triquetra closer to the mediation this chapter has described.

Position. In the triquetra the three arcs are symmetrical and interchangeable. In the three powers, the three positions are different. Heaven above, earth below and humanity between are not interchangeable, and the middle has a mediating role the others lack. A triadic structure in which one term mediates the other two (Peirce's interpretant mediating sign and object, the blending breath mediating yin and yang, humanity mediating heaven and earth) is not symmetrical. Its third is special. The triquetra's symmetry, attractive as it is, may obscure the asymmetry of mediation.

Practice. The triquetra is a static image, three arcs fixed in their interlacing. The Confucian 和 and the Zhongyong's 中 are practices, the cook's blending and the dancer's balance, continually performed by an agent. A triad that stays open does so not because of its geometry but because someone keeps it open, by supplementing and reducing, by finding the centre again and again. The triquetra needs a cook.

The human in the knot. The triquetra, like the waterfall image of Chapter 1, has no human in it. It is a diagram of three elements, viewed from outside. The three powers place the human inside the triad, as its middle term. A trialectic of human systems in the Chinese mode would place the human, and the practitioner, inside the knot, as the strand that mediates and is mediated, not outside it as the one who draws the diagram.

The Simplex Again

Thesis VII argued that the barycentric triad, the instrument through which SenseMaker embodies the trialectic, is a simplex. Its closure builds in a zero-sum trade-off, it conflates strong-in-all with weak-in-all at the centre, and it places the interpretant with the analyst. The Chinese resources sharpen the point.

The simplex is a geometry of shares. Three components compete for a fixed total, like sand, silt and clay in soil. The three powers are not shares. Heaven, earth and humanity do not compete for a fixed total of being. Each is fully itself, and their relation is mediation, not allocation. The blending breath of Laozi 42 does not take a share from yin and yang. It circulates between them, and makes harmony from their full presence. The Confucian soup is not a mixture of fixed proportions. It is a blend in which each ingredient contributes its full character. In every case, the Chinese model of mediation is a model of full presence in relation, not competing shares of a whole.

The simplex therefore represents the opposite of what the Chinese sources mean by mediation. It represents allocation, where they mean harmonisation. A narrative instrument that honoured the Chinese model would let each element be fully present (independent scales, Thesis VII's remedy) and would represent their relation, how the narrator experienced them coming together or pulling apart, rather than their proportions.

Verdict on the Dyad-and-Triad Point

The November post's dyad-and-triad point is the one Snowden holds most loosely, and this chapter's verdict honours the looseness.

On the dyad-and-triad point. partial

Snowden is right that mediation is irreducible, that a system stays open only when a third is present, and that Ing's dyadicism, insofar as it reduces Chinese logic to yin and yang, misses the third. The Laozi's blending breath, the Xici's three powers and the Confucian 和 all confirm it. Snowden's objection to the trigrams as 2³ holds against a combinatorial reading but not against the positional reading of the classical commentaries, in which the trigram's lines are heaven, humanity and earth, and the middle mediates. Snowden's concern about dichotomising is reframed by 和 and 同: the danger is collapse into conformity, which can befall a triad as easily as a dyad, and the remedy is the practice of harmonising, not the number three. Finally, the Chinese sources place mediation paradigmatically in the human, distinctive by her position between, which supports the relational account of the human and corrects the possessive grammar of the human cluster.

Book Six Concluded

Book Six has read the disputation through 道理勢 and through the Chinese resources for mediation. Its findings converge with Book Five's.

  • 勢 includes the strategist, read by others, and the political 勢 of position. A 勢 read from outside is terrain, 地形.
  • 理 is known by the knife, slowly at the knots. Wang Yangming's bamboo shows that investigation from outside fails.
  • 道 is made by walking, by many walkers. Wúwéi is the third option between leaving alone and stimulating. Wǒ zìrán is the test of recognition. 經營 is the loom.
  • Mediation is real and irreducible, and its paradigm is the human between heaven and earth.

Book Seven turns to the observer inside, through the physics of measurement and the political theory of crowds, and then gathers the adversary's whole case.

The adversary will say that this chapter has done what Snowden warned against: absorbed Chinese concepts into a Western argument, read them through Peirce and Mead, and made them say what the author wanted. The three powers are a cosmology of a pre-modern agrarian empire. The Laozi's "one, two, three" is notoriously obscure, and its commentators disagree. 和 and 同 are terms of Confucian court politics. To enlist them in a critique of SenseMaker is anachronism.

The adversary is right that the chapter reads Chinese sources in the service of a contemporary argument, and right that their meanings are contested. But Snowden asked the question, whether the Chinese sources have a way of handling mediation that he is missing, and the question deserves an answer from the sources, not a refusal to engage them. The answer offered is a reading. The Xici's positional account of the trigram, the Laozi's "two gives rise to three" and the Analects' harmony-not-sameness are standard passages, read in standard ways. That they bear on a contemporary dispute is not anachronism. It is the reason classics are read. Anthro-complexity itself reads Sunzi and the Zhuangzi through Jullien for its own purposes, so this chapter does no more. partial

TWO GIVES RISE TO THREE — THREE TO THE TEN THOUSAND THINGS —

HEAVEN ABOVE, EARTH BELOW, AND THE HUMAN BETWEEN —

THE EXEMPLARY PERSON HARMONISES BUT DOES NOT CONFORM.

Addition: Cross-Lights From Book Nine

The triquetra. Since this chapter was written, Snowden has drawn his three positions of practice as a triquetra with an empty centre, and has given a test for a real third term: it must be neither a midpoint between the first two nor a dissolution of the frame. That is a statement about mediation and irreducibility, which is where this chapter located the issue. The count of terms is secondary. Chapter 50 asks whether his own third position passes the test.

The empty centre. The raked gravel at Ryōan-ji, where nobody may stand, is offered in Chapter 53 as an image for a centre that cannot be occupied.

Chapter 39

Wigner's Friend and the Claim to Stand Neither

4,101 words, about 18 minutes of reading

Book Seven: The Observer Inside

The eight theses of Book Four converged on a single point. The human distinction, followed through, re-enters among humans as an asymmetry between describers and described, and anthro-complexity's instruments keep the describers outside. Book Five found the same point in Western social thought, and Book Six found it in Chinese thought. Book Seven approaches it from two further directions. The first is the physics of measurement, where the observer's place in the system observed has been a central and unresolved problem for a century. The second is political theory, where the question of who speaks for a collective, and from where, is the question of representation and of the state. The Book ends with the adversary's whole case.

The physics comes with a warning. The Compass and the Loom devoted a Part to quantum measurement as a source of analogies for social sensing, and insisted on an analogical discipline: a quantum image may cross into social life only when its borrowed structure is real in the source domain and the inference does not depend on the source's physics. That treatise also stressed that the physics evidence on observer-relative facts is contested. This Book keeps both cautions. The physics is not offered as proof of anything about organisations. It is offered as a domain in which the problem of the observer inside has been posed with unusual precision, and from which some structures can be honestly borrowed.

The Measurement Problem

Quantum mechanics describes physical systems by states, mathematical objects that encode the probabilities of the outcomes of possible measurements. Between measurements, a system's state evolves smoothly and deterministically according to the Schrödinger equation. A system can be in a superposition, a state that is a combination of several possible outcomes, none of which is actual. When a measurement is made, one outcome is observed, and the state is updated to reflect it, the so-called collapse. The theory gives no account of what a measurement is, why it produces one outcome and not a superposition of outcomes, or where in the chain from system to apparatus to observer the collapse occurs. This is the measurement problem, and after a century it remains unresolved, with competing interpretations offering different answers.

Schrödinger's cat (1935) dramatised the problem. A cat is sealed in a box with a radioactive atom, a detector and a vial of poison. If the atom decays, the detector triggers the release of the poison and the cat dies. Quantum mechanics says the atom, after an hour, is in a superposition of decayed and not-decayed. If the theory applies to everything, the detector, the poison and the cat are in a superposition too, of alive and dead, until someone opens the box and looks. Schrödinger meant the example as a reductio. Cats are not in superpositions. Something in the theory, or its interpretation, must give.

Wigner's Friend

Eugene Wigner, in "Remarks on the Mind-Body Question" (1961), sharpened the problem by putting a person inside the box. Wigner's friend is in a sealed laboratory, performing a measurement on a quantum system, say the spin of an electron. The friend observes an outcome, up or down. Wigner is outside the laboratory, and from his perspective the whole laboratory, electron, apparatus and friend, is a quantum system that has not been measured. If quantum mechanics applies to everything, then from Wigner's perspective the laboratory is in a superposition: friend-saw-up plus friend-saw-down. When Wigner opens the door and asks his friend what she saw, the superposition collapses, from Wigner's perspective, to one outcome.

The puzzle is this. What was the state of the laboratory before Wigner opened the door? From the friend's perspective, she had seen a definite outcome. The measurement was done. From Wigner's perspective, the laboratory was in a superposition, and no outcome was yet definite. Both perspectives use the same theory, correctly. They disagree about what is real. Either the friend's observation was definite and Wigner's description incomplete, or Wigner's description was complete and the friend's experience of a definite outcome was in some sense not real until he observed it. Wigner himself concluded, at the time, that consciousness must play a special role in collapse. He later abandoned the view. The puzzle remains.

Recent work

In the past decade, Wigner's friend has been the subject of a series of theorems and experiments that sharpen the puzzle further. Daniela Frauchiger and Renato Renner's 2018 paper in Nature Communications constructed an extended Wigner's-friend scenario, with two friends and two outside observers, and showed that three seemingly reasonable assumptions cannot all hold. The assumptions are, roughly, that quantum mechanics applies universally, including to observers; that observers can reason consistently about one another's conclusions; and that a measurement has a single outcome. Kok-Wei Bong and colleagues, in a 2020 Nature Physics paper, proved a related "Local Friendliness" no-go theorem and reported an experiment, with photons standing in for friends, whose results violated inequalities that "Local Friendliness" would require. The experiment challenges the assumption that observed events are absolute, the same for all observers.

These results are contested. They depend on idealisations (a photon is not a friend), and different interpretations of quantum mechanics respond to them differently. The Compass and the Loom was careful to say so, and this chapter repeats the caution. But they show that the question of whether observations are absolute or observer-relative is a live question in the foundations of physics, not a philosophical fancy.

Relational and Agent-Centred Interpretations

Two interpretations of quantum mechanics respond to Wigner's friend in ways that bear on this disputation, and The Compass and the Loom drew on both.

Relational quantum mechanics, proposed by Carlo Rovelli in 1996, holds that quantum states are not absolute properties of systems. They are relations between systems. A system has a definite value of some quantity only relative to another system with which it has interacted. In Wigner's scenario, the electron's spin is definite relative to the friend, who has interacted with it, and indefinite relative to Wigner, who has not. There is no contradiction, because there is no absolute state. There are only relative states. Rovelli has argued that this is the deepest lesson of quantum mechanics. The world is made not of things with properties but of relations, and facts are relational. (Rovelli's popular books, Helgoland among them, develop the view for a general audience.)

QBism, developed by Christopher Fuchs, Rüdiger Schack and N. David Mermin, holds that quantum states are not descriptions of systems at all. They are an agent's degrees of belief about the outcomes of her own future actions on the world, her personal, Bayesian probabilities. A quantum state belongs to the agent who assigns it, not to the system. In Wigner's scenario, the friend and Wigner assign different states because they are different agents with different information, and both are right, because each state expresses its agent's beliefs. Measurement is not a passive observation. It is an action by the agent on the world, and its outcome is a personal experience of the agent who acts. QBism's slogan, in Fuchs's phrase, is that quantum mechanics is a user's manual for agents acting in the world, not a description of the world from nowhere.

The Compass and the Loom used both to question the "single source of truth" dashboard, and drew a distinction from QBism that bears on the human/AI question. An AI can output probabilities but lacks the stake that makes them its own, so it lacks sentience in that sense. QBism's probabilities are an agent's own, because the agent has something at stake in the outcomes, namely her own future experience. A system that computes probabilities without a stake in their outcomes is not an agent in QBism's sense, however accurate its probabilities.

The Claim to "Neither"

Chapter 20 introduced the parallel The Compass and the Loom drew between Wigner's position and Snowden's claim to "neither". The claim to stand outside the lab parallels Snowden's claim to "neither". This chapter can make the parallel precise.

Wigner, outside the laboratory, describes it as a system in superposition. His description is correct by the theory. But it is a description from a position, outside the lab, and from that position the friend's definite experience is invisible. Wigner's description is complete relative to him and incomplete relative to the friend. If Wigner forgets that his description is relative to his position, and takes it as the absolute truth about the lab, he denies the reality of his friend's experience. That is the error the relational and agent-centred interpretations diagnose. No observer's description is absolute, and every description is from somewhere.

Snowden's claim to "neither" (neither baggy nor boxed, neither systems thinking nor flat ontology, neither dyadic nor dualistic) describes two positions from a third, which is presented as outside both. From the third position, both appear as errors. But the third position is a position. It has its own perspective, its own form, which it does not see. To the Strait-Jacketed, the Other Always Seems Baggy found the structure exactly: each side sees the other's form as "baggy" or "boxed", while its own form feels like freedom. The claim to "neither" is Wigner's error in the field of ideas. It takes a description from one position as an absolute description of the others.

Snowden does not, in fact, make Wigner's error crudely. His postscripts, revisions and invitations to correction (Chapter 20) acknowledge that his descriptions are from a position. But the instruments, unlike their author, do not acknowledge it. A SenseMaker dashboard presents patterns as the state of the organisation, not as the state relative to the instrument's position. An estuarine map presents constraints as the organisation's constraints, not as constraints relative to the mapping session. A Cynefin categorisation presents the situation as complex, not as complex relative to the categoriser. Each is Wigner outside the lab, describing what he sees as what is.

The Single Source of Truth

The phrase "single source of truth" is common in contemporary data management. It is the aspiration that an organisation's data should come from one authoritative source, so that everyone sees the same numbers and decisions are made on a consistent basis. The aspiration is reasonable for some data, such as inventory counts, account balances and transaction records, which are facts that do not depend on who records them.

For narrative data, the aspiration is Wigner's error institutionalised. A single source of truth about an organisation's culture, morale or risks presents one perspective, the instrument's or the analyst's or leadership's, as the organisation's state. But on the relational account, the organisation has no single state. It has states relative to its members, each of whom has interacted with it differently. The established and the outsiders (Chapter 28) live in different organisations, though they work in the same building. The frontline and the boardroom see different patterns, because they stand in different relations. A single dashboard that presents one pattern as the truth erases the relativity, as Wigner erases his friend's experience by taking his outside description as absolute.

The relational alternative is not relativism, the claim that every perspective is equally valid and none can be wrong. Rovelli's relational quantum mechanics is not relativist. Relative states are objective facts about relations, and they can be checked when systems interact. Nor is QBism relativist. Agents' beliefs are disciplined by the outcomes of their actions. The relational alternative is perspectivism with interaction. Each perspective is a real relation to the organisation, perspectives can be compared when their holders interact, and the comparison is itself an inquiry (Peirce's community, Chapter 32). A relational dashboard would present the organisation's states relative to its different members and groups, and would show where they agree and where they differ. It would let the holders of different perspectives interact, comparing and contesting them, so that the organisation's self-understanding is the outcome of the interaction, not the imposition of one perspective as truth.

The View From Somewhere

The physics of observer-relative facts has a counterpart in philosophy and feminist epistemology that makes the point without quantum mechanics.

Thomas Nagel's The View from Nowhere (1986) examined the aspiration to objectivity as the attempt to view the world from no particular perspective, detached from the viewpoint of any particular person. Nagel took the aspiration seriously. Objectivity, he argued, is achieved by stepping back from one's perspective and including it in a wider view. But he also argued that the view from nowhere can never be complete, because there are facts, the facts of subjective experience, that are essentially perspectival and cannot be captured from outside. A complete account of the world must include both the view from nowhere and the views from somewhere, and their relation is a permanent philosophical problem.

Donna Haraway's "Situated Knowledges" (1988) put the critique more sharply. The view from nowhere, she argued, is a god trick: the pretence of seeing everything from nowhere, which is in fact a view from a particular, powerful position that has disguised its particularity. The scientist, the state and the institution that claim a view from nowhere are claiming the authority of a god while occupying the position of a person, with that person's interests and blind spots. Haraway's alternative is situated knowledge: knowledge that acknowledges its location, takes responsibility for its perspective, and gains objectivity not by escaping location but by connecting partial perspectives, through what she called "webs of connections" among situated knowers. Objectivity is the outcome of accountable partiality, not of the god trick.

The Compass and the Loom used QBism for a "view from somewhere", with skin in the game. Haraway supplies the political dimension. The view from nowhere is a claim to power, and situated knowledge is its accountable alternative. A dashboard that presents a single source of truth about an organisation's culture performs the god trick. It presents one perspective, the analyst's or leadership's, as the view from nowhere. A relational dashboard practises situated knowledge. It shows each perspective's location and connects them in a web.

Mowles's staying with the trouble

Haraway's later work, Staying with the Trouble (2016), gave Chris Mowles the motto The Compass and the Loom and To the Strait-Jacketed both discuss. Staying with the trouble means remaining in the difficulty, the entanglement, the situated mess of a problem, rather than escaping into a framework that promises to resolve it from outside. Mowles used it against frameworks that offer clarity at the cost of situatedness. The phrase connects to Wigner's friend. The friend in the laboratory is staying with the trouble. She is inside the system, with a definite experience, and does not have Wigner's luxury of describing the whole as a superposition. Staying with the trouble is choosing the friend's position over Wigner's: the situated, inside, accountable view.

What Crosses, and What Does Not

The analogical discipline requires an explicit account of what crosses from the physics to the social, and what does not.

Quantum structure Real in physics? Social analogue Depends on physics? Crosses?
Observer-relative states (RQM) Yes, as an interpretation; contested An organisation's state is relative to the member who stands in relation to it No; follows from members' different interactions holds
States as an agent's beliefs (QBism) Yes, as an interpretation; contested A pattern is the analyst's belief about the organisation, not its state No; follows from the pattern being produced by the analyst holds
Wigner's friend disagreement Yes; central to foundations Inside and outside descriptions of an organisation can both be correct and disagree No; follows from different positions holds
Local Friendliness violation Reported experimentally; contested Observed events are not absolute for all observers Partly; the social version is weaker, since social facts are shared through interaction partial
Consciousness causes collapse (early Wigner) No; abandoned by Wigner; fringe Attention changes what is attended to The social version does not need collapse; attention effects are psychological fails as physics; the social claim stands on its own grounds
Superposition of organisational states No; organisations are not quantum systems An organisation is "in many states at once" until sensed Yes; the inference depends on quantum physics fails

What crosses is the structure of observer-relativity. Descriptions are from positions, positions differ, and descriptions from different positions can both be correct and disagree. That structure is real in physics, under several interpretations, and real in social systems, for social reasons. What does not cross is the idea that organisations are literally in superposition, or that sensing literally collapses them. Those inferences depend on physics that does not apply.

Wigner and the Theses

Thesis II (asymmetry applied to itself). Wigner outside the lab describes from a position, and so does the practitioner outside the organisation. Each description is correct relative to its position and incomplete relative to others. The asymmetry principle applied to itself is the recognition that the practitioner is Wigner, and the narrators are friends whose experience her description does not capture.

Thesis III (estuaries). The sailor on the deck is Wigner, describing the estuary from outside. The fisherman in the maelstrom is the friend, inside, with a definite experience. Elias's choice of the fisherman is Wigner's friend chosen over Wigner.

Thesis V (power). Haraway's god trick names the power in the view from nowhere. The single source of truth is a claim to authority disguised as a claim to objectivity.

Thesis VI (computation). QBism's stake: an AI's probabilities lack the stake that makes them an agent's own, so the human/AI hybrid must keep the stake with humans.

The adversary will say that quantum analogies in management are a notorious source of nonsense, and that this chapter, for all its analogical discipline, adds to the pile. The social conclusions (descriptions are from positions, dashboards should show perspectives, objectivity is accountable partiality) can all be reached from Nagel, Haraway, Mead and Elias without any physics. Bringing in Wigner and Rovelli adds glamour, not argument, and lends the conclusions an authority they have not earned.

The adversary wins a strong half. The social conclusions do not need the physics, and the chapter's table says so: every crossing that holds holds "for social reasons". The physics is offered not as proof but as precision. The measurement problem poses the question of the observer inside with a rigour that social theory rarely achieves, and the recent theorems show that the question is not merely philosophical. If the physics adds glamour, the analogical table is the discipline that keeps the glamour from doing the argument's work. The reader who distrusts quantum analogies may skip to the Haraway section and lose nothing essential. partial

WIGNER OUTSIDE THE DOOR — THE FRIEND INSIDE THE LAB —

BOTH CORRECT, AND THEY DISAGREE —

THERE IS NO VIEW FROM NOWHERE, ONLY THE GOD TRICK.

Addition: Sketch of a Relational Dashboard

The chapter argued that a single source of truth about an organisation's culture is Wigner's error institutionalised, and that the relational alternative is perspectivism with interaction. This addition sketches what a relational dashboard might look like. It is offered as a design proposal, by someone outside the practice, to be tested and improved by those inside it.

Design principles

The dashboard follows from five principles drawn from the disputation.

  1. Every pattern is relative to a position. Patterns are shown for groups, not of the organisation: for the frontline, for middle management, for leadership, for the established and the outsiders, for each function. The organisation-wide pattern is shown, but labelled as one perspective among others, an aggregate from nowhere in particular (Chapter 39).
  2. Perspectives are shown side by side. Where groups' patterns differ, the difference is the finding. The dashboard highlights where perspectives diverge, not only where they converge.
  3. Readers are readable. Each group's reading of the patterns, what they make of them, is recorded and shown, alongside leadership's reading and the analyst's. The interpretant is visible (Chapter 32).
  4. Horizons overlap. Every group sees every other group's patterns and readings, at the same time, within the limits of anonymity and confidentiality. No group, leadership included, sees what others cannot (Chapter 31).
  5. The dashboard shows its own position. The declination table, the signifier set's date, the order randomisation and the crowd crystal are visible on the dashboard itself, so that anyone reading it knows from where it was drawn (Chapter 45).

Layout

The relational dashboard has five panels.

Panel 1: Positions. A grid of the groups whose narratives contributed, with the number of narratives from each and the date range. It also shows the groups who did not contribute, or contributed few narratives, as the margin in Midgley's sense, and states that their perspective is absent.

Panel 2: Patterns by position. For each triad and dyad, the distribution of significations shown separately for each group, in small multiples, side by side. The organisation-wide aggregate is shown last, labelled aggregate: not any group's view.

Panel 3: Divergences. A ranked list of the signifiers on which groups' patterns diverge most, with the divergence shown visually. Divergence is presented as information about the organisation's figuration (Elias), not as noise to be averaged away.

Panel 4: Readings. For each group, a summary of how its members read the patterns when they were returned: what they said fits, what they said does not, what they said is missing. Leadership's reading and decisions are shown here too, in the same format, with the date.

Panel 5: The instrument. The declination table, the signifier set and its expiry date, the consent tiers used, the crowd crystal named, the order randomisation report and a note of any reflexive content (stories about previous rounds' effects).

What the relational dashboard does not do

It does not produce a single score, index or headline. There is no "culture health score" and no "engagement index". Those are the god trick in numerical form (Haraway, Chapter 39). If a single number is demanded, by a board, a regulator or a client, the dashboard can supply the aggregate, labelled, but it will not present the aggregate as the truth.

It does not rank groups. Divergences are shown, but no group's perspective is marked as correct. The established's and the outsiders' patterns are shown side by side, and the dashboard does not say which is the organisation's reality (Chapter 28).

It does not conceal leadership's reading. Leadership sees what everyone sees, and everyone sees what leadership concluded and did.

Who would resist it

A relational dashboard would be resisted, and the resistance is predictable from the disputation's analysis.

Leadership may resist because it loses the asymmetry of visibility (Foucault, Chapter 23). It sees no more than others, and its readings are exposed.

Boards and regulators may resist because they want a single number, a single source of truth, for accountability.

Commissioners of studies may resist because the relational dashboard is harder to act on. It shows divergence where they hoped for direction.

Analysts may resist because their interpretive authority is diminished. Their reading is one of several, shown beside the narrators' own.

Each form of resistance is a sign that the dashboard is doing what it was designed to do: distributing the horizon of observation, declaring position and refusing the god trick. A practitioner deploying it would need to negotiate with each resisting party, in the light of CST's participants question (Chapter 12). In a coercive context, the full relational dashboard may be the most important protection the population has. In a unitary context, a lighter version may serve.

A note on feasibility

Nothing in the relational dashboard requires new technology. Small multiples, divergence rankings and side-by-side panels are standard data-visualisation techniques. The readings panel requires a second round of collection, in which narrators respond to the patterns, and that has a cost. The instrument panel requires that the declination table and other records be kept and published, which costs almost nothing. The main obstacle is not technical. It is the willingness of those who commission studies to see their organisation as a set of perspectives rather than a single picture.

NOT THE ORGANISATION'S PATTERN — THE PATTERNS FOR EACH POSITION —

DIVERGENCE IS THE FINDING —

AND THE DASHBOARD SHOWS FROM WHERE IT WAS DRAWN.

Addition: Cross-Lights From Book Nine

The practitioner inside, in Snowden's words. The post on Luhmann says that nobody stands outside, the practitioner least of all, and that both he and Luhmann reject a view from above. They differ on what the observer is part of: a system of communication for Luhmann, an entanglement of people, stories and material conditions for Snowden. The argument of this chapter is to that extent agreed. Chapter 52 asks the remaining question, which concerns the designer of the instrument.

Chapter 40

Order, Zeno, Decoherence: The Physics of Asking

3,469 words, about 15 minutes of reading

The Physics of Asking

Chapter 39 took the observer inside from the side of what is observed: the relativity of states and facts to the observer's position. This chapter takes it from the side of the act of observing: what asking a question, taking a measurement or sensing a population does to the thing asked, measured or sensed. In quantum mechanics, measurement is not passive. It changes the system measured, and the change has a precise structure that physicists have studied for a century. In human systems, asking is not passive either. It changes the person asked, and the change has a structure that cognitive scientists have recently modelled using the mathematics of quantum measurement.

The chapter takes four phenomena: question-order effects, the quantum Zeno effect, decoherence and einselection, and weak measurement. The Compass and the Loom's Uncertain Compass drew its "four laws of the uncertain needle" from them. For each, the chapter states the physics, the social analogue and the analogical discipline's verdict, and draws the consequence for anthro-complexity's instruments.

Order: Non-Commuting Questions

The physics

In quantum mechanics, some pairs of measurements do not commute. The order in which they are made matters. Measuring a particle's position and then its momentum gives different results, in general, from measuring momentum and then position. This is because each measurement changes the system's state, and the state the second measurement meets depends on which measurement came first. Non-commutativity is the mathematical root of Heisenberg's uncertainty principle and of Bohr's complementarity. Some properties cannot be jointly determined, because determining one disturbs the other.

The social finding

Survey researchers have long known that the order of questions affects answers. A question about a specific issue asked before a general question shifts the answer to the general one, and the other way round. Classical models of judgement had trouble explaining the pattern of these order effects, because, on a classical probability model, the joint probability of answering yes to A and yes to B should not depend on the order in which A and B are asked.

Zheng Wang, Tyler Solloway, Richard Shiffrin and Jerome Busemeyer, in a 2014 paper in the Proceedings of the National Academy of Sciences, proposed that order effects could be modelled by treating survey questions as non-commuting measurements on a cognitive state, using the mathematics of quantum probability, though not, they emphasised, implying that the brain is a quantum system. Their model made a distinctive prediction, which they called the QQ equality: a specific combination of the order effects (for yes-yes and no-no answers) should cancel exactly. They tested it against data from about seventy national surveys and found that the equality held, in a way that classical models did not predict. The Compass and the Loom summarised the finding: order effects in yes-yes and no-no answers offset exactly.

The work belongs to the wider programme of quantum cognition, set out by Busemeyer and Peter Bruza in Quantum Models of Cognition and Decision (2012), which uses quantum probability's mathematics to model judgements that violate classical probability. Examples are the conjunction fallacy, the disjunction effect and order effects. The programme is debated, but its order-effect results are among its most robust.

The analogue and its discipline

The borrowed structure is this: a question changes the state of the person answering it, so the state a second question meets depends on the first. In physics the structure is real. In cognition it is supported by the order-effect literature, independent of quantum physics, because asking a question makes some considerations salient and those considerations shape later answers. The quantum mathematics is a model of the structure, not its cause. The analogy passes.

The consequence for instruments

The Compass and the Loom drew the design consequence: signifier sets are sequences of interventions that change the respondent. A SenseMaker signifier set is not a battery of independent measures applied to a fixed state. Each triad, dyad and question changes the narrator's state, by making some aspects of her story salient and others recede, and the next signifier meets the changed state. The order of signifiers therefore shapes the signification. A narrator who first places her story on a triad of "trust, compliance, fear" and then on a dyad from "too much control" to "too little" will signify differently from one who meets the dyad first, because the triad has framed her story in terms that colour her response to the dyad.

The practical remedies are known from survey methodology: randomise the order of signifiers across narrators, or at least across subgroups, and report order effects. A more radical remedy follows from the relational account. Treat each signification as a sequence and record the sequence, so that the analyst can see not only where each story was placed but in what order the placements were made, and how earlier placements shaped later ones.

Zeno: The Watched Pot

The physics

The quantum Zeno effect, named by Baidyanath Misra and George Sudarshan in 1977 after Zeno's paradox of the arrow, is the phenomenon whereby frequent measurement of a quantum system inhibits its evolution. A system that would, left alone, make a transition (an atom decaying from an excited state, say) is prevented from doing so if it is measured often enough, because each measurement projects it back onto its initial state before the transition has had time to develop. In the limit of continuous measurement, the transition never happens. The effect was demonstrated experimentally by Wayne Itano and colleagues in 1990, using trapped ions whose transitions were inhibited by frequent laser pulses. The popular gloss is a watched pot never boils, which in quantum mechanics is, under the right conditions, literally true.

The social analogue

The Compass and the Loom's Zeno law states it: over-frequent monitoring freezes change, so monitor less often than the change you hope to see. Chapter 26 argued that the borrowed structure, frequent measurement resetting a system to the measured state and so inhibiting transitions that need time to develop, has a social counterpart independent of quantum physics. A frequently asked question resets the respondent's attention to the question's frame, and the emergence of new noticing needs time free of the frame.

The social Zeno effect has several mechanisms. Frequent pulse surveys keep employees' attention on the dimensions surveyed, crowding out attention to emergent matters. Frequent sentiment monitoring makes people self-conscious about expressing sentiment, so that expression conforms to what is monitored. Frequent narrative collection trains narrators in the instrument's categories (Thesis VIII's looping), so that new experiences are categorised before they can be noticed as new. In each case the monitoring holds the population in the state the monitoring expects, and the transition to a new state, the emergence the monitoring was meant to detect, is inhibited.

The consequence for instruments

The Zeno law gives a design rule that is rare in management practice: monitor less. Continuous sensing, real-time dashboards and always-on narrative capture are often presented as improvements on periodic sensing, because they detect change sooner. The Zeno law says they may detect less change, because they inhibit it. The right frequency of sensing depends on the timescale of the change one hopes to see, and it should be lower than that timescale, so that change has time to develop between measurements.

Decoherence: The Selection of Cliché

The physics

Decoherence is the process by which a quantum system's interaction with its environment destroys the coherence, the definite phase relations between components of a superposition, that makes interference effects possible. A system in isolation can be in a superposition whose components interfere, producing the characteristic patterns of quantum behaviour (the double-slit fringes, for example). A system that interacts with its environment becomes entangled with it, and the interference effects are suppressed, because the information about which component the system is in has leaked into the environment. Decoherence explains why macroscopic objects do not show quantum interference. They interact too strongly with their environments.

Wojciech Zurek's work on einselection (environment-induced superselection) showed that decoherence does not destroy all states equally. It selects a small set of pointer states, states that are robust against the environment's monitoring and survive the interaction, while superpositions of pointer states are rapidly destroyed. Zurek's later "quantum Darwinism" proposed that pointer states are those whose information is copied redundantly into the environment, so that many observers can learn about them without disturbing them. The classical world, on this view, is the world of pointer states, selected by the environment's continuous monitoring.

The social analogue

The Compass and the Loom's decoherence law states it: saturating expectation selects safe, reportable clichés and destroys interference, so protect pockets of low monitoring. The borrowed structure is this: continuous interaction with an environment that "monitors" selects robust states that survive monitoring and destroys fragile combinations. In social systems the analogue is that continuous exposure to expecting audiences, such as instruments, managers and peers, selects robust, expected, reportable states (clichés, safe opinions, the organisation's official vocabulary) and destroys the fragile, novel, interfering combinations from which new ideas emerge.

The social analogue has several forms. Organisational vocabulary becomes clichéd as it is repeated in every report and meeting, so that "agility", "resilience" and "empowerment" lose their power to mean. Narratives converge on safe forms that survive the instrument's expectations. Innovation is suppressed in environments where every half-formed idea is immediately exposed to evaluation. The interference that novelty needs, combining ideas that do not yet fit together and holding contradictions open long enough for something new to emerge, requires a degree of isolation from expectation, a protected space.

The analogical discipline

Does the analogy pass? The structure, interaction with an expecting environment selecting robust states and destroying fragile combinations, has a social basis independent of quantum physics, in social desirability, conformity pressures and the evaluation apprehension documented in the creativity literature. The analogy passes, with the caution that "interference" in the social case is a metaphor for the combination of ideas, not a literal phase relation.

The consequence for instruments

The decoherence law gives a second design rule: protect pockets of low monitoring. Some parts of an organisation, some times and some topics should be lightly sensed or not sensed at all, so that novelty can form outside the instrument's expectations. That is the opposite of comprehensive sensing, the aspiration to see everything. A practice faithful to the decoherence law would map where its sensing is not, as well as where it is, and would protect the unsensed spaces as the places where the organisation's future is being made.

Complementarity: Path or Place

The physics

Niels Bohr's principle of complementarity holds that some properties of a quantum system are complementary. They cannot be observed simultaneously, because the experimental arrangements needed to observe them are mutually exclusive. The paradigm is wave and particle. An experiment that shows which slit a particle went through (its path) destroys the interference pattern (its wave character), and an experiment that shows the interference pattern cannot show the path. Both are real aspects of the system, but they cannot be seen together. The Compass and the Loom stressed that complementarity is distinct from measurement disturbance, though often confused with it. Complementarity is about mutually exclusive experimental arrangements, not about one measurement disturbing another.

The social analogue

The Compass and the Loom's complementarity law states it: a reading chooses between knowing where and knowing how, Path or Place. An instrument that shows where things are (a snapshot of the organisation's state, a map of constraints, a cluster of narratives on a triad) cannot at the same time show how things are moving (the vector, the trajectory, the process by which the state is changing). The arrangements are different. A snapshot requires freezing the motion, and seeing the motion requires not freezing it. Anthro-complexity's Element 4, the shift from state to vector, is a choice of path over place. But many of its instruments are place instruments (maps, dashboards, clusters) that show where, not how.

The consequence for instruments

Complementarity counsels declaring which reading an instrument makes. An estuarine map is a place instrument. It shows where constraints are, not how they are changing. A narrative instrument used repeatedly can be a path instrument, if it tracks how significations move across rounds. But it cannot show both the current state and its trajectory in a single reading. Practitioners should know which they are getting and not mistake a place reading for a path reading.

No-Signalling: Harmony at the Rendezvous

The physics

Quantum entanglement produces correlations between distant particles that cannot be explained by any local hidden variables, as Bell's theorem shows. But entanglement cannot be used to send signals. The no-signalling theorem holds that no measurement on one particle of an entangled pair can transmit information to an observer of the other. The correlations are real, but they become visible only when the two observers' results are brought together and compared. Each observer, alone, sees only random outcomes. The harmony appears at the rendezvous.

The social analogue

The Compass and the Loom's no-signalling law states it: harmony between distant parties is recognised at the rendezvous, not transmitted. Two teams that share a common culture, a common history, or a common reading of a situation may act in correlated ways without communicating. But neither can send its alignment to the other, and the correlation becomes visible only when they meet and compare. Alignment cannot be transmitted from a centre. It can only be recognised when the parties come together.

The consequence for anthro-complexity's instruments is a critique of alignment by broadcast: the practice of sensing a population, deriving a direction and transmitting it to the population as the direction to align with. The no-signalling law says this cannot produce real alignment, only compliance with a broadcast. Real alignment, the correlation of independent action, is recognised at the rendezvous, when parties who have acted independently meet and find that their actions fit. That is recognition, not alignment (Chapter 37). The practice it recommends is to create rendezvous: occasions when parties who have acted independently come together and see whether their actions fit.

Weak Measurement and the Light Touch

The physics

A weak measurement, introduced by Yakir Aharonov, David Albert and Lev Vaidman in 1988, is a measurement that couples only weakly to the system measured. It extracts a small amount of information while disturbing the system only slightly. A single weak measurement is very noisy and tells little. But many weak measurements on an ensemble of identically prepared systems, averaged, can reveal information, the so-called weak values, that a strong measurement would destroy. Weak measurement has been used to trace the average trajectories of photons in a double-slit experiment without destroying the interference pattern, which strong which-path measurement would do.

The social analogue

The Compass and the Loom's Pyxis recast weak measurement as wúwéi (Chapter 37): the leader who measures weakly learns without freezing. A weak social measurement is a light touch. It is a question asked casually, a conversation that does not record, an observation that does not report, a sensing that informs without exerting the pressure of evaluation. A single light touch tells little. Many light touches, from many people, over time, can reveal patterns that a strong measurement, a formal survey or an audit, would distort or destroy.

The analogical discipline

The borrowed structure is that weak coupling extracts information with little disturbance, and that aggregated weak readings reveal what strong readings destroy. In social systems the structure is supported by research on reactivity: unobtrusive measures, informal conversation and ethnographic observation disturb less than formal instruments. The aggregation claim is plausible but less established. The analogy passes in its core, with the aggregation claim marked as partial.

The consequence for instruments

Weak measurement recommends a portfolio of light touches over a single strong instrument: many informal, low-stakes, unrecorded or lightly recorded sensings, aggregated over time by people who know the organisation, rather than one formal, high-stakes, comprehensive survey. Snowden's anecdote circles and Stage One narrative methods were closer to weak measurement than SenseMaker at scale, and this suggests a return to them. Not as a replacement for scale, but as the light touch that keeps the strong instruments honest.

The Four Laws Together

The Compass and the Loom's Uncertain Compass stated four laws of the uncertain needle. This chapter has added weak measurement as a fifth practice. Together they form a discipline of asking:

Law / practice Physics Social structure Design rule
Order Non-commuting observables A question changes the state the next question meets Randomise and record signifier order
Zeno Frequent measurement inhibits transitions Frequent sensing freezes emergent change Monitor less often than the change you hope to see
Decoherence Environmental monitoring selects pointer states Saturating expectation selects cliché Protect pockets of low monitoring
Complementarity Mutually exclusive arrangements: path or place A reading shows where or how, not both Declare which reading an instrument makes
No-signalling Correlations visible only at comparison Alignment cannot be broadcast; harmony is recognised at the rendezvous Create rendezvous, not broadcasts
Weak measurement Weak coupling, aggregated, reveals what strong destroys Light touches disturb less A portfolio of light touches beside strong instruments

The Pyxis Game

The Compass and the Loom turned the laws into a game, the Pyxis game, with seven rules, Attunement dice, a Zeno-Lock, Decoherence, the Twin (no-signalling), Turing's Glass (the Good–Turing estimate of missing mass) and Weak Reading. The game is a training device. It lets practitioners experience the laws by playing them, rather than reading about them, and so develops the practical sense, in Bourdieu's terms, that the laws require. It is mentioned here because it embodies a principle this chapter supports. The discipline of asking is a skill, not a set of rules, and skills are acquired by practice. A practitioner who has played the Pyxis game has felt the Zeno-Lock freeze her play and the Decoherence collapse her options, and will recognise them in an organisation.

The Physics of Asking and the Human Distinction

The physics of asking returns the disputation to its central claim. Human systems differ in kind from other complex systems because they are descriptively reflexive (Chapter 14): they are changed by descriptions of themselves. This chapter has shown that they are also changed by the act of describing, by asking, sensing and measuring, in ways that have a precise structure: order, Zeno, decoherence, complementarity, no-signalling. The structure is borrowed from physics as a model, not as a cause, and the social mechanisms are psychological and social.

The starling flock filmed from the Palazzo Massimo is not changed by the filming. A population sensed by a narrative instrument is changed by the sensing, in the order of the questions, the frequency of the sensing, the saturation of expectation, the choice of reading and the mode of feedback. Anthro-complexity's human distinction implies that its instruments are not cameras but probes, in the physicist's sense of instruments that interact with what they measure. A probe changes the system probed, and the physics of probing is a discipline anthro-complexity's instruments need.

The adversary will repeat the objection to Chapter 39: the physics adds glamour, and the social findings stand on their own. Order effects are known from survey methodology, observation effects from the Hawthorne literature, conformity from social psychology. Calling them "Zeno" and "decoherence" adds nothing but an aura of scientific authority to familiar observations, and risks the kind of quantum mysticism that brings complexity thinking into disrepute.

The adversary wins half again. The social findings do stand on their own, and the chapter's analogical discipline says so at each step. But the physics adds one thing beyond glamour: organisation. Order, Zeno, decoherence, complementarity and no-signalling are five distinct structures, each with a precise definition, and the physics shows how they differ and how they relate. Survey methodology knows order effects, social psychology knows conformity, and the Hawthorne literature knows observation effects. But they are scattered across literatures and rarely related. The physics offers a single framework in which the different ways asking changes the asked can be distinguished and designed for. That is a genuine contribution, provided the discipline holds. partial

THE ORDER OF QUESTIONS CHANGES THE ANSWERS —

THE WATCHED POT DOES NOT BOIL —

ALIGNMENT CANNOT BE BROADCAST; IT IS RECOGNISED AT THE RENDEZVOUS.

Addition: Cross-Lights From Book Nine

A second physical analogy. The Zen Garden study uses Maxwell's demon and Landauer's principle to argue that looking and sorting are never free and that the sorter is inside the box. It stands beside the Zeno effect of this chapter, with the same caution that these are analogies argued on independent grounds.

Parallel probes. Snowden's reason for running probes in parallel includes guarding against the effect by which anything new seems to work at first. That is an acknowledgement, in the method, of what this chapter describes.

Chapter 41

The Mariner Inside the Whale: Gratings and Mesh

4,169 words, about 18 minutes of reading

How the Whale Got His Throat

The first of Rudyard Kipling's Just So Stories (1902) tells how the whale got his small throat. In the sea, once upon a time, there was a whale who ate fishes: the starfish and the garfish, the crab and the dab, the plaice and the dace, the skate and his mate, the mackerel and the pickerel, and the really truly twirly-whirly eel. All the fishes he could find in all the sea he ate with his mouth, until at last there was only one small fish left, a small 'Stute Fish, who swam a little behind the whale's right ear, to be out of harm's way. The whale, hungry, asked the 'Stute Fish whether there was anything left to eat. The 'Stute Fish, being astute, told him of a man, a shipwrecked Mariner, sitting on a raft in the middle of the sea, with nothing on but a pair of blue canvas breeches, a pair of suspenders (which Kipling tells the reader, his Best Beloved, not to forget) and a jack-knife. The Mariner, the 'Stute Fish said, was a man of infinite-resource-and-sagacity.

The whale swam to the raft and swallowed the Mariner, raft, breeches, suspenders, jack-knife and all. Once inside, the Mariner stumped and jumped, thumped and bumped, pranced and danced, banged and clanged, hit and bit, leaped and creeped, prowled and howled, hopped and dropped, cried and sighed, crawled and bawled, stepped and lepped, and danced hornpipes where he shouldn't. The whale felt most unhappy, and asked the 'Stute Fish what to do. The 'Stute Fish told him to tell the Mariner to come out. The whale did, and the Mariner said he would come out only if the whale took him to his natal shore and the white cliffs of Albion. The whale swam there, as fast as he could. And the Mariner, before he stepped out, took his jack-knife and cut up the raft into a little square grating all running criss-cross, tied it firm with his suspenders, and dragged the grating good and tight into the whale's throat, where it stuck.

The grating is still there. From that day the whale could not cough it up or swallow it down, and could eat only very, very small fish. That is why whales nowadays never eat men or boys or little girls. The Mariner went home to his mother, who had given him leave to trail his toes in the water, and he married and lived happily ever after. So did the whale. But from that day on, the grating in his throat, which he could neither cough up nor swallow down, prevented him eating anything except very, very small fish. And the small 'Stute Fish went and hid himself in the mud under the doorsills of the Equator, because he was afraid the whale might be angry with him.

The Mariner as Figure

The Compass and the Loom made Kipling's Mariner the figure of the observer inside the system. Unlike Wigner, who describes from outside, the Mariner has existential stakes in the outcome. He is inside the whale. His fate depends on what happens there. He cannot describe the whale's digestive system from a neutral vantage, because he is in it, and its outcome is his survival.

The treatise drew three lessons from the story.

He intervenes by changing a constraint, not by targeting an outcome. The Mariner does not try to make the whale behave well, persuade it to stop eating, or redesign its appetite. He installs a grating, a constraint that limits what can pass, and the grating changes what the whale can do for ever after. That is anthro-complexity's own principle (Element 18): change the conditions, not the behaviour. But it is applied by someone inside the system, with stakes, using what he has to hand, not by a consultant designing interventions from outside.

He uses what is to hand. The grating is made of the raft and the suspenders, the only materials the Mariner has. Nothing is purpose-built. The treatise names the method in three words: bricolage, exaptation, détournement. Claude Lévi-Strauss's bricoleur, in The Savage Mind (1962), is the handyman who makes do with whatever is at hand, recombining materials that were made for other purposes, as against the engineer, who designs from first principles with purpose-made components. Stephen Jay Gould and Elisabeth Vrba's exaptation (1982) is the evolutionary co-optation of a trait for a function other than the one it evolved for, as feathers, perhaps evolved for warmth, came to serve flight. Snowden has made exaptation a central concept of his own practice: innovation as the repurposing of what exists. Guy Debord's détournement is the Situationist practice of turning existing images and texts against their original meaning. The Mariner is all three. He makes do, he repurposes the raft and suspenders, and he turns the whale's own appetite against it.

He leaves, and the constraint remains. The Mariner goes home, and the grating stays. His intervention does not require his continuing presence. It has changed the whale's constitutive constraints, its throat, so that the whale is now, permanently, a whale that eats only small fish. The intervention is not a programme that must be sustained. It is a change in what the system is.

The Mariner Against the Sailor

Set the Mariner beside the estuary's sailor (Thesis III) and Elias's fisherman (Chapter 28).

Estuary's sailor Elias's fisherman Kipling's Mariner
Position On deck, outside the medium In the maelstrom Inside the whale
Stakes The voyage Survival Survival
Relation to the system Reads it; waits for the tide Observes from inside; binds himself to what descends slowly Acts from inside; changes the system's constraint
Method Timing Detachment within involvement Bricolage; exaptation; détournement
Intervention Moves when the tide turns Changes his own relation to the system Changes the system's throat
What remains The voyage completed The fisherman saved; the maelstrom unchanged The grating, for ever

The three figures form a progression. The sailor reads from outside. The fisherman reads from inside and changes his own relation to the system. The Mariner acts from inside and changes the system's constraints. A reflexive anthro-complexity would take the fisherman's detachment within involvement and the Mariner's constraint-changing bricolage, and would leave the sailor's position on deck to the non-reflexive media where it belongs.

Escape, Conquest, Grating

The Compass and the Loom developed the Mariner's lesson into a typology of stances towards automated systems, its Table 24: three stances, each defined by the position from which one relates to the system.

Escape is the stance from outside: refusing the system, leaving it, opting out. The escaper does not engage with the whale. She stays out of the sea, or swims away. Escape is sometimes right. Some systems are best refused. But it leaves the whale to eat everyone else, and it is available only to those who can afford to leave.

Conquest is the stance from above: steering the system, mastering it, bending it to one's purposes. The conqueror captures the whale and harnesses it. Conquest is the stance of the steersman, the kybernetes of The Compass and the Loom's critique, who reads the system and sets its course. Its characteristic failure, the treatise noted, is Goodhart's and Lucas's: the steered system learns the steering. A whale harnessed for one purpose learns to serve it in ways that defeat the harnesser's intentions. The conquered system adapts to the conquest.

Grating is the stance from inside, with stakes. The Mariner neither escapes nor conquers. He is inside the whale, and he installs a constraint that changes what the whale can do, without trying to steer it. The whale remains a whale, eating small fish, living happily ever after. It has not been conquered, only constrained. And the constraint protects everyone else in the sea.

The typology has a precedent in political economy. Albert Hirschman's Exit, Voice, and Loyalty (1970) analysed how members of organisations respond to decline. They can exit, leaving for a competitor, or exercise voice, complaining, protesting and trying to change the organisation from within. Hirschman showed that the balance between exit and voice depends on loyalty. Loyal members stay and use voice, and less loyal members exit. Organisations need both. Exit signals failure through the market, and voice signals it through politics. Escape is Hirschman's exit. The grating is something like voice made structural: an intervention from inside, by a member with stakes, that changes the organisation's constraints rather than leaving it. Conquest has no place in Hirschman's typology, because it is not a member's response to the organisation. It is an owner's.

Requisite Friction

The Compass and the Loom defined the grating precisely: a checkpoint where a human must sense before a machine decides. The "suspenders" that tie it in place are trust, informal networks and people who notice. Chapter 24 used the grating as the remedy for the harvest loop, placing human checkpoints between narrative and data, between pattern and action, between model and collection, and between typical and atypical.

The grating has a property that runs against much contemporary design thinking: it introduces friction. Contemporary technology design prizes frictionlessness, removing every obstacle between a user's intention and its fulfilment, every delay, every checkpoint. The treatise's warning is that frictionless adoption removes the gratings for the sake of efficiency, which lets the whale eat everything in the sea. A system without friction is a system without gratings, in which nothing slows the whale's appetite.

The concept can be called requisite friction, after Ashby's requisite variety. W. Ross Ashby's law of requisite variety (1956) holds that a regulator must have at least as much variety, as many possible states, as the system it regulates, if it is to control it. Only variety can absorb variety. Requisite friction is the corresponding principle for systems whose appetite must be limited. A system must contain at least as much friction as is needed to prevent its appetite from consuming what it should not. Friction is not a defect to be engineered away. It is a design feature that protects what lies outside the system's proper scope.

Mesh Size

The grating's crucial design parameter is its mesh size, the size of the openings through which things can pass. The Compass and the Loom stated the trade-off: too wide and the system swallows everything, too fine and it starves, which echoes the Zeno effect. A grating with openings too wide is no grating. The whale swallows men through it. A grating with openings too fine starves the whale. Nothing passes, and the system cannot function. The right mesh size lets through what the system should take in and stops what it should not.

The echo of the Zeno effect is precise. A grating too fine is like monitoring too frequent. It inhibits the system's normal functioning, freezing it, as continuous measurement freezes a quantum transition. A human checkpoint at every step of a process, with every narrative reviewed, every pattern checked and every decision approved, would make the process too slow to work. The grating must be placed where it matters and sized so that it stops what matters, and it must let the rest through.

Mesh size cannot be specified in advance. It depends on the system, on what it should and should not take in, and on the costs of friction and of its absence. It is a matter of judgement: the phronesis and mētis of Element 11, the Mariner's infinite resource and sagacity. And it must be revisited as the system changes. A mesh size right for one stage of a system's life may be wrong for another.

Mesh sizes for narrative instruments

Chapter 24 proposed four gratings for narrative instruments. Each has a mesh-size question.

Grating Too wide Too fine Judgement needed
Narrative → data (consent to harvest) All narratives become training data by default Every use requires fresh individual consent, and nothing can be learned Which uses need fresh consent; which are covered by the original
Pattern → action (return first) Patterns acted on before narrators see them Every pattern must be ratified by every narrator before action Which patterns need return before action; how long return takes
Model → collection (protect the tails) Model's expectations shape signifiers and flagging unchecked No model allowed to influence anything Where the model's influence is checked; how tails are monitored
Typical → atypical (route outliers to humans) All outliers handled by the model Every narrative read by a human Which outliers are routed; how many humans; how fast

The Suspenders

Kipling tells his Best Beloved not to forget the suspenders, and The Compass and the Loom took the instruction seriously. The suspenders are what tie the grating in place: trust, informal networks and people who notice. A grating without suspenders falls out. The whale coughs it up, and the system returns to eating everything.

The suspenders are the social infrastructure that makes structural constraints hold. A rule that no one trusts is evaded. A checkpoint staffed by people who do not notice is passed without a check. A consent procedure that no one believes in is a formality. The grating, the formal constraint, holds only if it is tied in place by the informal fabric of trust, relationship and attention.

This bears on anthro-complexity's instruments directly. Every grating proposed in this disputation (consent to the harvest, return before action, protection of the tails, routing outliers to humans, mutual dashboards, declination tables, the reading of the reader) is a formal constraint. Each will hold only if it is tied in place by suspenders. Narrators must trust that consent means something. Analysts must notice the outliers. Leadership must actually return patterns before acting. The convenors' own significations must be read and not merely recorded. A reflexive anthro-complexity cannot be achieved by design alone. It requires the suspenders: the cultivation of trust, the maintenance of informal networks, and the presence of people who notice. These are not instruments. They are the social conditions within which instruments can be honest.

That returns the disputation to CRP (Chapter 11), which has always insisted that the quality of relationships, the conversation, the trust, the attention to what is happening in the room, is the medium in which any practice succeeds or fails. CRP has the suspenders and few gratings. Anthro-complexity has the gratings, or could have, and needs the suspenders. The Mariner needed both. The grating was made from his raft, and the suspenders were his own.

The Mariner's Limits

The Mariner is an attractive figure, and it is worth noting his limits, since this disputation has asked anthro-complexity to notice the limits of its own figures.

The Mariner acts on a non-human system. The whale is not a human system. It does not tell stories about the grating, form coalitions against it, or redescribe its own appetite. The Mariner's grating works because the whale is, in Hacking's terms, an indifferent kind. It does not respond to being constrained as a human system would. In a human system a grating would be read, contested, gamed, circumvented and renegotiated. The workforce would find ways round the checkpoint, the data team would lobby to widen the mesh, and leadership would read the grating as an obstacle to efficiency. A grating in a human system is a gesture, and its meaning is completed by the response. The Mariner's story, like the estuary, is an image from a non-reflexive medium, and its lesson must be carried into human systems with the reflexivity the story does not contain.

The Mariner acts alone. He has no colleagues inside the whale, no one to consult, and no one whose stakes differ from his. In a human system the person inside is one of many, and their stakes differ. A grating that serves the Mariner's interest (not being eaten) may not serve the interests of others inside (the small fish who are still eaten). The design of gratings in human systems is a political question. Whose interests does this constraint protect, and whose does it sacrifice? That brings back CST's participants dimension (Chapter 12) and Ulrich's boundary questions.

The Mariner leaves. His intervention is permanent, and he is not there to see its consequences. A practitioner who installs a grating and leaves is not accountable for how it works out. The whale lives happily ever after, Kipling says, but we have only the 'Stute Fish's evidence for that, and the 'Stute Fish has hidden in the mud. A reflexive practitioner stays to see what her gratings do, or arranges for someone to.

These limits do not undo the Mariner's lesson. He remains the best available figure for the observer inside with stakes, who changes constraints by bricolage. They show that the lesson must be completed by what this disputation has drawn from Mead, Elias and CST: reflexivity, plurality and politics.

The Mariner and the Theses

Thesis The Mariner's contribution
II: asymmetry applied to itself The Mariner is inside, with stakes; his intervention is from within, not from Wigner's outside
III: humans are not estuaries The Mariner against the sailor: inside the medium, changing it, not reading it from deck
IV: constraints The grating is a constitutive constraint, a change in what the whale is
VI: computation Grating, mesh size, requisite friction; the human checkpoint before the machine decides
VIII: degradation Conquest fails by Goodhart and Lucas; the steered system learns the steering

The adversary will say that a children's story is a slender basis for a theory of practice, and that the Mariner's lesson, that one should change constraints from inside using what is to hand, is anthro-complexity's own lesson (exaptation, constraint-shifting, probe-sense-respond), restated with a whale. Snowden has preached exaptation for twenty years. The Mariner adds nothing but charm.

The adversary is right that the Mariner restates much of anthro-complexity's own practice, and the chapter acknowledges that exaptation is Snowden's concept. What the Mariner adds is position. He does all this from inside, with stakes. Anthro-complexity's exaptation is often presented as a practitioner's technique, applied to an organisation by someone who sees what can be repurposed. The Mariner's exaptation is the act of someone whose survival depends on it, inside the system, with nothing but what is to hand. The difference is the difference between the sailor and the fisherman. The adversary has the method right and the figure wrong. partial

NEITHER ESCAPE NOR CONQUEST —

A GRATING IN THE THROAT, TIED WITH SUSPENDERS —

AND YOU MUST NOT FORGET THE SUSPENDERS, BEST BELOVED.

Addition: Gratings Already in the World

The grating is not a new invention. Human institutions are full of gratings: checkpoints where a person must sense before a process proceeds, constraints installed from inside to limit what a system can swallow. This addition collects examples from several fields, to show that the principle is familiar and that its failures are familiar too. Each example is described in general terms, as an illustration of the principle and not as an authoritative account of any particular regime.

Examples of gratings

The four-eyes principle. In banking, payments above a threshold require approval by two people. In aviation, critical actions are cross-checked by both pilots. In clinical practice, high-risk medications are checked by two nurses. The second pair of eyes is a grating. It stops what one person alone would pass, and its mesh size is the threshold above which the check applies.

The human in the loop for automated decisions. Data-protection law in the United Kingdom and the European Union has long given individuals rights in relation to decisions based solely on automated processing that produce legal or similarly significant effects, including, in broad terms, the right to obtain human intervention. That is a grating at the point where a machine decides about a person. Its mesh size is the definition of "significant effects".

The works council. In several European countries, works councils elected by employees must be informed and consulted, and in some matters must consent, before management makes certain changes affecting employees. The works council is a grating at the point where organisational decisions meet those affected. It is installed from inside, by employees, and it limits what management can do without engaging them.

Ethics review. Research involving human participants typically requires approval by an ethics committee before it begins. The committee is a grating between the researcher's design and the participants. It checks consent, risk and the protection of the vulnerable.

The duty to consult. Many public bodies are required to consult those affected before making certain decisions. Consultation is a grating, or at least it is meant to be. Whether it functions as one depends on whether the consultation can change the decision.

The receiver's duties. In the receivership of Chapter 46, the receiver's duty of good faith to the borrower, and her duty to manage with due diligence if she manages, are gratings installed by law. They limit what the receiver can do in the lender's interest without regard to the borrower.

How gratings fail

The examples also show how gratings fail, and the failures fall into types the Mariner's story helps to name.

Mesh too wide: the grating that does not stop. A four-eyes check in which the second person approves without examining, rubber-stamping, is a grating with mesh so wide it stops nothing. Ethics review that approves designs without engaging their substance is the same. The grating exists formally and does not function.

Mesh too fine: the grating that starves. A four-eyes check applied to every transaction, however trivial, slows the bank to a halt. Ethics review so demanding that low-risk research cannot proceed starves inquiry. Consultation requirements so extensive that no decision can be made paralyse the public body. The grating functions and the system cannot eat.

Suspenders missing: the grating that falls out. A works council that management routinely ignores, or whose members fear retaliation, is a grating without suspenders. Its formal powers exist and trust does not. Consultation in which the decision has already been made, and is announced as consultation, is a grating that the system has coughed up while keeping its shape for show. The Mariner's instruction applies: do not forget the suspenders. Trust, informal networks and people who notice are what make gratings hold.

Grating captured: the grating that serves the whale. A grating can be captured by the system it was meant to constrain. A second approver who is the first approver's subordinate, an ethics committee dominated by the researchers it reviews, or a consultation run by the body whose decision it scrutinises are all gratings turned. The checkpoint exists, and it serves the appetite it was meant to limit.

Failure types

Failure Mariner's terms Example Remedy
Rubber stamp Mesh too wide Second approver signs without looking Make the check substantive; rotate checkers
Paralysis Mesh too fine Every trivial action needs approval Set thresholds by judgement; revisit them
Ignored Suspenders missing Works council overruled routinely Cultivate trust; protect those who use the grating
Captured Grating serves the whale Checker subordinate to the checked Independence; checkers from outside the line of command

Gratings for anthro-complexity's instruments

Chapter 24 proposed four gratings for narrative instruments: consent to harvest, return before action, protection of the tails and routing outliers to humans. The examples above show what each will need to work.

  • Consent to harvest will fail as a rubber stamp if consent is a checkbox no one reads. It needs plain language, separate consent and a real option to refuse.
  • Return before action will fail as consultation-after-decision if patterns are returned after leadership has already acted. It needs a stated interval between return and action, and a mechanism by which narrators' responses can change the action.
  • Protection of the tails will fail if captured, for example if the people checking the tails report to those who want the typical patterns. It needs independence.
  • Routing outliers to humans will fail by paralysis if every outlier must be read in full, and by rubber stamp if the humans skim. It needs a mesh size set by judgement and revisited.

None of these failure modes is unique to narrative instruments. They are the ordinary ways gratings fail in every institution. A practice that installed gratings in its instruments would need to learn from the institutions that have installed gratings before, and from their failures.

FOUR EYES — HUMAN IN THE LOOP — WORKS COUNCIL — ETHICS REVIEW —

GRATINGS FAIL BY RUBBER STAMP, PARALYSIS, NEGLECT AND CAPTURE —

LEARN FROM THE GRATINGS ALREADY IN THE WORLD.

Addition: Cross-Lights From Book Nine

The walk. David Hockney photographed the garden at Ryōan-ji while moving and assembled the prints into one image with many vanishing points. The Zen Garden study takes it as the remedy for a limited seat. It is the mariner's situation seen from the other side: no chart, and a body that moves. Mapping from several positions is the walk, and a set of maps laid together is a joiner. No joiner is the garden.

The node. Foulkes's conductor, in Chapter 51, is another figure of the practitioner inside: one node in the web, tending the medium.

Chapter 42

Crowds and Leviathan: Where Only Humans Go

4,334 words, about 19 minutes of reading

Where Only Humans Go

Chapter 9 compared the starling murmuration with the flash mob and found the difference between them. The murmuration has distributed, unreflexive agency. The flash mob has reflexive agency, because its participants know they are a mob. It also found something the murmuration cannot show: the seed of an institution. A flash mob has an organiser, and once someone speaks for it, it carries the seed of something larger. This chapter follows that seed. It draws on The Compass and the Loom's Part VIII, on crowds and Leviathan, and argues that the decisive difference between human and non-human collectives is not that humans have richer capacities, but that human collectives can be represented. Someone can speak for them, and the speaking makes them one. Representation is the asymmetry principle at the scale of the collective, and it is where the human distinction meets the state.

The chapter takes, in turn, Hobbes's Leviathan read as a physics of motion and a theory of representation; Canetti's crowds and crowd crystals; Clastres's societies against the state; and the six thresholds by which a flash can become a Leviathan. It then asks what all this means for anthro-complexity's instruments, which aggregate the narratives of populations and present them, through dashboards, to those who act on the populations' behalf.

Hobbes as Physicist

Thomas Hobbes's Leviathan (1651) is usually read as political philosophy: the argument that, to escape the war of all against all, people must authorise a sovereign to act for them, and obey. The Compass and the Loom read it also as a physics, an ontology of motion. Hobbes was a mechanist, a friend of Galileo's admirers and a correspondent of Mersenne's circle, and Leviathan begins not with politics but with sense, imagination, speech and the passions, all of which Hobbes explains as motions.

Several of Hobbes's concepts, read as motion, bear on this disputation.

Endeavour (chapter 6). The small beginnings of motion within the body, before they appear in visible action: appetite towards and aversion from. Endeavour is the infinitesimal of motion, the propensity before the act. It is Hobbes's 勢, an incipient tendency, not yet an action.

Power (chapter 10). A man's present means to obtain some future apparent good. Power is not a substance. It is a relation to the future, the means one has now towards what one may get later. And it is relational: a man's power is measured by comparison with the power of others, and his worth is his price, what would be given for the use of his power, which is a judgement made by others.

War as disposition (chapter 13). War, Hobbes says, consists not in actual fighting but in the known disposition to fight, during all the time there is no assurance to the contrary. He compares it to foul weather. Foul weather is not a shower or two of rain but an inclination thereto of many days together. War is a disposition, a propensity, not an event. That is anthro-complexity's dispositional state (Element 4), in the language of 1651.

Representation (chapter 16). This is the most important of the concepts for this chapter. A person is one whose words or actions are considered as his own, or as representing the words or actions of another. When a person acts for others, by their authority, he represents them, and they are the authors of his actions. Hobbes then makes his decisive claim. A multitude of men are made one person when they are by one man, or one person, represented. It is the unity of the representer, not the unity of the represented, that makes the person one. A multitude has no unity of its own. It becomes one only by being represented by one.

Money as blood (chapter 24). The circulation of money through the commonwealth is like the circulation of blood through the body, nourishing every part. The commonwealth lives by circulation, and dies when it stops.

Three ontologies of the Leviathan

The Compass and the Loom drew three ontologies of the commonwealth from these concepts. The Leviathan is dynamic, a resultant of composed endeavours, the summed motions of its members. It is energetic, a store of conferred power, the power its members have authorised it to use. And it is kinetic, a circulation that dies when it stops, like blood and money. The treatise connected the kinetic reading to Thomas Nail's Being and Motion (2018), whose kinetic ontology describes the world in terms of flow, fold and field, so that the commonwealth is a field.

The frontispiece

The treatise also drew on Horst Bredekamp's study of the Leviathan's famous frontispiece, in which a giant crowned figure rises over a landscape, holding sword and crozier, his body composed of hundreds of tiny people. In the printed edition of 1651, the tiny people face inward, towards the sovereign's body, their backs to the viewer, looking at the head. In the manuscript copy Hobbes presented to the future Charles II, Bredekamp reported, the tiny figures are faces looking outward, towards the viewer. The treatise read the difference as two ontologies of the institution. In one, the people are absorbed into the sovereign, facing the head that represents them. In the other, they are the sovereign's surface, facing the world, the sovereign's eyes.

Representation as the Asymmetry Principle

Hobbes's chapter 16 is the asymmetry principle at the political scale. The November post's principle is that only one of the actants, the planners, can change the story it tells about the other. Hobbes's principle is that a multitude becomes one only by being represented, and it is the representer who makes it one. The representer tells the story of the multitude: who they are, what they want, what they authorise. In doing so the representer makes them into the subject of the story. The multitude, unrepresented, has no single story. It is many people with many stories. Represented, it has one story, the representer's, and it is one person.

This is the re-entered asymmetry of Chapter 18 at its sharpest. Among humans, some represent and others are represented. The representers tell the story of the represented, and the story makes the represented one. The represented cannot change the representer's story as easily as the representer can change theirs, because the representer speaks with the unity the representation confers, and the represented speak as many.

And this is what no non-human collective can do. A starling flock cannot be represented. No starling speaks for the flock, and the flock is not made one by being spoken for. Its unity, such as it is, is the unity of coupling, the topological rule of six or seven neighbours. An ant colony cannot be represented. The queen does not speak for it. She lays eggs. Human collectives can be represented, and the capacity for representation, someone speaking for many and the speaking making them one, is the human collective's distinctive property. Only the human collective can become a Leviathan, The Compass and the Loom concluded, and a flash mob carries the seed of one once someone speaks for it.

Canetti: Crowds and Crystals

Elias Canetti's Crowds and Power (1960) is a strange, brilliant, unclassifiable book, a phenomenology of crowds drawn from anthropology, history, myth and Canetti's own experience of crowds in Vienna and Frankfurt in the 1920s. Two of its concepts matter here.

The discharge. The crowd, Canetti argued, comes into being at the moment of discharge, when the differences between its members are shed and they feel themselves to be equal. Before the discharge there are many people, each with distinctions of rank, property and role. At the discharge, the distinctions fall away, and everyone feels the crowd's density and equality as release. The fear of being touched, which Canetti held to be fundamental to individual life, is reversed in the crowd. Each person presses against the others, and the pressure is relief.

The crowd crystal. Crowds do not form from nothing. They precipitate around crowd crystals: small, rigid, constant groups of people, defined and bounded, who serve as the nucleus around which a larger crowd forms. A band of musicians, a group of soldiers, a congregation's choir or a party cell are crowd crystals. They are small and disciplined, they persist when the crowd disperses, and they can call a crowd into being again. Crowds are transient, and crowd crystals endure.

The Compass and the Loom applied the crowd crystal to Snowden's own concern about entrainment, his charge that groups fall in behind the first confident voice. The treatise read the first confident voice as a crowd crystal, the seed of representation. A group in a workshop, a room in a facilitated session or a population in a SenseMaker study begins as many people. The first confident voice offers a nucleus, a definite position around which others can gather. The others gather, the room becomes a crowd around the voice, and the voice now speaks for it.

The treatise drew a further, sharper point. An undeclared form lets whoever holds it become the crowd's representative without authorisation. A crowd crystal that does not announce itself, the facilitator whose framing shapes what is said, the designer whose signifiers shape what is signified, the analyst whose patterns shape what is seen, can precipitate a crowd without anyone noticing that it has done so, and then speak for the crowd without anyone having authorised it to. Freeman's tyranny of structurelessness (Chapter 23) is the same point. In a group without declared structure, an undeclared structure forms, and it is unaccountable.

Clastres: Societies Against the State

Pierre Clastres, a French anthropologist who worked among peoples of the South American lowlands, argued in Society Against the State (1974) that the societies anthropologists had called "stateless" did not lack a state, as if they were primitive versions of state societies that had not yet developed one. They actively warded it off. Their institutions were designed to prevent the emergence of a separate political power. The chief, in Clastres's account, had prestige but no power. He was obliged to be generous, to speak eloquently and to keep the peace, but he could not command. If he tried to, he was abandoned. The chief's speech, Clastres observed, was often delivered to no one in particular and heeded by no one. It was a ritual of speaking that displayed the chief's lack of power. Society kept the chief, and with him the possibility of representation, in check.

The Compass and the Loom summarised: societies do not lack a State; they actively ward it off. The concept matters here because it suggests that the move from crowd to Leviathan, from many to one represented, is not inevitable. It can be warded off by institutions designed to prevent the representer from becoming the sovereign. The chief who must give away everything, the speech that commands no one and the abandonment of the chief who tries to rule are gratings (Chapter 41) on the throat of representation.

Six Thresholds

The Compass and the Loom traced six thresholds by which a momentary crowd can become a state. Each is a bifurcation, a point at which the collective can go one way or another, and each has a possible warding-off mechanism.

Threshold What happens Warding-off
1. Flash Many people come together, briefly, around a shared script Disperse quickly (Wasik's ten-minute rule)
2. Refrain The flash is repeated; it becomes a pattern, a ritual Vary the script; do not let it become routine
3. Crystal A small, constant group forms that calls the flash into being Rotate the organisers; declare who they are
4. Representation Someone speaks for the collective; the many become one Bind the speaker; make speech accountable; speech that commands no one (Clastres)
5. Capture The representer appropriates the collective's power and surplus (Deleuze and Guattari's apparatus of capture) Generosity obligations; the chief gives everything away
6. Leviathan The representer becomes sovereign; the collective is a state Exit; voice; the war machine exterior to the State (Deleuze and Guattari)

The treatise drew on Deleuze and Guattari's account of the State as an apparatus of capture, which springs up rather than growing gradually, appropriating the surplus of what it captures, and on their concept of the war machine, the nomadic, exterior force that the State tries to capture and that resists capture. It drew on Max Weber's routinisation of charisma, the process by which a charismatic leader's personal authority becomes institutional authority after his death or departure, which is thresholds 3 to 5. It drew on Victor Turner's communitas, the bond of the liminal crowd, which ritual can host only by fencing it in. And it drew on Critical Mass, the monthly cycling events that began in San Francisco in 1992, described by participants as an "organised coincidence", a crowd that recurs without organisers.

The treatise also proposed rules for anyone who convenes collectives, among them Rule 12: write the dispersal rule before convening. Before bringing people together, decide how they will be released, so that the convening does not slide, threshold by threshold, into capture.

The Flash Mob as Fruit Fly

The treatise defended the flash mob, against the objection that a ten-minute prank is trivial, as the fruit fly of institutions. Geneticists study the fruit fly not because it matters in itself but because it is small, short-lived and simple enough that processes common to all living things can be observed in it quickly. The flash mob is the institutional fruit fly. It passes through the first thresholds (flash, refrain, perhaps crystal) in minutes or weeks, so the dynamics of collective formation, representation and capture can be seen in miniature. Wasik's own description of the project as an experiment in conformity, and his decision to stop it when it began to be captured by marketers and media, is a case study in thresholds and their warding-off.

The Dashboard as Leviathan's Face

What does this mean for anthro-complexity's instruments?

A narrative instrument deployed across a population gathers many stories from many people and aggregates them into patterns. The patterns are presented, through a dashboard, to those who act on the population's behalf: leadership, a central function, a government department. In Hobbes's terms the dashboard represents the population. It makes the many into one, a pattern, a picture of "what our people are experiencing", and the representation is what leadership acts on.

The dashboard is therefore at threshold 4. It is a representation of a collective, and it makes the collective one by representing it. The question Hobbes's analysis forces is: who is the representer, and by what authority?

The narrators did not authorise the dashboard to represent them, in Hobbes's sense of making them its authors. They shared stories. They did not appoint the dashboard as their person. The analyst who reads the patterns and the leadership who acts on them speak for the population, as the representer speaks for the multitude, but without the authorisation that, in Hobbes's theory, makes representation legitimate. In the printed frontispiece's terms, the narrators face inward, their backs to the viewer, absorbed into the body of a representation they see only from behind.

The consequences follow from the thresholds.

The representation can become capture (threshold 5). The patterns, once aggregated, can be used for purposes the narrators did not intend: restructuring, performance management, training automated systems (Thesis VI's harvest). The surplus of the narrators' noticing is appropriated by the representer. That is Deleuze and Guattari's apparatus of capture, in miniature.

The representation can become sovereign (threshold 6). A dashboard that is accepted as the picture of the organisation, the single source of truth (Chapter 39), becomes the organisation's authoritative self-description. What it shows is what the organisation is. What it does not show is not. It has become, in a small way, a Leviathan, an artificial person made of the population's stories, speaking for them and governing what they can be.

The crowd crystal is undeclared (threshold 3). The designers of the signifiers, the convenors of the study and the analysts of the patterns are the crowd crystal around which the representation forms. If their role is not declared, as The Compass and the Loom warned, an undeclared form lets whoever holds it become the crowd's representative without authorisation.

Warding Off the Leviathan

Clastres's societies warded off the state by binding the chief: obliging him to generosity, making his speech command no one, abandoning him if he tried to rule. What would warding off look like for narrative instruments?

Declare the crystal. The designers, convenors and analysts are named, and their role in shaping the representation is stated. This is declination before direction and the reading of the reader.

Bind the representation. The dashboard does not speak for the population without the population's authorisation. Patterns are returned to narrators before they are used (return as design requirement), and narrators can contest, qualify or refuse them. The representation is a proposal, not a verdict.

Oblige generosity. The representer gives back what it takes. Narrators receive the patterns, the analysis and the decisions made from them, and share in the benefits of what their stories produce (consent to the harvest, a share in the outcome).

Write the dispersal rule. Before a study is convened, state how it will end: when the data will be deleted, when the patterns will be retired, and when the representation will cease to speak for the population. That is Rule 12, and it is also Thesis VIII's dated instruments.

Keep exit and voice open. Narrators can leave the study (exit), and can speak against its representation (voice), and both are honoured.

Prefer the outward-facing frontispiece. Design the representation so that the people in it face outward, so that their stories are visible as their own and not absorbed into an aggregate whose face is someone else's. Mutual dashboards, in which each group sees its own patterns and the others', are the outward-facing frontispiece.

The Three Levers

The Compass and the Loom's Movement 34 proposed that leverage in collectives needs three things: agency (the act), adjacency (the neighbourhood the act reaches) and tendency (the lean of the field). Each alone fails. The treatise mapped them onto Snowden's own triad of agency, affordance and assemblage. It kept agency, read affordance as a face of adjacency, and replaced assemblage with tendency, arguing that an assemblage without a lean is "a diagram, not a motion". It noted that most interventions push only one: campaigns push agency, reorganisations push adjacency, and strategy decks describe tendency.

The three levers bear on representation. A representer who has all three (the agency to speak, the adjacency of a dashboard that reaches everyone, and the tendency of an organisation already leaning towards whatever the dashboard shows) has great leverage. That is the leverage of the Leviathan. A practice that wards off the Leviathan distributes the levers, so that narrators too have agency (they can speak back), adjacency (they can reach one another, not only the centre) and a share in shaping the tendency (they can say which way the organisation should lean). Hobbes's commonwealth concentrates the levers in the sovereign. Clastres's society disperses them.

The Human Distinction, Politically

The chapter can end by stating the human distinction in political terms.

The political form of the human distinction. Human collectives differ in kind from non-human collectives because they can be represented: one can speak for many, and the speaking makes them one (Hobbes). Representation is the asymmetry principle at the scale of the collective. The representer tells the collective's story, and the represented cannot easily change it. Only human collectives can pass through the thresholds from flash to Leviathan, and only human collectives can ward the Leviathan off (Clastres). Anthro-complexity's instruments, which aggregate populations' narratives into patterns that speak for them, stand at the threshold of representation. A reflexive anthro-complexity would declare its crowd crystals, bind its representations, oblige generosity, write its dispersal rules and keep exit and voice open.

The adversary will say that this chapter inflates a dashboard into a state. A SenseMaker study is not a Leviathan. It has no sword, levies no taxes and commands no armies. Its "representation" of a population is a picture, not a sovereign. To invoke Hobbes, Canetti, Clastres and Deleuze is to dress a methodological question in apocalyptic clothes.

The adversary is right that a dashboard is not a state, and the chapter does not claim it is. It claims that the dashboard passes through some of the same thresholds as a state (crystal, representation, capture), in miniature, and that the thresholds are worth knowing because they are where the risks lie. The flash mob is not a state either. It is the fruit fly, in which the thresholds can be seen. The dashboard is another fruit fly. The apocalyptic clothing is the adversary's description. The chapter's is the more modest claim that whoever speaks for many, by whatever means, should know what Hobbes, Canetti and Clastres knew about speaking for many. partial

A MULTITUDE IS MADE ONE BY ONE WHO REPRESENTS IT —

THE CROWD CRYSTAL MUST BE DECLARED —

WRITE THE DISPERSAL RULE BEFORE YOU CONVENE.

Addition: The Six Thresholds as a Convener's Check

The six thresholds from flash to Leviathan can be read as a check for anyone who convenes people: a workshop, a sense-making session, a SenseMaker study, a community of practice. At each threshold the convener can ask whether the gathering is crossing it, whether it should, and what warding-off mechanism is in place. The check is offered as a set of questions, not a procedure.

Threshold 1: Flash

People come together around a shared prompt or task.

  • What brought them here, and did they choose to come?
  • Is there a stated end, a time when the gathering disperses?
  • If the gathering is a study, when does collection close?

Warding off: a stated end. Wasik's ten-minute rule, translated: say when it stops, before it starts.

Threshold 2: Refrain

The gathering is repeated; it becomes a pattern, a ritual.

  • Is this the first session, or the fifth? If the fifth, has its form become routine?
  • Are the same prompts, signifiers or exercises used each time?
  • Has the gathering acquired a name, a schedule or a place in the calendar that makes it expected?

Warding off: vary the form. Rotate the frames (Thesis VIII). Ask, at each repetition, whether the gathering still serves those who come, or has become a habit maintained for its own sake.

Threshold 3: Crystal

A small, constant group forms that calls the gathering into being.

  • Who convenes? Is it always the same people?
  • Are the conveners named, and is their role in shaping the gathering stated?
  • Could someone outside the convening group call a gathering, or propose a different form?

Warding off: name the crystal and rotate it. Declare who convenes, why and with what stake. Let others convene.

Threshold 4: Representation

Someone speaks for the gathering; the many become one.

  • Who reports what the gathering found? To whom?
  • Did those gathered authorise the report, or see it before it was delivered?
  • Does the report present one view, or the range of views expressed?

Warding off: bind the representation. Return the report to those gathered before it goes elsewhere. Show perspectives, not a verdict (Chapter 39). Let the gathered contest it.

Threshold 5: Capture

The representer appropriates the gathering's surplus: its ideas, its stories, its energy.

  • What happens to what the gathering produced? Who uses it, for what?
  • Do those who produced it share in what it produces?
  • Has any of it become training data, intellectual property or a product, without their knowledge?

Warding off: oblige generosity. Consent to the harvest. Return the surplus: patterns, analysis, benefits. Write down who owns what was produced.

Threshold 6: Leviathan

The representation becomes authoritative; it governs what the gathered can be.

  • Has the gathering's output become the account of the organisation, the single source of truth?
  • Are decisions made on its basis that the gathered cannot contest?
  • Has the convening group become an institution with authority over those it convenes?

Warding off: keep exit and voice open. Retire the representation on a stated date. Make sure the gathered can leave, can speak against the account given of them, and can call a different gathering.

The check as a whole

Threshold Question Warding-off
1 Flash When does it end? State the end before starting
2 Refrain Has it become routine? Vary the form; rotate frames
3 Crystal Who always convenes? Name and rotate the conveners
4 Representation Who speaks for it, with whose authority? Return first; perspectives; contest
5 Capture Who takes what it produced? Consent; return the surplus
6 Leviathan Does its account now govern? Exit, voice, retirement

Not every gathering should be stopped at the first threshold. Some should become refrains, some should crystallise, and some should be represented. A community of practice that meets every month for years, convened by a stable group that speaks for it to the organisation, has crossed thresholds 2, 3 and 4, and may be all the better for it. The check does not forbid crossing. It asks that each crossing be noticed, and that the warding-off mechanism for the next threshold be in place before it is reached. Most gatherings that end as Leviathans did not mean to. They crossed thresholds without noticing, because no one asked the questions.

FLASH — REFRAIN — CRYSTAL — REPRESENTATION — CAPTURE — LEVIATHAN —

NOTICE EACH CROSSING —

AND HAVE THE NEXT WARDING-OFF READY BEFORE YOU ARRIVE.

Addition: Cross-Lights From Book Nine

The party called "the system". Snowden's formative encounter is stated as a meeting of practitioner and system. Chapter 50 applies Hobbes to it. A system of many people becomes one party to an encounter only by being represented, and the unity is supplied by whoever speaks for it. In practice that is the commissioning few.

Impersonal fellowship. De Maré's word for the climate of a large group, discussed in Chapter 57, describes a human gathering that is neither crowd nor Leviathan: strangers thinking together without a representative.

Chapter 43

The Case for the Other Side: The Adversary's Whole Case

3,665 words, about 16 minutes of reading

The Rule of the Respondent

The first rule of the disputation, stated in the Preliminaries, is that the adversary's case must be stated at its strongest before it is answered. Throughout Books II to VII, boxes marked The Case for the Other Side have conceded to the adversary what could be conceded, thesis by thesis. This chapter gathers the adversary's case into one continuous argument, the case a thoughtful defender of anthro-complexity, perhaps Snowden himself, might make against this whole book. It is written in the adversary's voice, as the scholastic respondent was required to restate the opponent's argument to the opponent's satisfaction. The author's replies follow, and then a ledger of what the adversary wins.

The case is stated as strongly as the author can state it. If it is not strong enough, that is the author's failure, and a reader who can make it stronger is invited to.

The Adversary's Case

What follows is the adversary speaking.

Article 1. The theory is relational already.

You accuse anthro-complexity of a possessive fallacy, of placing the human distinction in properties persons have, not in relations. But read the whole corpus, not one post. The tacit–explicit–entangled trialectic places knowledge in interaction. Crews are contextual identities formed around tasks. The critique of codification, from the first knowledge-management papers onward, insists that knowledge cannot be extracted from the relations in which it lives. The asymmetry principle is about the relation between the planners and the river. Disintermediation is about the relation between experts and people. Every element of anthro-complexity that matters in practice is relational. You have built a thesis on the grammar of three sentences in a single post, written for a particular argument against Ing, and ignored twenty years of practice.

Article 2. The grammar is shorthand.

"Humans hold multiple identities" is how anyone would say it in a blog post. It is not a metaphysical commitment to the closed person. Elias wrote "the individual" too, and no one accuses him of possessive individualism. You have mistaken a figure of speech for an ontology.

Article 3. The instruments were designed against the outside position.

SenseMaker's self-signification exists precisely so that the analyst cannot impose meaning. Estuarine mapping is done by participants. Collective categorisation exists so that no expert decides what kind of situation a group is in. Ritual dissent exists so that the first confident voice does not dominate. Every instrument you criticise for reinstating the outside observer was designed to remove one. You describe what the instruments are designed against and call it what they are.

Article 4. The estuary is a heuristic.

No practitioner thinks an organisation is an estuary. The image draws attention to constraint energetics and timing, which it pictures well. Its limits are known. Snowden has spent his career warning against reifying metaphors. To hold a heuristic responsible for what it does not picture is to confuse a picture with a model.

Article 5. Power is everywhere in the practice.

Power is not missing from anthro-complexity. It is in the critique of the systems-change industry, in the warning about whoever defines balance, in the concern with entrainment and in every method designed to counter the first confident voice. Refusing to make power the master concept is a considered choice, not an omission. Theories that make power central see nothing else, and their practitioners become critics who cannot act.

Article 6. Scale requires the individual.

You want relational data: the addressee of every story, the others in it, the participatory sense-making that preceded it. That cannot be collected from ten thousand people. At scale, the individual narrative is the only unit that can be sampled. Your relational instruments would confine complexity practice to the small group, which is exactly CRP's limitation, and you concede the limitation in Chapter 11.

Article 7. Your alternative is CRP, and CRP cannot act.

Strip away the Chinese philosophy and the quantum physics, and your reconstruction is CRP with a theory of constraint borrowed from Juarrero. CRP has had twenty-five years to show what it can do in large organisations and governments, and it has produced reflective practitioners and doctoral theses. Anthro-complexity has produced instruments used in counter-terrorism, public health, international development and corporate strategy across the world. You ask the school that acts to adopt the philosophy of the school that reflects.

Article 8. Reflexivity is a regress.

You demand that every instrument account for its own effects, that the practitioner be included in what she senses, that the designer declare her expectations and that the effects of the declarations be tracked. Where does it stop? Every account of an instrument's effects is itself an intervention whose effects must be accounted for. Structural reflexivity, taken seriously, is an infinite regress that leaves nothing done. Probe-sense-respond is the practical answer to the regress: act, watch and adjust. It does not need your apparatus.

Article 9. The author is reflexive, and the instruments evolve.

You concede that Snowden is personally reflexive, revising in public, holding points loosely and inviting correction. You say this is the reflexivity of an author, not of an instrument. But instruments are made by authors, and they evolve. Cynefin has been revised many times, SenseMaker has been redesigned repeatedly, and estuarine mapping is new and still developing. The reflexivity you want in the instruments is happening, through the author's reflexivity, over time. You are criticising a snapshot of a moving practice.

Article 10. You are standing outside.

Finally, this book criticises anthro-complexity for keeping the observer outside, from outside. You are not inside the Cynefin Company. You have not designed or deployed its instruments, sat in its workshops or worked with its clients. You read a blog post and a brochure and built 200,000 words on them. Your disputation is Wigner outside the lab, describing the friend's experiment as a superposition. It is the scholastic fallacy, Bourdieu's skholè, projected onto practitioners who work in the urgency you have the leisure to reflect on. You accuse anthro-complexity of the god trick while performing it.

Article 11. Instruments must be sold.

The Cynefin Company is a business. Its instruments must be sold to clients, and clients buy what solves their problems. A Pelagian practice that sold nothing would help no one. The drift you trace from ontology to product is the price of reaching people. You have the luxury of an imprint that sells nothing and of a disputation that asks nothing of a client. The practice you criticise has to survive in a market.

Article 12. The human distinction does its job.

You concede that the human distinction is true, that its five reasons hold, and that humans are not ants has done real good. Everything else in this book is a demand that the distinction do more, be stated differently and be applied further. But a distinction that is true and has done good has done its job. The rest is your agenda, not anthro-complexity's failure.

Here the adversary ends.

The Replies

The adversary's case is strong, and parts of it are decisive. The replies follow article by article, and they concede where concession is due.

To Article 1: the theory is relational already

Conceded in large part. Much of anthro-complexity's practice and corpus is relational. The trialectic, crews, the critique of codification, disintermediation and the asymmetry principle all point that way, and this book has said so (Chapters 4, 13, 19). The disputation's claim is narrower than the adversary states. It is that the human cluster, the definition of what makes anthro-complexity anthro, as stated in the November post at the moment it defines the school against a neighbour, uses a possessive grammar. That grammar pulls the instruments' design towards collecting what individuals have (Chapter 18). The claim is about the self-definition, not the whole practice. partial

To Article 2: the grammar is shorthand

Partly conceded. Grammar is often shorthand, and the November post is a post, not a treatise. But this book's claim is that the grammar is consistent across the human cluster, the 3Is of The Compass and the Loom and the reply to Barad ("the people remain distinct"). A consistent grammar at the moment of self-definition is evidence of an ontology. Elias wrote "the individual", but he also wrote What Is Sociology? to argue against the closed person. Snowden's equivalent argument against the possessive person has not been made, and the November post's rejection of Barad suggests the opposite. partial

To Article 3: the instruments were designed against the outside position

Conceded as to intent; maintained as to structure. This is the adversary's strongest article, and it is true that the instruments were designed against the outside observer. Self-signification, participatory mapping, collective categorisation and ritual dissent are real achievements. The disputation's claim, from Chapter 6 onwards, is that intent sits inside structure. Someone still designs the signifiers, holds the dashboard, receives the patterns first and decides what the study is for (Theses II, V, VII). The instruments remove the outside observer from the act of interpretation and leave her in the frame. partial

To Article 4: the estuary is a heuristic

Conceded as to the heuristic; maintained as to the method. The estuary is a heuristic, and its limits are known. The thesis (Chapter 21) concerns the method built on it, estuarine mapping, which does not map the mapping's effects or the mapper's waiting. Pepper's point is that root metaphors carry their structure into methods whether or not their users intend it. partial

To Article 5: power is everywhere in the practice

Conceded as to practice; maintained as to theory. Anthro-complexity's practice is power-aware, more than most. The danger of power as master concept is real. The thesis (Chapter 23) asks that power be a dimension of the theory of the human, not its master concept, so that the instruments represent it and do not rely on practitioners' vigilance alone. partial

To Article 6: scale requires the individual

Largely conceded. This is the adversary's most practical article, and it wins much. The individual narrative is the natural sampling unit at scale, and relational data is costly. The disputation's reply (Chapters 19, 31) is that the individual may be the sampling unit without being the definition of the human. Light additions (the addressee, the others in the story, the narrator's reading of the pattern) can capture something of the relation at low cost, and Hutchins's overlapping horizons are achievable at scale through mutual dashboards. But the adversary is right that a fully relational instrument cannot be deployed to ten thousand people, and that the disputation's reconstruction must accept approximation. holds for the adversary, with the disputation's qualification.

To Article 7: your alternative is CRP

Rejected. The reconstruction is not CRP. It keeps anthro-complexity's instruments, its theory of constraint, its scale and its capacity to act, and adds to them what CRP, Elias, Mead, Peirce and the Chinese sources supply: relational persons, the observer inside, power as dimension, reflexive design. Chapter 11 accepted anthro-complexity's charge against CRP, that it is thin on instruments and cannot work at scale. The reconstruction is anthro-complexity with its own premises followed through, not CRP with borrowed instruments. Book Eight must show this, and the reader will judge. fails for the adversary.

To Article 8: reflexivity is a regress

Partly conceded. Full self-accounting is impossible, and the regress is real (Chapter 20). The disputation's reply is that its demands are finite: four structural additions (record descriptions, record effects, include the teller, declare position), not an infinite accounting. Probe-sense-respond is a partial answer to the regress, as the adversary says. But it probes the system's response to the experiment, not to the framing, so the descriptions that license the probes go unprobed. The finite additions extend probe-sense-respond to the descriptions. They do not replace it with a regress. partial

To Article 9: the author is reflexive, and the instruments evolve

Conceded as to evolution; maintained as to direction. The instruments do evolve, and Snowden's reflexivity drives the evolution. The disputation criticises a snapshot, and the snapshot will be outdated. But Chapter 6's genealogy traced the direction of the evolution, from the listener towards the steersman, and the direction matters more than the snapshot. If the instruments evolve towards the reconstruction this book proposes, the book will have been outdated in the best way. If they continue towards the steersman, its criticisms will stand. partial

To Article 10: you are standing outside

Conceded, and the concession is the most important in this chapter. The author is outside the Cynefin Company. He has read a post, a brochure, a blog and the published corpus. He has not designed SenseMaker studies or facilitated estuarine mapping sessions. The disputation is a description from a position, and it has the limits of its position. It cannot see what practitioners inside the practice see: the workshops in which the instruments are used with care, the clients whose populations are given the patterns first, and the adjustments made in the room that no brochure records. Wigner outside the lab describes a superposition, and the friend inside may have seen something definite that Wigner's description misses.

The concession does not refute the disputation, any more than Wigner's position refutes his description. Both descriptions, outside and inside, are correct relative to their positions. But it does require two things of the disputation. First, it must declare its position, which the Preliminaries and this chapter do: a critic outside, working from published texts. Second, it must invite the inside view, which this chapter now does. The author would welcome correction from practitioners inside, who can say where the instruments, as used, already do what this book asks, and where they do not. That is the reading of the reader applied to the disputation. Without the inside view, the disputation's description is Wigner's. With it, the two descriptions can be compared at the rendezvous, and the comparison is the inquiry. holds for the adversary, and the disputation accepts it.

To Article 11: instruments must be sold

Conceded as fact; not as justification. The Cynefin Company is a business, and its instruments must be sold. The author's imprint sells nothing, and this asymmetry of positions is real and should be stated. But the positional reason (Chapter 17) was offered as explanation, not accusation. The drift is the price of reaching people, as the adversary says. The disputation's claim is that the price includes a loss, the human moving from the centre to the periphery, and that the loss can be reduced without abandoning the market. A Pelagian practice that sells gratings, mutual dashboards and consent to the harvest is still a practice that sells. partial

To Article 12: the human distinction does its job

Rejected. The human distinction is true and has done good, and the disputation has said so repeatedly. But a distinction whose own logic, the reflexive reason, the asymmetry principle and Peirce, implies conclusions the school has not drawn has not finished its job. The disputation's agenda is the distinction's own, followed through. The adversary's Article 12 says the distinction should stop where it is. The disputation says the distinction does not stop there, and that this is what Snowden means when he says Peirce is still live. fails for the adversary.

The Ledger

The replies can be gathered into a ledger of what the adversary wins, what the disputation keeps, and what is shared.

Article Adversary's claim Verdict What the adversary wins What the disputation keeps
1 Theory relational already partial Much of the practice and corpus is relational The self-definition is possessive, and pulls design
2 Grammar is shorthand partial Posts are not treatises The grammar is consistent at self-definition
3 Instruments designed against outside partial The intent is real; self-signification, participation and ritual dissent are achievements Intent sits inside structure; the frame remains outside
4 Estuary is heuristic partial Its limits are known The method built on it does not cover its limits
5 Power in practice partial Practice is power-aware; master concept is a danger Power belongs in the theory as dimension
6 Scale requires individual holds (adversary) Relational data is costly; full relational instruments cannot scale Sampling unit is not definition; light relational additions are feasible
7 Alternative is CRP fails (adversary) — Reconstruction keeps instruments, constraint, scale
8 Reflexivity is regress partial Full self-accounting is impossible Four finite additions extend probe-sense-respond
9 Instruments evolve partial The book criticises a snapshot The direction of evolution matters
10 You are outside holds (adversary) The disputation is Wigner; it cannot see inside practice Both descriptions are correct relative to position; the inside view is invited
11 Instruments must be sold partial The market is real; the author's imprint sells nothing The loss can be reduced within the market
12 Distinction does its job fails (adversary) — The distinction's own logic leads further

The adversary wins two articles outright (6 and 10), loses two (7 and 12), and wins half of the other eight. That is a substantial win. The disputation's eight theses survive, but each is qualified by the adversary's case. The qualifications are part of what the disputation now holds.

Amendments Under the Covenant

The Additive Covenant requires that verdicts revised in light of the adversary's case be amended in place, with the earlier judgement kept visible. The following amendments are made to the theses of Book Four.

Thesis I (possessive fallacy). Earlier verdict: holds. Amended: holds as a claim about the human cluster's self-definition and its pull on instrument design. It does not hold as a claim about anthro-complexity's practice as a whole, which is substantially relational (Article 1). The scale objection (Article 6) limits the remedy to approximation.

Thesis II (asymmetry applied to itself). Earlier verdict: holds. Amended: holds, with the recognition that the disputation is itself subject to it (Article 10). This book is a constraint-changing story, told from outside, and its effects, if any, are not the author's to determine.

Thesis III (humans are not estuaries). Earlier verdict: holds. Unamended, with Article 4's acknowledgement that the heuristic's limits are known to practitioners.

Thesis V (power). Earlier verdict: holds. Amended: holds as a claim about the theory. As a claim about practice it is substantially met by anthro-complexity's power-aware methods (Article 5).

Thesis VII (triad as simplex). Earlier verdict: partial. Unamended.

Thesis VIII (degradation). Earlier verdict: holds. Amended: holds as a prediction. Its testing requires data held inside the practice, which the disputation does not have (Article 10). It is offered as a hypothesis to practitioners, not as a finding.

The Shared Ground

When the ledger is drawn, the adversary and the disputation share more ground than either began with.

  • Human complexity differs in kind from other complexity. Humans are not ants.
  • The difference is grounded, at least in part, in reflexivity: human systems are changed by their descriptions.
  • The outside observer is a problem, and anthro-complexity's instruments were designed, with real success, to reduce it.
  • Full reflexivity is impossible, and every practice must act with partial self-knowledge.
  • Scale imposes approximations that pure theory does not.
  • Power matters, and must be attended to without becoming the only thing attended to.
  • The practice evolves, and the evolution is driven by the reflexivity of those who practise it.
  • Every description, the disputation's included, is from a position, and the comparison of positions is the inquiry.

The last point is the most important, and it is the adversary's gift to the disputation. Article 10 is right. The disputation stands outside, and its description is one of two. The other, the inside view, is what practitioners of anthro-complexity know about their instruments in use. The disputation cannot supply it. It can only invite it. That invitation is now the disputation's central request. Like Snowden's at the head of the November post, it is meant: if I have misread, I would welcome the correction.

THE ADVERSARY WINS HALF —

AND THE HALF IT WINS IS THE OUTSIDE FROM WHICH THIS BOOK WAS WRITTEN —

THE INSIDE VIEW IS INVITED, AND THE RENDEZVOUS IS THE INQUIRY.

Addition: Four Further Articles, and the Whole Case in Cards

Book Nine adds four articles to the adversary's case. They are stated here in brief and argued in the chapters named.

Article 13. The trialectic already says it.

The formative encounter was published five months before this book was written. A practitioner with a stable outside view treats the complex as complicated. Capacity is held by neither party. The practitioner's formation is part of the work. Reply: credited in Chapter 48. The question becomes whether the instruments embody the norm, and Chapter 50 argues that they realise it in inverted form.

Article 14. The originator speaks first.

Self-signification gives the teller the first attribution and keeps it on record. That is epistemic justice built into a method. Reply: the right is real. The prompt and the signifiers were written before the teller spoke, and a record answers to her only if she can consult it. See Chapter 52.

Article 15. Formation is a priesthood.

The process school's remedy requires years in a community under supervision, which is the Augustinian model in its purest form. Reply: the point shows that neither school escapes a mediating figure. Method and formation are complements, and the formed must stay answerable to the unformed. See Chapters 51 and 57.

Article 16. Compression by intervention is what power does.

To say that compression alters complexity is to license making people regular. Reply: the objection strengthens the ethics. Whoever compresses answers for what people are made into. See Chapter 56.

Article 17. Participation is only another frame.

Someone decides who takes part and what is asked, so the critic's answerability is one more model. Reply: accepted nearly in full. An arrangement in which those affected can answer a frame still corrects itself by their action, which an attitude does not. See Chapter 58.

Chapter 44

Principles of a Reflexive Anthro-Complexity

4,224 words, about 18 minutes of reading

Book Eight: Reconstruction

The disputation has stated anthro-complexity's claim (Book One), placed it among its neighbours (Book Two), asked why it treats human complexity differently (Book Three), advanced eight theses against it (Book Four), brought the corpus to bear (Book Five), read it through 道理勢 (Book Six) and taken up the observer inside, the Leviathan and the adversary's whole case (Book Seven). Book Eight reconstructs. It does not ask anthro-complexity to give up anything it has rightly won. It asks it to finish its own thought, following its premises to the conclusions they imply.

This chapter states the principles of a reflexive anthro-complexity. Chapter 45 applies them to the instruments. Chapter 46 tests them on an illustrative case from the receiver's desk. Chapter 47 is the Coda.

The reconstruction is offered under the adversary's Article 10. It is written from outside the practice, by a critic who has not deployed the instruments, and it is a proposal, not a prescription. Practitioners inside will know where it is naive, where it is already practised and where it would not work. It is offered to them as a gesture whose meaning will be completed by their response.

What Is Kept

The first task of reconstruction is to say what is kept. That is nearly everything. Of the twenty-two elements of Chapters 2 to 5, all but three are kept as they stand, and the three are amended, not abandoned.

# Element Status Amendment, if any
1 Complexity is a claim about the world Kept Realism about interactive kinds (Hacking): real, and responsive to description
2 Coherence found, not presumed Kept, and extended Applied to persons: persons too are found, not presumed
3 Actants, not variables Kept, amended Actants as knots in a reflexive meshwork; constructors replaced by practices that change their performers
4 Dispositional states and vectors Kept The practitioner's disposition is part of the state
5 Far from equilibrium; no normal Kept The power critique of norms is extended to the instruments
6 The line is ordered/complex Kept The human/non-human line is drawn within the complex by descriptive reflexivity
7 Propensity over causality Kept, completed 勢 includes the strategist and the political 勢 of position
8 Coherence only in retrospect Kept Every present reconstructs its past (Mead); the instrument is part of the present
9 Propensity in complex systems as such Kept Human propensity is reflexive
10 Kairos Kept, completed Waiting is a gesture; the tide may turn because one waited
11 Phronesis and mētis Kept Mētis protected from legibility; the knife-holders consulted
12 Three marks of the human Amended Restated relationally: identity as position, intention as shared, abstraction as accepted
13 Asymmetry principle Kept, and turned on itself Applies to anthro-complexity's own descriptions; re-enters among humans
14 Entanglement without flat ontology Kept, refined Persons distinct as produced patterns (Elias, Juarrero), not as presumed containers
15 Against the programme Kept Extended to practice: the grating; protection of the tails
16 Peircean pragmatism Kept, completed The whole Peirce: the interpretant with the community; the self as sign
17 Trialectic Kept, refined Mediation as practice (和); the human as mediator (三才); instruments that do not impose closure
18 Behaviour as shadow of conditions Amended Change the conditions with the knowledge and participation of those whose conditions they are
19 Distributed sensing Amended Distribute the frame as well as the act; mutual horizons
20 Pelagian stance Kept Tested by recognition: we did it ourselves
21 Sequence depends on system Kept, extended Add CST's participants dimension: unitary, pluralist, coercive
22 Resistance to commodification Kept, and turned inward Applied to anthro-complexity's own instruments

The table is the reconstruction's first claim. Reflexive anthro-complexity is anthro-complexity. Nineteen of its twenty-two elements are kept, several extended or completed from their own sources, and three amended. None is abandoned. The amendments are to the human cluster's grammar (Element 12), the shadow principle's ethics (Element 18) and distributed sensing's reach (Element 19). They are exactly where the disputation's eight theses converged.

What Is Added

The second task is to say what is added. Twelve principles follow. They are grouped under four headings: ontology, epistemology, practice and politics. Each is stated, grounded in the chapters that support it, and connected to the theses it answers.

Ontology

Principle 1. The human is a pattern of relation, distinct as produced. Persons are not presumed parts bearing identities, intentions and abstractions. They are persisting patterns of context-sensitive constraint (Juarrero), social processes of I and me (Mead), nexuses in figurations (Elias), knots in a meshwork (Ingold), signs in dialogue (Peirce) and mediators between (三才). They are fully real and distinct, but their distinctness is produced in relation, not given before it. Answers Thesis I. Grounded in Chapters 13, 19, 27–32, 38.

Principle 2. The human difference of kind is descriptive reflexivity. Human systems differ in kind from other complex systems because they are changed by public descriptions of themselves, which their members take up and act on, and because they can be represented, made one by one who speaks for them. The three marks (multiple identities, intention, abstraction) are the capacities that make descriptive reflexivity possible. The difference of kind is in the reflexive relation, not in the capacities. Answers Theses I, II. Grounded in Chapters 8, 14, 42.

Principle 3. Constraints include the constitutive, and they close. Beside governing and enabling constraints there are constitutive constraints, which make a community what it is, are carried in habitus and practical consciousness, and are reproduced in compliance and in mapping. Constraints maintain one another in closure, so energy cost is a property of position, not of the constraint. Practices change their performers, so everyone is a reagent. Answers Thesis IV. Grounded in Chapters 22, 29, 33, 36.

Epistemology

Principle 4. The observer is inside, and her description is one of many. There is no view from nowhere, only the god trick (Haraway). Every description of a human system is from a position: Wigner's or the friend's, the sailor's or the fisherman's. Descriptions from different positions can each be correct and still disagree. The practitioner's description is one among the system's many, and it is itself a move in the system. Answers Theses II, III. Grounded in Chapters 20, 21, 28, 39.

Principle 5. Meaning is completed by response; the interpretant belongs to the community. The meaning of a story, a pattern or a categorisation is not fixed by its author or analyst. Others complete it by how they respond (Mead), and it is approached as a final interpretant only through the community of inquiry, in which those described are inquirers too (Peirce). An instrument that captures meaning at one point and interprets it at another has broken the community of interpretation. Answers Theses I, VII. Grounded in Chapters 25, 27, 32.

Principle 6. Asking changes the asked. Every act of sensing a human system changes it. The order of questions matters, frequent monitoring freezes change, saturating expectation selects cliché, a reading shows where or how but not both, alignment cannot be broadcast, and light touches disturb less. Instruments are probes, not cameras. Answers Thesis VIII. Grounded in Chapters 26, 34, 40.

Practice

Principle 7. Declare: the reading of the reader, declination before direction. Every instrument declares its position. Its designers state what they expect to find, what they hope to find and what they would prefer not to find. Its convenors signify their own experience, published alongside everyone else's. Its intended performativity, what it is meant to do to the system, is stated. Its crowd crystal is named. Answers Theses II, V. Grounded in Chapters 20, 33, 34, 42; the Tidal Loom.

Principle 8. Make the horizons overlap: mutual dashboards, return before action. Those whose narratives produce the patterns see the patterns at the same time as those who act on them, and before action is taken. Leadership's readings and decisions are visible to the population and open to response. Hutchins's bridge, not the panopticon. Answers Theses V, VI. Grounded in Chapters 23, 31, 32; the Tidal Loom.

Principle 9. Grate the harvest; protect the tails; sense lightly. Human checkpoints stand between narrative and data, pattern and action, model and collection, and typical and atypical. Their mesh size is set by judgement and revisited. Atypical narratives are routed to humans. The missing mass is estimated. Monitoring is less frequent than the change hoped for, and pockets of low monitoring are protected. A portfolio of light touches stands beside the strong instruments. Answers Theses VI, VIII. Grounded in Chapters 24, 26, 40, 41.

Principle 10. Date the instruments; rotate the frames; track the loop. Every signifier set, map and framework has an expiry. Frames rotate so that no single set of categories loops into self-description. Reflexive content, stories about the instrument's own previous effects, is tracked separately and read as wake, not water. The type of performativity (generic, effective, Barnesian, counter) is tracked over time. Answers Thesis VIII. Grounded in Chapters 26, 33, 34.

Politics

Principle 11. Power is a dimension: track the gradient, distribute the felicity conditions. Power is in the theory of the human as a dimension, not as master concept. Every deployment asks CST's participants question (is this situation unitary, pluralist or coercive?) and Ulrich's boundary questions (who is affected but not involved?). It tracks where exposure lands on the power gradient, and whose anxiety its probes create. It distributes the authority to make performatives stick, as well as the locution. Answers Thesis V. Grounded in Chapters 12, 23, 28, 34.

Principle 12. Consent, dispersal, recognition. No narrative becomes training data without informed, separate and revocable consent and a share in the outcome. Every convening writes its dispersal rule before it begins. The test of every intervention is recognition, not alignment: whether, when the work is done, the people say we did it ourselves. The light touch is declared, visible and contestable, so that recognition is true and not the effect of an invisible hand. Answers Theses V, VI. Grounded in Chapters 16, 37, 42; the Tidal Loom.

The Principles and the Theses

Thesis Principles that answer it
I: possessive fallacy 1, 2, 5
II: asymmetry applied to itself 2, 4, 7
III: humans are not estuaries 4 (and the figure of the fisherman, the Mariner, the calligrapher)
IV: under-socialised constraint 3
V: power 7, 8, 11, 12
VI: computation 8, 9, 12
VII: triad as simplex 5 (and Chapter 45's Peircean triad)
VIII: degradation 6, 9, 10

Every thesis is answered by at least one principle, and every principle answers at least one thesis. The reconstruction is the disputation's affirmative side. The theses showed where anthro-complexity's premises lead further than its practice goes, and the principles say what the practice would look like if it went there.

The Figure of the Practitioner

Every body of thought has a figure of its practitioner, an image of the person who does the work. The disputation has examined anthro-complexity's figures: the listener of Stage One, the leader of Stage Two, the analyst at the dashboard of Stage Three, the reader of terrain of Stage Four, and the steersman of Stage Six. It has also examined its governing image, the sailor in the estuary.

The reconstruction proposes a different figure, assembled from the figures the disputation has found in its sources.

Source Figure What it contributes
Elias The fisherman in the maelstrom Detachment achieved within involvement; reads from inside, binds himself to what descends slowly
Kipling The Mariner inside the whale Stakes; changes constraints from inside by bricolage; the grating and the suspenders
Sunzi The general within the 勢 Is read as he reads; his visibility is a force; deception and empty and full
Calligraphy The calligrapher Inside the stroke; responsive to material; cannot plan the detail; the trace is read afterwards
Zhuangzi Cook Ding The grain known by the knife; slows at the knot; the blade stays sharp
Ingold The wayfarer Knows the country along the way, not from the chart
Hutchins The navigator on the bridge Overlapping horizons; distributed cognition; the shared chart
Wigner The friend inside the lab The situated, definite, inside view
Laozi 17 The ruler barely known Action that does not force; we did it ourselves, provided she is still seen

The figure is the practitioner inside. She is in the system she works in, with stakes and seen by those she sees, and her detachment is an achievement of involvement, not a privilege of position. She changes constraints from within by working with what is to hand, along the grain, slowing at the knots. She knows the organisation as a wayfarer knows country, shares her chart with the crew, and declares her position. When her work is done, the people say they did it themselves, and they are right, because her touch was light, declared and refusable.

The practitioner inside is not a new figure in anthro-complexity. She is the listener of Stage One, restored, with the instruments and the theory of constraint of the later stages in her hands.

Cynefin Reconsidered

The reconstruction has a consequence for Cynefin itself, the framework from which anthro-complexity grew. It does not require Cynefin to be abandoned or redesigned. It requires two additions and a change of posture.

Addition one: the participants dimension. Cynefin's domains are defined by the relation of cause and effect. CST's system of systems methodologies (Chapter 12) adds a second dimension, the relations among participants: unitary, pluralist or coercive. A reflexive Cynefin would ask both questions. What is the relation of cause and effect here? What are the relations of power among those affected? The decision model would then depend on both. Probe-sense-respond in a unitary complex situation, where participants share purposes, differs from probe-sense-respond in a coercive complex situation, where some can impose on others. In the coercive case the probes need safeguards: who designs them, who bears their risk and who sees their results.

Addition two: the categoriser in the category. Every Cynefin categorisation is a performative (Chapter 34), a status-function declaration (Chapter 13) and a move in the field (Chapter 20). A reflexive Cynefin would record each categorisation as an act: who categorised, with what authority, and with what consequences for whose licences. It would treat the categoriser's position as part of the situation categorised. The aporetic domain at the centre of the framework, where it is not clear what kind of situation one is in, already acknowledges the difficulty of categorisation. A reflexive Cynefin would add that the categoriser is always partly in that domain, because her categorising is part of what makes the situation what it is.

The change of posture. Cynefin has often been read, especially in its HBR form, as a framework for leaders: a tool with which a leader diagnoses her situation and chooses her response. Snowden has resisted that reading, insisting on collective sense-making. A reflexive Cynefin would complete the resistance. The framework would be a shared chart on the bridge, a means by which everyone in the situation can see what kind of situation it is for them, compare their readings and find, at the rendezvous, where they agree and where they differ. It would not be the leader's diagnostic instrument but the crew's shared surface.

Estuarine Mapping Reconsidered

Estuarine mapping, the most developed recent method, is reconsidered in detail in Chapter 45. In outline, a reflexive estuarine map would keep the axes of energy and time and add:

  • a third axis, hysteresis (Bourdieu): how far the habitus carrying a constraint lags behind its formal conditions;
  • closure links: which constraints maintain which, so that energy is read as a property of position;
  • constitutive constraints: the deference, doxa and habitus reproduced in the mapping session itself;
  • dated limits: every "cannot be changed" with a date, a condition and the name of whoever judged it so;
  • the mappers on the map: whose constraints these are, who bears the cost of shifting them, who has the power to shift them;
  • the wayfarer's reading: where the knife-holders say the grain actually runs, as against where the map shows it.

The map would become a wayfarer's map. It would be a record of where people have been and found their way, made by them and shared among them, not a navigator's chart drawn from above.

SenseMaker Reconsidered

SenseMaker is reconsidered in Chapter 45. In outline, a reflexive SenseMaker would keep self-signification, micro-narratives, triads and scale, and add:

  • the address: to whom the story is told, and who else is in it;
  • independent scales beside the triads: so that strong-in-all and weak-in-all are distinguishable;
  • a proposed fourth vertex: what is missing from the signifier set;
  • mutual dashboards: narrators see the patterns when, or before, leadership does;
  • the narrator's reading of the pattern: the second-order interpretant, recorded;
  • order randomisation: signifier order varied and recorded;
  • dated signifier sets: retired before they become self-description;
  • gratings at the harvest: separate, revocable consent; atypical stories routed to humans; tails protected;
  • the reading of the reader: convenors and designers signify too.

Is It Still Anthro-Complexity?

The question the reconstruction must face is whether what it proposes is still anthro-complexity, or whether, as the adversary's Article 7 charged, it is CRP with borrowed instruments.

The answer depends on what makes anthro-complexity what it is. If anthro-complexity is defined by its ontology (complexity as a claim about the world, coherence found, constraints, dispositions, propensity, retrospect), the reconstruction keeps it whole. If it is defined by its instruments (Cynefin, SenseMaker, estuarine mapping), the reconstruction keeps them and extends them. If it is defined by its scale and capacity to act, the reconstruction keeps both, with approximations the adversary rightly demanded. If it is defined by its human distinction, the reconstruction keeps it and grounds it more deeply, in descriptive reflexivity and representation, where the difference of kind is secure against the CAS theorist's reply.

What the reconstruction changes is the grammar of the human cluster, the ethics of the shadow principle and the reach of distributed sensing. These are anthro-complexity's own concepts, followed through. The reconstruction takes from CRP its relational account of the human, and from Elias his account of power. It does not take CRP's refusal of instruments, its analogical relation to complexity science or its scale. It is anthro-complexity with its premises followed through, and if that makes it closer to CRP than it was, the closeness is the premises' doing.

The adversary will say that the reconstruction is a list of good intentions, each of which costs time, money and attention that practitioners do not have. Every principle adds a step, a record, a consultation or a delay. A SenseMaker study with mutual dashboards, return before action, independent scales, a fourth vertex, order randomisation, dated signifiers, gratings at the harvest and the reading of the reader would cost twice as much and take twice as long, and the client would buy a cheaper instrument from a less scrupulous vendor. The reconstruction is a counsel of perfection that, if adopted, would price anthro-complexity out of its market.

The adversary wins a real half. The principles cost something, and the market is real. But not all principles cost equally. Some are nearly free: declaring the designers' expectations, recording signifier order, dating signifier sets, adding "who else is in this story?" to the prompt. Some cost more: mutual dashboards, return before action, gratings. Practitioners would choose among them according to the situation, as Cynefin teaches them to choose decision models. The reconstruction is not a mandatory protocol. It is a set of principles from which a practice can select, guided by the participants dimension. In coercive situations, where the costs of omission are highest, more principles are needed. In unitary situations, fewer. The adversary has priced the full set. The practice would buy what the situation requires. partial

NINETEEN ELEMENTS KEPT — THREE AMENDED — NONE ABANDONED —

TWELVE PRINCIPLES ADDED —

THE PRACTITIONER INSIDE IS THE LISTENER OF STAGE ONE, RESTORED.

Addition: The Reconstruction on One Page

For practitioners who want the reconstruction in a form they can carry, the following summary condenses Chapters 44 and 45. It is offered as a card, not a protocol. The protocols are in Chapter 45, and the reasons are in the rest of the book.

Before you begin

Ask the participants question. Is this situation unitary, pluralist or coercive? Who is affected but not involved? Let the answer decide how much of what follows you need.

Declare yourself. Write down what you expect to find, what you hope to find and what you would prefer not to find. Name who commissioned the work and why. Name the convenors and designers, the crowd crystal. Say what the work is meant to do, not only what it is meant to reveal.

Write the dispersal rule. Say how and when the work will end, the data be deleted, the patterns retired and the representation stop speaking for anyone.

Separate the consents. Consent to share is not consent to harvest. If any story will train any model, ask separately, make it revocable and say what the narrator receives in return.

While you work

Ask about the address. To whom would this story usually be told? Who else is in it?

Let them name what is missing. Offer a fourth vertex. Record what people say your signifiers leave out.

Measure strength as well as proportion. Beside every triad, independent scales. Where tension matters, ask whether the elements pulled together or apart.

Randomise and record order. Questions change the questioned.

Signify yourself. Convenors and designers tell and place their own stories, shown beside everyone else's.

Record who placed what. On every map, every judgement of what is fixed carries a name and a date.

Ask the knife-holders. Those who work the system daily know where the grain runs. Put their reading on the map in its own colour.

Ask about hysteresis and closure. Is this still how people behave, or only how the rules say they should? What holds this constraint in place?

When you have results

Return before action. Those whose stories produced the patterns see them first, or at least at the same time as those who will act on them.

Ask them what the patterns mean. Their reading is an interpretant too. Record it.

Show perspectives, not a verdict. Patterns for each group, side by side. Divergence is the finding.

Make your reading visible. What leadership concluded, and why, is shown to everyone, who may respond.

Route the outliers to people. Stories in the corners are read as stories, not discounted as noise. Estimate what you are not seeing.

Apply the established-and-outsiders lens. Before reporting a pattern as a fact about a group, ask whether it is a trace of a power ratio.

When you act

Change conditions openly. Say what you are changing and why, and ask those whose conditions they are.

Declare your waiting. If you hold back, say so, say why and say what would end it. Your silence is a gesture.

Track the gradient. Who bears the risk of each probe? Whose anxiety does it create?

Prefer gratings to conquest. Change a constraint from inside, with what is to hand. Tie it in with trust.

Probe by asking where you can. The best probe addresses people and does not only condition them.

Over time

Monitor less often than the change you hope to see. Watched pots do not boil.

Protect pockets of low monitoring. The new forms where it is not yet expected.

Rotate the frames. No signifier set should outlive its usefulness. Date it, and retire it on time.

Read the wake. Separate stories about your own previous effects from stories about everything else.

Watch your vocabulary. When a word stops doing work and becomes an excuse, retire it.

The test

When the work is done, will the people say we did it ourselves, and will they be right?

The figure

You are not the sailor on the deck. You are the fisherman in the vortex, the Mariner in the whale, the cook with the knife in the ox and the resident on the street. You are read as you read. Your detachment is something you achieve inside your involvement, not a privilege of where you stand.

When the card becomes rote

Put it down and pick up the knife.

DECLARE — RETURN — GRATE — DATE —

ASK THE KNIFE-HOLDERS — SHOW PERSPECTIVES — DECLARE THE WAIT —

AND TEST BY RECOGNITION.

Addition: A Thirteenth Principle

Book Nine adds one principle to the twelve of this chapter. It is stated in full in Chapter 55.

Principle 13. The practitioner is formed, in company, and returns to give an account. Method does not suffice, and neither does character. Those who design and convene inquiries undergo a continuing formation of perception and judgement among peers. They enter each engagement expecting to be changed, and say afterwards how they were. Answers Thesis IX. Grounded in Snowden's trialectic, in A Web of Relations and in Flinn's account of convening.

Its safeguard is that formation must not become a priesthood. Instruments stay usable by anyone, and the formed are answerable to the unformed.

Two rules of Snowden's are adopted beside it. Every intervention is to be designed as a diagnostic instrument, the instruments included. And every instrument, like every probe, is to carry stated signs of its own failure.

Chapter 45

The Tidal Loom Applied: Redesigning the Instruments

3,505 words, about 15 minutes of reading

From Principles to Protocols

Chapter 44 stated twelve principles of a reflexive anthro-complexity. This chapter turns them into protocols for the three instruments the disputation has examined most closely: SenseMaker, estuarine mapping and the Cynefin sense-making session. It also shows how The Compass and the Loom's Tidal Loom fits around them. The protocols are offered as sketches, under the adversary's Article 10, by someone outside the practice. Practitioners will know which steps are already done, which are impractical and which would need redesign. They are offered as a basis for that conversation.

Each protocol is given in three tiers, following the reply to the adversary at the end of Chapter 44. A light tier uses principles that are nearly free and suits unitary situations, where participants share purposes. A standard tier adds principles of moderate cost and suits pluralist situations, where participants differ but can reach accommodation. A full tier adds the costlier principles and suits coercive situations, where some participants can impose on others and the costs of omission are highest. The tier is chosen by asking CST's participants question at the start of every deployment.

Protocol One: Reflexive SenseMaker

Before deployment

1. The participants question. Before anything else, the convenors, with representatives of the population if possible, ask: is this situation unitary, pluralist or coercive? Who is affected but not involved (Ulrich)? The answer sets the tier.

2. The declination table. The designers complete a table, kept with the study's records and shown to narrators with the results.

Declination Designers' answer
What we expect to find
What we hope to find
What we would prefer not to find
Who commissioned this study, and why
What the study is intended to do (its intended performativity)
What decisions the results may inform
Who will see the results, in what order
What the signifier set was designed to make sayable
What the signifier set leaves out, as far as we know

3. The crowd crystal named. The designers, convenors and analysts are named, with their roles and their relation to the commissioner.

4. The dispersal rule. The study's end is stated before it begins: when collection closes, when the patterns will be returned, when the data will be deleted or archived, and when the signifier set will be retired.

5. Consent tiers. Narrators are offered separate consents: to share the story; to have it aggregated into patterns returned to the population; to have it seen by the commissioner; and, if relevant, to have it used for training any automated system. Each is revocable, with a stated mechanism. If harvest is intended, the narrators' share in the outcome is stated.

The prompt and signifiers

6. The address. The prompt asks, alongside the story: to whom would you usually tell this story? who else is in it? (Bakhtin, Chapter 19; Elias, Chapter 28.)

7. Independent scales beside the triads. For each triad, narrators may also mark each vertex's strength independently, from "not at all" to "strongly". An explicit "none of these" and "all strongly" are offered. (Thesis VII; Chapter 25.)

8. The fourth vertex. After the triads, narrators are asked: is there something your story is about that none of these words name? Responses are recorded as data about the signifier set's agenda. (Thesis V.)

9. Order randomisation. Signifier order is randomised across narrators, and the order each narrator met is recorded. Order effects are reported. (Chapter 40.)

10. The reading of the reader. Convenors and designers share and signify their own stories about the situation, in the same instrument. Their significations are shown alongside the population's. (Principle 7.)

After collection

11. Return before action. Patterns are shown to the population before the commissioner acts, and at least at the same time as the commissioner sees them. The population sees the declination table too. (Principle 8.)

12. The narrators' reading. Narrators are asked, in the return: what do you make of this pattern? does it fit your experience? what is missing? Their readings are recorded as a second-order interpretant. (Peirce, Chapter 32.)

13. Leadership's reading, visible. The commissioner's reading of the patterns, and the decisions made from it, are shared with the population, which can respond. (Hutchins, Chapter 31.)

14. The atypical routed. Narratives in the corners, outliers and singletons are read as stories by people who know the context. They are not discounted as noise. The missing mass is estimated (Good–Turing). (Thesis VI.)

15. The harvest grating. If any automated system is to learn from the narratives, a human checkpoint confirms consent, checks that the tails are protected and records what is passed. (Chapter 41.)

Across rounds

16. Reflexive content tracked. In repeated deployments, stories that concern the organisation's responses to previous rounds are coded separately and read as wake, not water. (Thesis VIII.)

17. Frame rotation. Signifier sets are varied between rounds, and at least some signifiers are replaced each time. (Tidal Loom: frames, not seers.)

18. Retirement. Each signifier set is retired by its stated date, or earlier if its signification has converged (Thesis VIII's prediction 1) or its outliers have thinned (prediction 2).

19. Monitoring interval. Rounds are spaced more widely than the change hoped for, and pockets of the population, topics or times are left unsensed. (Zeno and decoherence, Chapter 40.)

Tiers

Step Light (unitary) Standard (pluralist) Full (coercive)
1 Participants question ✓ ✓ ✓
2 Declination table ✓ (short form) ✓ ✓
3 Crowd crystal named ✓ ✓ ✓
4 Dispersal rule ✓ ✓ ✓
5 Consent tiers Share + aggregate + commissioner + harvest; revocable; share in outcome
6 Address ✓ ✓ ✓
7 Independent scales Optional ✓ ✓
8 Fourth vertex ✓ ✓ ✓
9 Order randomisation ✓ ✓ ✓
10 Reading of the reader Optional ✓ ✓
11 Return before action Same time Before Before, with right to contest
12 Narrators' reading Optional ✓ ✓
13 Leadership's reading visible ✓ ✓ ✓
14 Atypical routed ✓ ✓ ✓
15 Harvest grating If harvest If harvest Mandatory
16–19 Across rounds ✓ ✓ ✓

The light tier is nearly free. Most of its steps are additions to forms and prompts that cost little more than the time to write them. The full tier costs more, and it is reserved for the situations in which, on CST's analysis and this disputation's, it is most needed.

Protocol Two: Reflexive Estuarine Mapping

Estuarine mapping places constraints on axes of energy cost and timescale, so that participants can see which constraints can be shifted, at what cost, on what timescale, and where they have room to act. The reflexive protocol keeps the axes and the participatory practice, and adds the following.

Before the session

1. The participants question and the boundary. As for SenseMaker. Who is in the room, who is affected but not in the room, and is the situation unitary, pluralist or coercive?

2. The convenor's own map. The convenor, before the session, places on a map the constraints she expects the group to name, with her own judgement of their energy and time. The map is sealed and opened at the end, so that the group can compare its map with the convenor's expectations. (Reading of the reader; declination.)

During the session

3. Who placed what. Each constraint placed on the map is tagged with who placed it, and each judgement of its energy cost is tagged with who made it. Where a placement is challenged, the challenge is recorded. Where it is not, the absence of challenge is recorded. (Thesis IV; status characteristics, Chapter 23.)

4. Dated limits. Every constraint placed in the zone of "cannot be changed" is recorded with a date, a condition ("cannot be changed until…" or "cannot be changed unless…") and the name of whoever judged it so. (Tidal Loom: dated limits.)

5. The hysteresis question. For each constraint, the group asks: is this still how people actually behave, or only how the rules say they should? Constraints whose formal conditions have changed but whose habitus persists, and constraints whose habitus has shifted ahead of their formal conditions, are marked. This gives a third axis, hysteresis, beside energy and time. (Bourdieu, Chapter 33.)

6. Closure links. For each constraint, the group asks: what holds this in place? Links are drawn between constraints that maintain one another. The energy cost of a constraint is then read as depending on whether the constraints that hold it can be loosened. (Constraint closure, Chapters 22, 29.)

7. The knife-holders' grain. The group includes, or consults, those who work the system daily, and asks them: where does work actually flow, and where does it stall? where could change pass easily? Their answers are placed on the map in a distinct colour, as the grain, against the formal constraints. (理, Chapter 36.)

8. The mappers on the map. The map includes, as constraints, the positions of the people mapping: whose constraints these are, who would bear the cost of shifting each, and who has the power to shift each. The session's own constitutive constraints (who speaks first, whose judgement stands, what goes without saying) are named, at least briefly, at the end. (Chapters 22, 33.)

After the session

9. The convenor's sealed map opened. The group compares its map with the convenor's expectations. Where they differ, the difference is discussed. Did the convenor's framing shape the session? Did the group see what the convenor missed? (Reading of the reader.)

10. The map as wayfarer's record. The map is presented to the organisation not as a chart of the terrain but as a record of where the group found its way: provisional, dated and made by named people. It is shared with everyone whose constraints it names, and they can respond. (Ingold, Chapter 30.)

11. Revisit. The map is dated and revisited at a stated interval, and dated limits are checked against their conditions. (Dated instruments.)

Tiers

Step Light Standard Full
1 Participants and boundary ✓ ✓ ✓
2 Convenor's sealed map Optional ✓ ✓
3 Who placed what ✓ ✓ ✓
4 Dated limits ✓ ✓ ✓
5 Hysteresis Optional ✓ ✓
6 Closure links Optional ✓ ✓
7 Knife-holders' grain ✓ ✓ ✓, with knife-holders in the room
8 Mappers on the map Brief ✓ ✓, with power to shift named
9 Sealed map opened If made ✓ ✓
10 Map shared ✓ ✓ ✓, with right to respond
11 Revisit ✓ ✓ ✓

Protocol Three: Reflexive Cynefin Session

A Cynefin sense-making session helps a group see what kind of situation it is in and what kind of response the situation calls for. The reflexive protocol adds the following.

1. Two questions. The group asks not only what is the relation of cause and effect here? but what are the relations of power among those affected? Is the situation unitary, pluralist or coercive? (Chapter 44, Addition one.)

2. Categorisation as act. Each categorisation ("this is complex", "this part is complicated") is recorded with who proposed it, who agreed and what licences it creates: to probe, to fail, to proceed without a plan. (Chapter 13; Chapter 34.)

3. Perspectives, not verdict. Where members of the group categorise differently, the differences are recorded as perspectives, not resolved by vote or authority. A situation may be complex for one group and complicated for another, because they stand in different relations to it. (Chapter 39.)

4. The categoriser in the category. The group asks: how does our categorising change the situation? who gains and who loses authority if we call this complex? (Chapter 20.)

5. Probes with safeguards. In complex-coercive situations, every probe is designed with those who will experience it, its risk is tracked along the power gradient and its results are returned to them first. (Chapter 16; Principle 11.)

6. The dispersal of the category. Every categorisation is dated, and revisited. A situation categorised as complex in January may be complicated by June, partly because it was categorised as complex and probed. (Chapter 34: Barnesian performativity.)

The Tidal Loom Around the Instruments

The Compass and the Loom's Tidal Loom was offered as a reconstruction of the two instruments that treatise examined, SwarmCompass and WRAS. Its warp-and-weft structure can also serve as the frame around all three protocols above.

Tidal Loom principle Where it appears in the protocols
Reading of the reader SenseMaker 10; Estuarine 2, 9; Cynefin 4
Declination before direction SenseMaker 2; Estuarine 2
Frames, not seers SenseMaker 17; Cynefin 3
Quorum with inhibition Decision after return: decide by threshold, with stop signals; seek what people can live with
Dated limits Estuarine 4, 11; SenseMaker 18; Cynefin 6
Noticing, not mapping SenseMaker 16, 19; measure latency between novelty and recognition
Risk follows anxiety Cynefin 5; Estuarine 8
Return as design requirement SenseMaker 11–13; Estuarine 10
Consent to the harvest SenseMaker 5, 15
Recognition, not alignment The test of every deployment: do people say we did this ourselves?

The Tidal Loom is the warp, and the three protocols are ways of weaving on it. The convenors lay the warp: the participants question, the declination table, the dispersal rule, the consent tiers. Everyone weaves the weft: the stories, the significations, the placements, the readings, the decisions. The convenors weave too.

Four Instruments Not Yet Built

The protocols above extend existing instruments. The disputation's principles also suggest instruments that do not yet exist, at least in anthro-complexity's repertoire. Four are sketched here, as invitations.

The Rendezvous

The no-signalling law (Chapter 40) says that alignment cannot be broadcast. Harmony is recognised at the rendezvous, when parties who have acted independently come together and compare. A rendezvous instrument would create such occasions deliberately. Teams that have acted independently on a shared challenge would be brought together to compare what they did, see where their actions fit and where they diverge, and recognise, or fail to recognise, a shared direction they did not coordinate in advance. The rendezvous replaces the broadcast of a direction with the discovery of a harmony. Its output is not alignment but recognition: we were going the same way, and did not know it, or we were not, and now we see why.

The Latency Gauge

The Tidal Loom's noticing, not mapping asks that instruments measure the latency between something new becoming actual and someone recognising it. A latency gauge would do this. For each novel pattern detected (in narratives, in practice, in the knife-holders' reports), it would record when the pattern first appeared, in retrospect, and when it was first recognised by the organisation. Rising latency would signal decoherence (Chapter 40) and the thinning of the tails (Thesis VI), the organisation losing its capacity to notice. Falling latency would signal improving attention. The gauge measures the organisation's noticing rather than its state, so it is a path instrument, not a place instrument (Chapter 40: complementarity).

The Established-and-Outsiders Lens

Elias's Established and Outsiders (Chapter 28) showed that patterns of praise and blame gossip reflect power ratios, not facts about groups. An established-and-outsiders lens would be applied to any narrative pattern that distinguishes groups. It would ask whose vocabulary the signifiers reflect, whether one group's stories carry group charisma and another's group disgrace, whether the less cohesive group's counter-stories can be heard, and whether the pattern is a description of the organisation or of one group's description of it. The lens is not a separate instrument but a discipline of reading, applied to every pattern before it is reported.

The Wake Reader

Thesis VIII predicted that repeated deployments would increasingly capture the organisation's response to its own previous readings, its wake. A wake reader would be an analysis, run on every repeated deployment, that separates stories about the instrument's previous effects (the culture programme that followed the last round, the manager who was moved, the sense that "they only act when the numbers look bad") from stories about experience independent of the instrument. It would present the two separately, so that the organisation can see how much of what it is sensing is its own wake. It would also track MacKenzie's performativity types across rounds: is the instrument's fit improving because the organisation is becoming more like the picture (Barnesian), or degrading because the organisation is responding to the picture (counter-performative)?

What the Protocols Do Not Do

The protocols should be read with their limits in view, which are the limits the adversary identified.

They do not remove the outside position. Someone still designs, convenes and analyses. The protocols declare the outside position, make it visible and make its readings contestable. They do not abolish it, because it cannot be abolished. Every practice has convenors. The aim is not a practice without convenors but one in which convenors are inside the cloth, weaving as well as warping.

They do not capture the relational fully. The address, the others in the story and the narrator's reading of the pattern are light approximations of the relational account. They do not capture participatory sense-making, constitutive constraints or the generalised other. Those remain beyond any instrument's reach, and the protocols mark their absence. They do not fill it.

They depend on suspenders. Every protocol step is a grating, and every grating needs suspenders: trust, informal networks and people who notice (Chapter 41). A declination table filled in by rote, a return before action that is a formality, or consent tiers that no one reads are gratings without suspenders, and they will fall out. The protocols cannot supply the suspenders. They can only make clear where suspenders are needed.

They cost something. The light tier is nearly free, and the full tier is not. The choice of tier is a judgement, made by asking the participants question honestly, and it will sometimes be made wrongly, under commercial pressure, in favour of the cheaper tier. The protocols cannot prevent that. They can make the choice explicit, so that its consequences can be seen.

The Instruments and the Figure

Chapter 44 described the figure of the practitioner inside, assembled from the fisherman, the Mariner, the general, the calligrapher, the cook, the wayfarer, the navigator on the bridge, the friend in the lab and the barely known ruler. The protocols are that figure's tools. The fisherman needs to know what descends slowly, which is the latency gauge and the wake reader. The Mariner needs a grating and suspenders, which are the harvest grating and consent tiers, held by trust. The general needs to know how he is read, which is the reading of the reader and the convenor's sealed map. The cook needs to find the grain, which is the knife-holders' reading. The wayfarer needs a record of the way, which is the map as wayfarer's record. The navigator on the bridge needs the shared chart, which is the mutual dashboard. The friend in the lab needs her definite experience to count, which is perspectives, not verdict. The ruler barely known needs to be seen, so that recognition is true, which is declination, the named crystal and the dispersal rule.

The adversary will say that these protocols, however modest each step, add up to bureaucracy. A practice that began as a liberation from frameworks (Cynefin against best practice, narrative against surveys, probes against plans) now has nineteen steps for SenseMaker, eleven for estuarine mapping and six for Cynefin, with tiers and tables and templates. The reflexive anthro-complexity has become the systems thinking it set out to escape: models, boundaries, procedures and a method that can be taught and rolled out. The irony of Chapter 10 has come home.

The adversary wins this half, and it is a deep half. Protocols are structure, structure is what institutions host easily, and a reflexive practice written as protocols risks becoming the next commodity (Element 22). The disputation can only answer that the protocols are warp, not cloth. They are offered to be woven on, varied and discarded when they no longer serve, not to be followed as procedure. That answer depends on the suspenders, practitioners who use the protocols as Cook Ding uses his knife, with judgement, slowing at the knots and changing course when the grain requires. The protocols cannot guarantee that. Nothing can. The best the disputation can do is say plainly, here, that protocols applied without judgement would reproduce the very thing this book criticises. Practitioners who find themselves following these steps by rote should stop, and ask what the situation needs. partial

DECLARE THE CRYSTAL — RETURN BEFORE ACTION — GRATE THE HARVEST —

DATE THE FRAME — ROUTE THE OUTLIER — READ THE WAKE —

AND WHEN THE PROTOCOL BECOMES ROTE, PUT IT DOWN AND PICK UP THE KNIFE.

Addition: Four Lines for the Protocols From Book Nine

From Snowden. Each protocol gains a section headed signs of failure, written before deployment, as his method already requires for a probe. For a narrative instrument the four predictions of Thesis VIII serve.

From Flinn. The declination table gains a line: what the convenor was tempted to do under pressure, and whether she did it. Every invitation to speak carries his caution that speaking is a practical judgement with unforeseeable results.

From de Maré. The return of patterns aims at outsight: a population's recognition of the dominating and silencing habits forming among its members. That is distinct from a sponsor's reading and from private insight.

From the Zen Garden study. Each map is labelled with the seat it was drawn from. Several maps are laid side by side as a joiner, and the joiner is labelled as one assembly among possible others.

Chapter 46

From the Receiver's Desk: An Illustrative Case

4,789 words, about 21 minutes of reading

Why a Case, and Why This One

A disputation about practice should be tested on practice. This chapter tests the principles of Chapter 44 and the protocols of Chapter 45 on a case. It is not a case from the Cynefin Company's practice, which the author has not seen (Article 10), but one from the author's own field: the appointment of a receiver over a charged property, after a borrower's default. The case is a composite. It describes no real receivership, borrower, lender, property, tenant or town, and its details have been invented for the argument. Any resemblance to a particular case is accidental. The legal framework is described in broad terms only, as the setting for the argument. Nothing here is legal advice, and the duties of receivers in any real appointment depend on the instrument of appointment, the law and the facts.

The receivership is chosen for three reasons. First, it is a field in which the author works, and so the one in which the disputation's Article 10 bites least: here the author is inside. Second, a receivership is a human complex system of a sharp and instructive kind. It is a configuration of people with different stakes, unequal power and conflicting stories, into which a new actor, the receiver, arrives with authority and a duty, and whose every move is read. Third, the receivership tests the disputation's central claim directly. It can be read as an estuary, a landscape of constraints to be mapped and navigated by the receiver from the deck. Or it can be read as reflexive human complexity, a configuration in which the receiver is inside, read as she reads, and in which her readings change what she reads. The two readings lead to different practices, and the difference can be seen.

The Setting

In broad terms, when a borrower defaults on a loan secured by a mortgage or fixed charge over property, the lender may, under the powers in the mortgage deed and statute, appoint a receiver. In England and Wales such receivers are often called LPA receivers, after the Law of Property Act 1925, whose provisions give the mortgagee a statutory power to appoint a receiver of income. Wider powers are usually conferred by the mortgage deed, in which case the receiver may be called a fixed charge receiver. The receiver's task is, broadly, to take control of the charged property, collect its income, manage it as the appointment requires and, often, realise it by sale, so that the proceeds go towards the debt.

The receiver's legal position is distinctive and itself a small study in reflexive complexity. By statute and usually by the deed, the receiver is deemed the agent of the borrower, not of the lender who appointed her. The borrower is responsible for her acts and defaults, unless the deed provides otherwise. Yet her primary duty is to bring about the repayment of the secured debt, in the lender's interest. The courts have held that she owes the borrower, and others interested in the equity of redemption, a duty of good faith, and, where she chooses to manage the property, a duty to manage it with due diligence. She is appointed by one party, deemed the agent of another and owes duties to both, and she is read by all of them, and by the property's occupiers, through each of those relations. No outside position is available to her. From the moment of appointment she is inside a configuration that her appointment has changed.

The Case

The property is a mixed-use building on the high street of a market town. It was built in the 1880s as a draper's shop with living quarters above, and was converted in the 1990s. On the ground floor are two retail units. One is let to a long-established independent bakery on a lease with nine years to run. The other is vacant, after a charity shop left eighteen months ago. Above are six residential flats, let on residential tenancies. Three tenants have lived there more than ten years. Two moved in within the past two years, at higher rents. One flat is empty, with damp penetration through the rear wall.

The owner and borrower is a small family property company, run by a widow in her seventies and her son, who took over the day-to-day management five years ago. The company owns four other properties in the town, mortgaged to other lenders. The loan secured on this building, from a regional lender, fell into arrears two years ago, when the charity shop left and the son, under pressure across the portfolio, began to fall behind with repairs and with the mortgage. Repeated forbearance arrangements failed. The lender's credit committee has now appointed a receiver.

The relevant actors, in the vocabulary of Element 3, are many:

  • The lender's credit team, who want the debt repaid, or the loss minimised, and who will read the receiver's monthly reports;
  • The borrower family: the widow, who bought the building with her late husband and regards it as part of the family's history, and the son, who feels he has failed and expects the receiver to sell the building from under him;
  • The bakery, whose owners have traded there for twenty-two years, know every tenant and are the street's informal memory, and whose lease gives them security but whose business depends on the high street's health;
  • The long-standing residential tenants: a retired postman, a nurse and a young family who took over a flat from the wife's mother. Two of them pay below-market rents under an informal understanding with the widow, unrecorded in writing, that rents would rise slowly because the tenants looked after the building;
  • The newer tenants, a couple who work remotely and a student, who pay market rents and have no relationship with the family;
  • The managing agent, a local firm engaged by the son two years ago, who collects rents, has not been paid in full for six months and has a contractor on standby for the damp;
  • The local authority's housing team, which received a complaint about damp from the empty flat's last occupant and has an open file;
  • The building: its Victorian rear wall, a gutter that overflows in heavy rain, and the vacant shop whose shutters have attracted graffiti;
  • The town: a high street with three other vacant units, a council regeneration plan, and local estate agents who know the building and the family.

Reading A: The Estuary

The first reading treats the receivership as an estuary: a landscape of constraints to be mapped and navigated by the receiver, who stands, by virtue of her appointment and her duty, on the deck. It is not a straw man. It is a competent and conventional approach, and in many receiverships it is what happens.

The map. The receiver maps the constraints on axes of energy and time. Low energy, short time: collect rents directly, instruct the managing agent, secure the vacant shop. Medium energy, medium time: remedy the damp, close the housing file, re-let the empty flat at market rent, market the vacant shop. High energy, long time: the bakery's lease, which cannot be changed and constrains any sale to a buyer willing to take it; the informal below-market rents, which are a problem for valuation and may be reviewed under the tenancies; the borrower's other properties and other lenders, which are outside the receiver's control.

The strategy. Stabilise income. Remedy the compliance issue. Re-let to market. Review below-market rents within the tenancies' terms. Market the building as a mixed-use investment with a secure retail tenant. Report monthly to the lender. Sell within twelve months.

The timing. Wait for the right moment to sell. The receiver watches the local market, the regeneration plan and buyer interest. She does not sell into a weak market. She waits for the tide.

The dashboard. A monthly report to the lender: income collected, arrears, costs, works completed, letting progress, marketing progress, estimated realisation. The report is the receiver's representation of the property to the party that appointed her.

Reading A is competent, and it may produce a good outcome for the lender. But it has the estuary's two properties, a medium treated as non-reflexive and a reader standing outside the reading, and they produce predictable effects.

The medium reads back. The receiver's appointment is read by every actor at once. The son reads it as the end, and stops engaging, then starts moving what money he can into the other properties' arrears. The widow reads it as a betrayal by a lender she had banked with for thirty years. The long-standing tenants read it as a threat to their informal understanding, and the retired postman stops doing the small repairs he had done for years, the gutter clearing among them, because "it's the bank's building now". The bakery reads it as a sign the building will be sold to someone who will not care about the street, and its owners begin, quietly, to look at a unit two doors down. The newer tenants read nothing, because no one has told them anything. The housing team reads the appointment as an opportunity to push the file, and serves a notice.

The reader is read. Each of these readings changes the constraints the receiver mapped. The gutter overflows in the first winter storm, because the postman no longer clears it, and the damp spreads to a second flat. The bakery's owners, having begun to look elsewhere, become less willing to cooperate with a sale that requires their lease to be attractive. The son's disengagement means the receiver lacks information about the informal rent understandings, which she discovers only when the long-standing tenants contest her rent review. The housing notice adds cost and time. The map's "low energy, short time" constraints become "medium" and "high". The tide the receiver was waiting for does not turn, because the receiver's own presence, read by everyone, has changed the water.

The dashboard represents. The monthly report to the lender shows rising costs, falling cooperation and a slipping timetable. The lender reads it and presses for a quicker sale. The receiver, under pressure, sells at a discount to an investor who will not take the bakery's goodwill into account. The bakery's owners, at the end of their lease, leave. The building is realised, and the debt is partly recovered. The town has a fourth vacant unit and a fifth in prospect.

Nothing in Reading A was incompetent. Each step was conventional, and each was done well. The outcome followed from the reading's structure, a receiver on the deck navigating an estuary that, being human, read her and responded.

Reading B: The Receiver Inside

The second reading treats the receivership as reflexive human complexity, applying the principles of Chapter 44. It does not change the receiver's legal duties, which remain what they are. It changes how she understands her position and therefore what she does.

The participants question. Before anything else, the receiver asks: is this situation unitary, pluralist or coercive? The honest answer is coercive. The receivership itself is an exercise of power by the lender over the borrower, and the receiver's arrival changes the power ratios among everyone else (Elias, Chapter 28). Tenants, bakery and borrower are all affected, and none was involved in the decision. The full tier is indicated, at least for the first weeks.

The receiver's own reading, declared. The receiver writes down, for her own file, what she expects to find, what she hopes to find and what she would prefer not to find, a declination table (Principle 7). She expects arrears, disrepair and an uncooperative borrower. She hopes for a secure retail tenant and a sale within a year. She would prefer not to find informal arrangements that complicate valuation, or tenants in hardship. Writing it down makes her expectations visible to herself, and so makes it possible for her to notice when they shape what she sees.

The appointment as gesture. The receiver recognises that her appointment is a gesture (Mead, Chapter 27) whose meaning will be completed by how each actor responds, and that she can shape the meaning by how she arrives. In the first week, before any strategy is set, she visits. She meets the widow and the son, separately and together. She explains her role and its limits honestly: she acts for the purpose of the appointment and is deemed the borrower's agent, and she will act in good faith. She asks what they know about the building that she should know. She meets each tenant, including the newer ones, who have been told nothing. She explains what the appointment means for them, what it does not mean and how to reach her. She calls on the bakery's owners and asks them about the street. She calls the housing team and tells them what she intends.

The knife-holders' grain. In these visits she learns the grain of the building (理, Chapter 36). The postman has cleared the gutter every autumn for eleven years. The informal rent understanding exists, and the widow considers it a moral commitment. The bakery's owners know which of the high street's landlords are talking to the council about the regeneration plan. The damp in the empty flat began when a previous contractor, cutting costs, repointed the rear wall with cement mortar instead of lime. The managing agent knew, and said nothing, because he was owed money.

Constitutive constraints named. She sees that the building has been governed for thirty years by the family's way of managing: informal, personal, relational, with rents traded for care. This is a constitutive constraint (Thesis IV). It is not in the leases, but it is how the building works, and it is carried in the habitus of the widow and the long-standing tenants (Bourdieu, Chapter 33). It has hysteresis. The family's way persists in the tenants' expectations even though the family no longer has control. The receiver cannot simply override it without breaking something that, unrecorded, holds the building together. She cannot simply honour it either, because her duties do not allow her to sustain below-market arrangements indefinitely without regard to the debt.

Established and outsiders. She sees an established-and-outsiders figuration (Elias, Chapter 28) among the tenants. The long-standing tenants and the bakery are the established, with the family's trust and the street's memory. The newer tenants are outsiders, paying more, unconsulted, unknown. Any change she makes will be read through this figuration. A rent review that falls on the established will be read as the bank's betrayal of the old order. Silence towards the outsiders will be read as confirmation that they do not count.

Gratings, not conquest. She decides that her first interventions will be gratings (Chapter 41): changes to constraints, made from inside, using what is to hand, that limit what the system can do without trying to steer it. She asks the postman whether he would continue to clear the gutter, and arranges a modest rent credit in return, recorded in writing, so that the informal care becomes a formal arrangement. She instructs a specialist to repair the rear wall with lime mortar, on the bakery owners' recommendation of a local builder who knows Victorian brick, and tells the housing team the timetable. She pays the managing agent's arrears from the first rents, on condition that he hands over the full history of the building's repairs. She secures the vacant shop and lets it, on a short licence, to a pop-up run by a local craft collective the bakery's owners know, which brings footfall and covers the rates.

Waiting, declared. On the question of sale she does not simply wait for the tide. She knows her waiting will be read (Chapter 21). So she tells the lender, the borrower and the bakery, each in appropriate terms, what she is waiting for and why. The regeneration plan's next stage is due in eight months. A sale before then would be into a weak market, and a sale after, if the plan proceeds, might be stronger. She also tells them what would change her mind. Her waiting is a declared gesture, and its meaning is completed by their responses. The lender, informed, accepts a longer timetable. The bakery, informed, stops looking at the unit two doors down. The son, informed and consulted, begins to engage again, and tells her about the other lenders' positions, which affects the family's options.

The report as two-way. Her monthly report to the lender is honest about the constraints and their changes, including the changes her own presence has caused. It includes what she has learned from those inside the building. The lender's credit team, whose reading of her reports is itself a force in the receivership, is invited to visit, and does. With the borrower's agreement, she shares a summary of the report's key points with the borrower, so that the family knows what the lender is being told (Principle 8: mutual horizons, within the limits of confidentiality).

The rent question, harmonised. On the informal below-market rents, she does not impose an immediate review, and she does not ignore the issue. She meets the long-standing tenants together, explains the constraint (her duty in respect of the debt) and the options, and asks what they can live with. They agree to a stepped increase over two years, recorded in writing, in exchange for the formalised care arrangements. This is quorum with inhibition (Chapter 9) and 和, not 同 (Chapter 38). It is not consensus, and not imposition, but a decision people can live with, which keeps the differences without collapsing them.

The outcome. Ten months later the rear wall is repaired, the housing file is closed, the empty flat is let and the pop-up has become a twelve-month lease to the craft collective. The bakery has agreed, in principle, to a lease extension that makes the building more attractive to buyers. The regeneration plan's next stage has been approved. The receiver markets the building, and sells it to a local investor who knows the street and wants the bakery to stay, at a price that repays the lender in full, with a small surplus to the borrower. The widow, at the completion meeting, tells the receiver that she had expected to hate her.

What the Two Readings Show

Feature Reading A: estuary Reading B: receiver inside
Receiver's position On deck; outside Inside; read as she reads
Appointment An event; then navigation A gesture; its meaning shaped by arrival
Map Constraints by energy and time Constraints, plus the grain, constitutive constraints, hysteresis, the established and outsiders
Waiting Silent; for the tide Declared; a gesture with anticipated readings
First interventions Collect, secure, remedy, review Gratings: formalise care, repair with the grain, let to the street
Rent question Review under terms Harmonise: a stepped decision people can live with
Report One-way to lender Two-way; mutual horizons within confidentiality
Tenants' response Withdraw care; contest Continue care; agree to stepped increase
Bakery's response Look elsewhere; leave Stay; extend
Outcome Discounted sale; vacant unit Full repayment; surplus; street sustained

The comparison is illustrative. It is a composite, constructed to show the difference, and real receiverships are messier, more constrained and less obliging. In many, Reading B's patience is not available. The lender cannot wait, the borrower will not engage, or the tenants are in conflict. Reading A is sometimes the right reading, when the situation is clear or complicated, not complex, and when speed matters more than texture.

But the comparison shows what the disputation's principles change in practice. They do not change the receiver's duties, powers or goals. They change her understanding of her position, from the deck to the vortex, from Wigner to the friend, from the sailor to the fisherman. That change of understanding changes what she notices (the grain, the constitutive constraints, the figuration), how she arrives (as a gesture to be shaped), how she waits (declared), how she intervenes (gratings) and how she reports (two-way). The outcome is different, not because she is kinder, though she may be, but because she has read the system as a system that reads her.

The Receiver as Figure

The receiver is, in one respect, the purest case of the practitioner inside. She cannot pretend to stand outside. Her appointment places her in the configuration, and her legal position (agent of one party, appointed by another, owing duties to both) makes her the mediator between them, the 人 between heaven and earth (Chapter 38). Every move she makes is read by parties with conflicting stakes. Her waiting is a gesture, and her reports are representations that speak for the property to the lender (Chapter 42). She is Kipling's Mariner, inside the whale, with stakes, able to change the system's throat only with what is to hand.

The figure has a lesson for anthro-complexity's practitioners, whose position is less obviously inside than the receiver's but is inside nonetheless. A consultant appointed by a client to sense a population is, like the receiver, appointed by one party and operating among others, read by all of them, her presence a gesture and her reports representations. The receiver's legal position makes explicit what the consultant's commercial position leaves implicit. You are inside, you are read, and your duty runs in more than one direction.

The adversary will say the case is rigged. Reading A was written to fail and Reading B to succeed, and the composite's details (the gutter-clearing postman, the lime mortar, the bakery's builder, the approved regeneration plan) were chosen to reward the reflexive reading. Real receiverships rarely offer such obliging facts. Reading B's patience depends on a lender willing to wait, a market that rises and tenants who negotiate, and when those are absent, Reading B is merely slower.

The adversary is right, and the chapter has said so. The case is a composite, constructed to show a difference, and its facts are obliging. A fair test would require real cases, compared systematically, which this book cannot supply. But the case's claim is not that Reading B always succeeds. It is that Reading A's failures, when they occur, follow from its structure, the estuary's two properties, and that Reading B's attention to being read is available even when its patience is not. A receiver who must sell quickly can still arrive as a declared gesture, still learn the grain, still formalise the gutter-clearing and still tell the bakery what she is doing. Those cost little and change much. The adversary has shown that the composite flatters the reflexive reading. It has not shown that the reflexive reading's core, you are inside and you are read, is wrong. partial

APPOINTED BY ONE — AGENT OF ANOTHER — READ BY ALL —

THE RECEIVER IS ALWAYS INSIDE THE WHALE —

FORMALISE THE GUTTER-CLEARING, AND DO NOT FORGET THE SUSPENDERS.

Addition: What the Receiver's Position Teaches the Consultant

The chapter ended by observing that the receiver's legal position, appointed by one party, deemed the agent of another and owing duties to both, makes explicit what a consultant's commercial position leaves implicit. This addition develops the comparison, because it offers anthro-complexity's practitioners a model of their own position that is more honest than the one their instruments usually assume.

The receiver's position, schematically

In broad terms, and subject always to the instrument of appointment and the law, the receiver's position has the following structure:

  • Appointed by the lender, whose interest in repayment is the purpose of the appointment.
  • Deemed agent of the borrower, whose property she manages and who is liable for her acts.
  • Owes the lender the duty of pursuing the purpose of the appointment.
  • Owes the borrower, and others interested in the property, duties such as good faith and, where she manages, due diligence in managing.
  • Acts among occupiers, tenants, contractors, authorities and others who are not parties to the appointment but whose interests her actions affect.
  • Reports to the lender, and accounts, in the ways the law requires, to the borrower.

The structure is multi-directional. The receiver cannot serve one party alone. Her legitimacy depends on honouring duties that run in more than one direction, and the law has developed, over a long history of cases, doctrines that police the balance.

The consultant's position, schematically

A consultant deploying anthro-complexity's instruments in an organisation has a position whose structure is similar, though rarely made explicit:

  • Appointed by the client, typically the organisation's leadership, whose purposes the engagement serves.
  • Acts upon the organisation's members, whose narratives she gathers, whose constraints she maps and whose situations she categorises.
  • Owes the client the delivery of the engagement's purpose.
  • Owes the members, by the ethics of her practice and the Pelagian stance, respect for their capacity to sense and heal themselves.
  • Acts among others affected but not involved: customers, communities, regulators, future employees.
  • Reports to the client and, in good practice, returns patterns to the members.

The structure is also multi-directional. But the consultant's duties to the members are not policed by law as the receiver's are. They rest on professional ethics, the practitioner's own values and the client's tolerance. The consultant is appointed by the client and acts upon the members, and nothing in the structure of the engagement requires her to treat the members as parties to whom she owes duties of the kind the receiver owes the borrower.

What the comparison suggests

The comparison suggests that anthro-complexity's practice could make its multi-directional duties explicit, as the receiver's law has done. The following duties, modelled loosely on the receiver's, might be stated at the start of every engagement:

Receiver's duty (broad analogue) Consultant's corresponding duty
Purpose of appointment Deliver the engagement's stated purpose to the client
Good faith to the borrower Good faith to the members whose narratives and constraints are sensed: no deception about purpose; no use of their material against them without their knowledge
Due diligence in managing Due care in sensing: protect the tails; route the atypical; apply the established-and-outsiders lens
Account to the borrower Return patterns to members before action; make the client's reading visible
Regard for others interested Witness for those affected but not involved (Ulrich, Chapter 12)

Stated at the start of an engagement, these duties would do for the consultant what the receiver's law does for the receiver. They would make the multi-directional structure of her position explicit, give the members a basis on which to hold her to account, and give the consultant a basis on which to resist a client's demand that she serve the client alone.

Why the receiver is honest about her position

The receiver's position is honest about its multi-directionality because the law made it so, over a long history of cases in which receivers who served lenders alone were held to have breached duties to borrowers. The honesty was not chosen. It was imposed by the consequences of dishonesty.

Anthro-complexity's practice has no such history of cases, and its honesty about its position must be chosen. The disputation's request is that it be chosen, before a history of failures imposes it. The principles of Chapter 44 and the protocols of Chapter 45 are one way of choosing. The duties above are another, simpler and closer to the practitioner's own experience of her position: you were appointed by one party; you act upon others; you owe duties to both.

The receiver as the figure of the consultant

The receiver is the practitioner inside in its legal form. She is inside the configuration from the moment of appointment, read by every party, her waiting a gesture and her reports representations, with duties that run in more than one direction and are enforced by law. The consultant is the practitioner inside in its commercial form, in the same position in every respect except the enforcement. The disputation offers the receiver to the consultant as a mirror. This is your position, made visible. These are your duties, stated. This is what honesty about where you stand looks like.

APPOINTED BY ONE — ACTING UPON OTHERS — OWING DUTIES TO BOTH —

THE RECEIVER'S HONESTY WAS IMPOSED BY THE LAW —

THE CONSULTANT'S MUST BE CHOSEN.

Addition: Cross-Lights From Book Nine

The case as a joiner. The receivership of this chapter was read twice, as an estuary and as a figuration. The Zen Garden study gives a name to the second reading: it is the garden walked. The lender, the borrower, the tenants and the receiver each have a seat, and from each one stone is hidden. A receiver's report that lays their views side by side is a joiner.

Truth to power. Flinn's caution applies to the tenants of the case. What they tell a receiver is a practical judgement about what it is wise to say to someone appointed by the lender.

Chapter 47

Coda: Humans Are Not Estuaries

3,681 words, about 16 minutes of reading

The Tree, the Rock, the Water, and the One Who Looks

The disputation began with a photograph: the roots of a tree gripping an isolated rock before a waterfall. Snowden chose it to show how complexity and systems thinking stand to each other. They are three actants, each constraining and enabling the others, and none contains or descends from the others. They are neighbours, not branches. Chapter 1 accepted the picture and noted what it left out: the human, the viewer, the conversation. It proposed a second image, two people at the pool's edge, one photographing the tree, one reading a map drawn by a third who was absent, each reading the arrangement differently and each reading changing what the arrangement would become.

Forty-six chapters later, the second image can be completed.

The photographer is the theorist, framing the arrangement as three actants and calling it a picture of how traditions stand to each other. Her frame is a gesture. Published, it becomes an actant in the relation it depicts (Thesis II). She stands on the bank, and the waterfall does not read her, but the two traditions she depicts do. Ing's readers and Snowden's will see the photograph and respond, and the relation will change because it was pictured.

The map-reader is the practitioner, navigating the river by a chart. If she reads the chart as a navigator, plotting a route through features fixed in advance, she is the sailor of the estuary (Thesis III). If she reads it as a wayfarer, knowing the river by moving through it and adjusting to what she meets, she is inside, and her reading is part of the river's life (Chapter 30).

The absent cartographer is the designer of instruments, whose map has been adopted by an authority and so shapes what happens to the rock more than either person present (Thesis V). Her map was drawn from somewhere, with categories she chose, and it now governs what the river is taken to be. The map has become a trail others follow (Chapter 30), a representation that speaks for the river (Chapter 42), an engine and not a camera (Chapter 34).

And there is a fourth figure the first image could not show, and the second only implied. It is the person in the water: the fisherman in the maelstrom (Chapter 28), the friend in Wigner's lab (Chapter 39), the Mariner in the whale (Chapter 41), the cook with his knife in the ox (Chapter 36), the calligrapher whose arm is inside the stroke (Chapter 35), the receiver in the building whose every move is read (Chapter 46). She cannot photograph the arrangement, because she is part of it. She cannot read the chart from above, because she is in the current. Her knowledge is the knife's and the wayfarer's, acquired by contact, shown in action and incomplete. It is the only knowledge of the arrangement that includes what it is like to be in it.

Anthro-complexity's theory is about the fourth figure. Its human cluster, its asymmetry principle, its insistence that humans are not ants and its Peircean pragmatism all describe human systems as systems of people who are in the water: narrating, intending, abstracting and changing the constraints by changing the story. Its instruments, in the drift this book has traced, are drawn for the photographer, the map-reader and the cartographer, the figures on the bank. The disputation's argument, in one sentence, is that the theory is right about the fourth figure, and that the instruments should be redrawn for her.

The Argument Gathered

The argument can now be gathered into its sequence.

  1. Anthro-complexity is a body of thought with twenty-two elements in four clusters. It is not a brand, and it deserves to be argued with on its own terms (Book One).
  2. It is right that complexity is not a kind of systems thinking (Thesis 2), and right that human complexity differs in kind from other complexity (Thesis 3) (Books II–III).
  3. Its five reasons for treating human complexity differently (scientific, reflexive, anti-computational, ethical, positional) all hold. Four of them, followed through, ground the difference relationally, in shared intentionality, collective acceptance, descriptive reflexivity, participatory sense-making and the relation between intervener and intervened. The fifth explains why the difference is stated in a grammar of possession (Book Three).
  4. The difference stated relationally re-enters among humans, dividing them into describers and described, representers and represented, readers and read (Chapter 18).
  5. Eight theses follow. The human cluster's grammar presumes persons (I). The asymmetry principle applies to anthro-complexity's own descriptions (II). The estuary places the reader outside a non-reflexive medium (III). Constitutive constraints are reproduced in compliance and mapping, and everyone is a reagent (IV). Power is missing from the theory of the human (V). The practice is computational where the ontology is not (VI). The triad is a simplex (VII). Reflexivity predicts the degradation of stable instruments (VIII) (Book Four).
  6. Mead, Elias, Juarrero, Ingold, Hutchins, Peirce, Bourdieu, Giddens, Hacking and the theorists of performativity converge. The human is relational, the observer is inside and the instrument is an actor (Book Five).
  7. 道理勢 converges with them. 勢 includes the strategist and the political 勢 of position, and a 勢 read from outside is terrain. 理 is known by the knife. 道 is made by walking, and the test is we did it ourselves. Mediation is real, and its paradigm is the human between heaven and earth (Book Six).
  8. The physics of measurement and the political theory of representation converge again. Descriptions are from positions. Asking changes the asked. Only human collectives can be represented, and so only they can become, or ward off, a Leviathan (Book Seven).
  9. The adversary wins half, and the half it wins is the outside from which this book was written. The inside view is invited (Chapter 43).
  10. A reflexive anthro-complexity keeps nineteen elements, amends three and adds twelve principles. Its figure is the practitioner inside, the listener of Stage One restored, with the instruments and the theory of constraint of the later stages in her hands (Book Eight).

Why Human Complexity Is Treated Differently

The question that began this inquiry, before it was a disputation, was simple: why does anthro-complexity treat human complexity differently? The disputation's answer can now be stated plainly.

Anthro-complexity treats human complexity differently because human complexity is different, and anthro-complexity has seen this more clearly than any other school that grew from the complexity sciences. Human systems are not populations of rule-following agents, as CAS models them. They are not wholes with parts serving functions, as systems thinking presumes them. They are not flat fields of intra-actions, as the new materialisms flatten them. They are systems of people who tell stories about the system, and whose stories change it. They are systems of people who can be represented, made one by one who speaks for them, and who can resist being represented. They are systems in which every description is an intervention, every observer is a participant and every instrument is an actor.

Anthro-complexity sees this, states it as the asymmetry principle and grounds it in the three marks of the human. Its grounding is partly misplaced, in the properties persons have and not in the relations through which persons are, and the misplacement has consequences for its instruments. But the seeing is right, and it is anthro-complexity's great contribution. Humans are not ants. That slogan, and the theory behind it, have freed a generation of managers from a misleading class of models.

The disputation's addition is one further step, which anthro-complexity's own premises require. Humans are not estuaries either. The medium of human complexity reads the reader. The practitioner is in the water. The instruments are part of what they sense. And the asymmetry between those who describe and those described, between the planners and the planned-for, the convenors and the convened, is a human–human asymmetry. The theory must name it, and the instruments must make it visible.

The Formulas

The disputation has gathered its findings, chapter by chapter, into short formulas, in capitals at each chapter's close. The principal formulas are collected here, so that the argument can be carried away in a few lines.

NEITHER WHOLES NOR PARTS PRESUMED — THEN NEITHER PERSONS.

WHERE THE DIFFERENCE IS IN THE AGENT, IT IS ONE OF DEGREE; WHERE IT IS IN THE RELATION, IT IS ONE OF KIND.

EVERYTHING SAID IS SAID BY AN OBSERVER; EVERY FRAMEWORK APPLIED BECOMES AN ACTOR.

AN ECOSYSTEM CAN ONLY BE CONDITIONED; A PERSON CAN BE ADDRESSED.

THE BORDER IT DRAWS IS BETWEEN HUMANS AND RIVERS; THE BORDER IT NEEDS IS BETWEEN READERS AND READ.

人 AND 二: THE HUMAN IS BETWEEN.

THE PLANNERS ARGUE ABOUT THE RIVER; THE CONSULTANTS ARGUE ABOUT THE PLANNERS.

THE TIDE DOES NOT READ THE TIDE-TABLE; THE ORGANISATION READS THE MAP.

IN HUMAN AFFAIRS EVERYONE IS A REAGENT.

WHOEVER DEFINES BALANCE, DESIGNS THE TRIAD, HOLDS THE DASHBOARD — WINS.

FIT A GRATING IN THE WHALE'S THROAT, AND DO NOT FORGET THE SUSPENDERS.

THE THIRD THAT MEDIATES SHOULD BELONG TO THE ONE WHO SPOKE.

THE STARLING DOES NOT LEARN THE CAMERA; THE WORKFORCE LEARNS THE TRIAD.

THE FISHERMAN SURVIVES BY OBSERVING FROM INSIDE THE VORTEX.

THE SHIP KNEW WHERE THE SHIP WAS, BECAUSE EVERYONE COULD SEE THE CHART.

A PERSON IS NOT ABSOLUTELY AN INDIVIDUAL.

地形 IS THE GROUND; 勢 IS WHAT THE GENERAL MAKES ON IT, WHILE THE ENEMY WATCHES.

THE GRAIN IS KNOWN BY THE KNIFE.

THE WAY IS MADE BY WALKING; WHEN THE WORK IS DONE, THE PEOPLE SAY: WE DID IT OURSELVES.

THE EXEMPLARY PERSON HARMONISES BUT DOES NOT CONFORM.

THERE IS NO VIEW FROM NOWHERE, ONLY THE GOD TRICK.

ALIGNMENT CANNOT BE BROADCAST; IT IS RECOGNISED AT THE RENDEZVOUS.

A MULTITUDE IS MADE ONE BY ONE WHO REPRESENTS IT; WRITE THE DISPERSAL RULE BEFORE YOU CONVENE.

THE ADVERSARY WINS HALF, AND THE HALF IT WINS IS THE OUTSIDE FROM WHICH THIS BOOK WAS WRITTEN.

A Letter, Not a Verdict

A disputation ends, in the scholastic form, with a ruling, the master's resolution of the question. This one ends instead with something closer to a letter, because the question it disputes is not closed. The adversary showed, in Chapter 43, that the disputation's view is from outside. The inside view, what practitioners of anthro-complexity know about their instruments in use, has not been heard. A determination without it would be Wigner's verdict on his friend's experiment.

So the disputation ends by addressing those inside, and Dave Snowden first among them.

The November post asked, about the Chinese sources: he would like to know whether the Chinese sources have a way of handling mediation that he has missed. Chapter 38 offered an answer, the three powers, the blending breath and harmony without conformity, and found that the paradigm of mediation in those sources is the human, standing between. The disputation offers the answer back as a gesture, whose meaning the response will complete.

The November post said that Peirce is still live. The disputation agrees, and has tried to show (Chapter 32) that Peirce, taken whole, holds that a person is not absolutely an individual, that the interpretant belongs to the community, and that the self is a sign in dialogue with the self to come. If that Peirce is still live, the human cluster can be restated in his terms, and the instruments redesigned around the community of inquiry he describes.

The November post said that a theory which sets intention aside cannot account for how human systems differ from ecological ones, and that this difference is what makes intervention in human systems both possible and dangerous. The disputation agrees, and has tried to show that the difference, followed through, applies to the theory's own interventions, its instruments, categories and maps, as much as to anyone's.

The November post ended its critique of Ing with a parallel line: complex systems can be looked at systemically, but complexity is not a kind of systems thinking. The disputation offers its own parallel: human systems can be sensed with instruments, but an instrument in a human system is not a camera. It is a participant.

And the November post opened with an invitation: If I have misread either source, I would welcome the correction. The disputation extends the same invitation, and means it. Where this book has misread anthro-complexity (its instruments in use, its practitioners' care, its evolution since the texts the author has read), the correction is welcome. The author is outside, and the correction from inside is what the disputation lacks. The rendezvous is the inquiry.

Under the Additive Covenant

This book is composed under the Additive Covenant: nothing is deleted, only inserted. Every verdict amended in Chapter 43 keeps its earlier judgement visible. Every thesis stands with its adversary's half conceded. The covenant applies to what comes after the book as well. If practitioners inside correct its readings, the corrections will be added, in place and dated, with the original readings kept. If the instruments evolve towards the reconstruction, the evolution will be recorded and the criticisms amended. If they evolve elsewhere, that too will be recorded. The book is not finished. It is a warp, laid by one convenor, on which others are invited to weave.

The covenant has a further meaning, which the disputation has tried to honour throughout. A book that argues against the outside observer cannot pretend to have been written from outside its own history. It was written by a particular author, at a particular time, in a particular field: a receiver and chartered surveyor in London, publishing under an imprint that sells nothing, working at the intersection of complexity theory, organisational philosophy, classical Chinese thought and continental philosophy, and indebted to the work it criticises. Its readings are from that position. They are offered as a gesture, in the hope that the response will complete them.

The Last Image

Return, finally, to the waterfall.

The tree grips the rock. The water falls behind them. Snowden's reading is that three actants constrain and enable one another, none containing the others, and that they are neighbours. The reading is true.

Now add the one who looks. She is not on the bank this time, with a camera. She is in the pool, waist-deep, feeling the current's pull and the stones underfoot. She sees the tree from below, its roots dark with spray, and the rock wet and slippery and nearer than it looked from the bank. She cannot frame the three actants, because she is too close and the water is in her eyes. But she knows something the photographer does not: what the current does to a body, where the stones are firm, how the spray changes with the wind, and which way to step when the pull is strong. Her knowledge is partial, situated and incomplete. It is also the only knowledge of the waterfall that includes what it is like to be in it.

She is not a fourth actant beside tree, rock and water. She is the one in whom their relation becomes meaning: the interpretant, in Peirce's terms; the 人 between heaven and earth; the fisherman who observes from inside the vortex and finds what descends slowly. Without her, the three actants constrain and enable one another, and nothing is understood. With her, the arrangement is read, from inside, and her reading changes how she moves in it, and her moving changes the water around her.

Human complexity is the complexity of systems that contain her. Anthro-complexity's theory knows this. Its instruments, in the drift this book has traced, were drawn for the photographer on the bank. The disputation's whole request is that they be redrawn for the one in the water. She should be the one who reads, who is read and who says, when the work is done, we did it ourselves.

ANTHRO-COMPLEXITY SAYS HUMANS ARE NOT ANTS.

THE HARDER CLAIM IS THAT HUMANS ARE NOT ESTUARIES EITHER —

AND NEITHER IS THE PERSON READING THE MAP.

NOT A BRANCH OF THE TREE —

BUT A ROOT IN THE SAME ROCK —

AND SOMEONE IN THE WATER, LOOKING UP.

Principia Humanitatis. London, November 2026.

Afterword: What Would Change the Author's Mind

A disputation that asks its adversary to be open to correction should say what would correct it. This afterword states, as precisely as the author can, the findings that would require the disputation's theses to be amended or withdrawn. It is offered in the spirit of Peirce's fallibilism and of the Additive Covenant. The theses are held as hypotheses, and these are the tests.

Thesis I (the possessive fallacy)

What would refute it: evidence that anthro-complexity's instruments, as designed and used, already collect relational data at scale (the addressee of stories, the others in them, joint signification, the narrator's reading of patterns), so that the possessive grammar of the human cluster has no effect on design. Equally, a statement by Snowden of the human cluster in relational terms, consistent with Element 2, would show that the grammar of the November post was shorthand, as the adversary's Article 2 claimed.

Thesis II (the asymmetry principle applies to itself)

What would refute it: evidence that anthro-complexity's instruments routinely record their own descriptions and their effects, include the convenor's own signification, and declare the designers' positions. The four structural requirements of Chapter 20 would be found already met.

Thesis III (humans are not estuaries)

What would refute it: evidence that estuarine mapping, as practised, maps the mappers, records the effects of the mapping on the constraints mapped, and treats the practitioner's waiting as a gesture with consequences. Alternatively, a showing that the estuary image has no effect on method, that practitioners trained in it behave no differently from practitioners trained in a human image such as the children's party or Jacobs's city.

Thesis IV (the under-socialised constraint)

What would refute it: a showing that constitutive constraints are already part of anthro-complexity's constraint vocabulary and are routinely mapped, or that constructor theory is used only at the level of possibility (what can and cannot be done) and not to classify human practices as unchanged transformers.

Thesis V (the missing term is power)

What would refute it: evidence that power-awareness is built into the instruments' design, not only practitioners' vigilance: signifier sets that declare their agenda, mutual dashboards, categorisations recorded as acts, mapping sessions that record who placed what, and risk tracked along the gradient. Also, a statement of the human cluster in which power appears as a dimension.

Thesis VI (computational practice)

What would refute it: evidence that the move towards human/AI hybrids includes gratings: separate and revocable consent to harvest, return before action, protection of the tails, routing of atypical narratives to humans. Also, evidence that the tails of narrative distributions do not thin when instruments are coupled to models over repeated rounds.

Thesis VII (the triad is a simplex)

What would refute it: evidence that triads, as used, are always accompanied by independent scales or other signifiers that make strong-in-all and weak-in-all distinguishable, and that analysts never read closure-induced negative associations as findings. Also, a mathematical argument that the closure constraint does not, in practice, distort narrators' signification. The author would welcome one, and does not expect it.

Thesis VIII (degradation)

What would refute it: data from repeated deployments showing that the five predicted patterns do not occur: signification does not converge on salient positions, outliers do not thin, signifier vocabulary does not appear in free text, reflexive content does not grow, and frame changes do not produce recovery. This is the most testable of the theses, and the data are held inside the practice. If practitioners ran the test and found the predictions false, the thesis would be withdrawn, and the disputation's account of descriptive reflexivity would need revision.

The reconstruction

What would refute it: evidence that the protocols of Chapter 45, applied in practice, produce worse outcomes than current practice. They might be more costly without commensurate benefit, more bureaucratic without more honesty, or more paralysing without more recognition. The adversary's objections to cost (Chapter 44) and to bureaucracy (Chapter 45) are serious, and only practice can test them.

What would not change the author's mind

It is also worth saying what would not count as refutation, so that the disputation's claims are not mistaken for something they are not.

  • Evidence that anthro-complexity's practitioners are careful, ethical and reflexive. The disputation has never doubted it. The theses concern design and theory, not character.
  • Evidence that anthro-complexity's instruments produce better outcomes than surveys, expert diagnosis or agent-based models. The disputation agrees, and has said so. The theses concern the gap between anthro-complexity's theory and its instruments, not between anthro-complexity and its inferior rivals.
  • Evidence that Snowden has addressed some of these points in writings the author has not read. Welcome, and likely. That would be a correction of the disputation's reading, not of its theses, unless the writings showed that the theses' targets do not exist.

The rendezvous

The tests above cannot be run from outside. They need the data, the deployments and the experience of practitioners inside. The disputation's final request is that some of them be run, and the results shared, so that the outside view and the inside view can be compared at the rendezvous. The author will add what is learned, under the Covenant, in place and dated, with these theses left visible.

STATE WHAT WOULD REFUTE YOU —

HOLD THE THESES AS HYPOTHESES —

AND MEET AT THE RENDEZVOUS WITH THE DATA.

Addition: After the Coda

The book was to end here. A ninth book follows, written because two further posts by Snowden required an answer and because the argument needed to be tested against Rahel Jaeggi's account of criticism, against A Web of Relations and Kevin Flinn's book, against the Zen Garden study, and against the author's own objection to the doctrine of incompressibility. Its finding is that the distance between this disputation and anthro-complexity is smaller than the coda supposed, and that it now lies in the instruments.

Chapter 48

The Trialectic: Three Architectures of Practice

1,927 words, about 8 minutes of reading

Book Nine: After the Coda

The disputation closed at Chapter 47. This ninth book is an addition made under the Additive Covenant, and nothing before it has been altered. It exists because the argument has to answer two further texts by Dave Snowden. One is earlier than the post that occasioned the disputation, and one is later.

The earlier text is "A Trialectic for organisational practice & coaching", published on 31 May 2026. It extends the Augustine and Pelagius distinction of Element 20 with a third term, and that third term comes closer to the disputation's own position than anything else in anthro-complexity. A critique that ignored it would be unfair. The later text is "Irritation is not an alibi", published on 14 November 2026, the day after "Not a branch of the tree". It sets anthro-complexity against the sociology of Niklas Luhmann, and it contains the plainest statement yet that the practitioner does not stand outside. A critique that ignored that would be out of date.

Two resources are brought to the task. The first is Rahel Jaeggi's Critique of Forms of Life, which supplies an exact account of what kind of criticism the disputation has been practising and what such criticism may claim. The second is A Web of Relations, the web edition of Kevin Flinn's work on complexity and organisational development, which Mark Downham edits. It contains the fullest statement of how the process schools reply to anthro-complexity, and a proposal, formation, that bears directly on the trialectic.

The book has eleven chapters. This one states the trialectic. Chapter 49 sets out Jaeggi. Chapter 50 applies her immanent criticism to the trialectic. Chapter 51 brings in A Web of Relations. Chapter 52 examines the Luhmann post. Chapter 53 applies the Zen Garden study throughout. Chapter 54 asks whether the practice blocks its own learning. Chapter 55 gathers the dialectics and revises the verdicts. Chapter 56 challenges the doctrine of incompressibility, Chapter 57 examines Flinn's book, and Chapter 58 interrogates Cilliers's ethics of framing in full.

The Occasion of the Trialectic

Snowden had used Augustine and Pelagius for some years as a diagnostic pair, and Chapter 5 recorded it as Element 20. In his own account two objections kept coming back. The first was that the scheme is a binary, and he distrusts binaries himself. The second was that fifth-century theology is a closed book to most practitioners. He treats these as separate complaints with separate remedies. The binary is cured by adding a third term. The inaccessibility is cured by giving a plain-language version that can stand alone.

He considers two candidates for the third term and rejects both, and his reasons are a useful statement of what he thinks a trialectic is.

John Cassian, the fifth-century monk usually called semi-Pelagian, held that the human will begins and grace completes. Snowden rejects him because he stands between the two poles. A point on a line between two others belongs to the same line. The third term of a trialectic must be irreducible to the other two. A compromise assembled from them does not qualify.

Nāgārjuna, the second-century Buddhist philosopher, is rejected for the opposite reason. His doctrine of emptiness denies that system, practitioner or capacity has any inherent existence. That removes the ground on which all three positions stand. Snowden's judgement is that cancelling a frame is different from taking a position within it. He adds that he thinks the doctrine mistaken, though interesting.

The test that both fail may be called the triquetra test, after the three-cornered knot Snowden uses to draw the scheme. A real third term must be neither a midpoint nor a dissolution. His candidate is Maximus the Confessor, the seventh-century Byzantine theologian.

Three Positions in Plain Words

The assumed deficit model. The practitioner arrives believing that the system or person lacks something needed for change. The practitioner's job is to supply it: the method, the insight, the missing capability. Snowden observes that the belief is seldom stated and nearly always present, and that it sustains itself. Without a deficit there is no role for an intermediary, so the deficit tends to survive whatever happens. He thinks most consulting, most coaching and much facilitation work this way without knowing it.

The innate capacity model. The practitioner's job is to arrange conditions under which what is already there can operate. Snowden's word for this is wayshaping: working with constraints, context and the existing grain. The system is treated as capable.

The formative encounter model. The meeting between practitioner and system produces something that could not have been credited to either beforehand. Both are participants in something neither controls or holds in advance. What emerges belongs to the coupling. The practitioner's own formation is part of the work. The boundary between practitioner and system is the place where something new comes into being. It does not divide two prior stocks of capacity.

Snowden insists that these are not points on a scale. To move from the first to the second one must change what one believes about the system. To move from the second to the third one must accept two things: that one will be changed, and that what both parties become could not have been predicted or possessed by either.

The Theological Genealogy

The three are then located in history. Snowden's stated purpose is to show that they are old and distinct modes with deep roots, and that the dominance of the first has an explanation.

The deficit model is Augustinian. For Augustine, original sin so damaged the will that real transformation needs something the human cannot supply. Grace comes from outside and through authorised channels. Snowden's translation is direct: put expert method where grace was, and one has a familiar model of organisational development, in which the intermediary's authority rests on exclusive access to what is needed.

The capacity model is Pelagian. Pelagius held that people have real, if imperfect, resources of their own, and that grace enables what is present. Snowden notes that Pelagius was from Britain, and probably Welsh, and that his theology came from a Celtic Christianity with a stronger sense of the goodness of creation. Augustine won the institutional argument, and Snowden's remark is that Western organisational practice has been Augustinian ever since.

The encounter model is Maximian, with a caveat. Maximus's account of synergeia, working-together, describes a transformation that neither the human nor the divine party produces alone. Theosis, deification, is neither delivered to a passive recipient nor achieved by an autonomous will.

The caveat is stated at the end of the post, and it matters. Synergeia can be read as two wills, each real, each bringing its own capacity, cooperating and being changed. Snowden says that this is narrower than what he means. His position is "stranger": the capacity does not exist within either boundary before the coupling. Both parties remain real, which he calls the guardrail against dissolving into undifferentiated co-creation. But what becomes possible could not have been predicted from the properties of either. He allows that Maximus may be only the best available approximation, and asks for help in finding better resources, perhaps from the Eastern tradition read alongside enactive cognitive science.

How the Three Are to Be Used

Four instructions follow, and each is to the post's credit.

All three stay in play. Within one engagement there are moments when something really is missing and must be supplied, moments when a capacity is present and only needs to stop being suppressed, and moments when the encounter itself is what matters. The skill is to hold the tension and to notice which mode is doing the work.

The centre is aporetic. The middle of the triquetra is empty. It is neither a fourth position nor a synthesis, and it gives no higher level from which to manage the other three. Snowden connects this to the central domain of Cynefin and says the emptiness is the generative starting point of the whole design.

It is a diagnostic and no ladder. To read the three as stages of a practitioner's growth would bring back the Augustinian logic: a correct destination at which the expert has already arrived.

The team should examine itself. A consultancy should ask which mode it defaults to and why. Snowden's answer for most teams is the deficit mode, because the market rewards it and clients expect it. He adds that the encounter mode is often available and not taken because it feels professionally risky.

One paragraph applies the scheme to anthro-complexity's own instruments, and the disputation will return to it several times. A single SenseMaker deployment, Snowden says, passes through all three modes. Choosing the area of inquiry may be Augustinian in nature or motive. Having people signify their own stories is Pelagian. Intervention designs using what he calls entangled trios, extended into distributed work in one domain of the WRASSE framework, are co-creative. With SwarmCompass added, the practice keeps what he calls the agility of aporetic positioning.

The post ends by describing its banner image: three interiors. A Roman basilica, long and directional, with light falling on a distant altar, is the Augustinian space. An Irish monastic enclosure on an Atlantic shore, with a round tower and small oratories inside a low wall under open sky, is the Pelagian space. A Byzantine domed church, with light diffused from a ring of windows and an iconostasis marking a threshold, is the Maximian space.

What the Trialectic Concedes to the Disputation

The disputation's theses were framed against "Not a branch of the tree" and the published instruments. The trialectic was written five months before that post. Read beside the theses, it grants a good deal, and honesty requires a list.

Concessions, thesis by thesis.

To Thesis III (the observer inside). The post says that a practitioner who arrives with a stable external perspective, however humble, is still treating the complex as if it were complicated. That is Thesis III in Snowden's words. It also says that in the complex domain the relation of probe and response is co-constitutive, which is the probe-as-gesture argument of Chapter 27.

To Thesis IV (no unchanged constructor). The practitioner's formation is "part of the work". The principle that everyone is a reagent says no more.

To Thesis I (the possessive fallacy). The formative encounter is defined by a capacity held by neither party. That is an explicit departure from the grammar of possession at the level of practice.

To the positional reason (Chapter 17). The post states that the market rewards the deficit mode, and that this shapes what consultancies do.

To Thesis V (power), in part. The deficit model is analysed as an arrangement that maintains the intermediary's authority. That is an analysis of power, applied to the consulting relation.

So the disputation's opening charge needs restating. It cannot say that anthro-complexity has no account of the practitioner inside. It has one, and a subtle one. The revised question is narrower and harder. The trialectic states a norm: in complex situations the encounter forms both parties, and nothing useful happens unless the practitioner is willing to be changed. Do anthro-complexity's instruments and arrangements embody that norm? Or does the practice, as built, work against the standard that its own author has set?

That is a question about the relation between a practice and the norms internal to it. There is a philosophy of exactly that relation, and the next chapter sets it out.

THREE ARCHITECTURES, ONE EMPTY CENTRE —

THE THIRD IS THE DISPUTATION'S OWN POSITION, NEARLY —

THE QUESTION IS WHETHER THE INSTRUMENTS LIVE THERE

Chapter 49

Jaeggi: Forms of Life and the Three Kinds of Criticism

3,460 words, about 15 minutes of reading

Why Jaeggi

Rahel Jaeggi is a German philosopher in the tradition of Frankfurt critical theory. Critique of Forms of Life appeared in German in 2014 and in Ciaran Cronin's English translation in 2018. Its question is whether ways of living can be criticised as better or worse, and by what standard, without either imposing an external ideal or collapsing into the view that each way of life is right by its own lights.

She is useful to this disputation for three reasons. Her account of a form of life as an ensemble of practices that solves problems fits a professional practice such as anthro-complexity closely. Her distinction among three kinds of criticism says precisely what the disputation has been doing since Chapter 19 and what it is entitled to conclude. And her criterion of success, which is that a form of life should not block its own learning, gives a test that is neither a rival doctrine nor mere taste.

This chapter sets out her argument. The next applies it.

What a Form of Life Is

Jaeggi defines a form of life as a cluster of social practices, together with the attitudes and interpretations that go with them, which has some permanence and some depth and which carries normative expectations. The bourgeois nuclear family is one of her main examples. So are the city as a way of living, and the working life organised around wage labour. A fashion is too shallow and brief to count. A whole civilisation is too large to be criticised as one thing.

Four features of the concept matter here.

An ensemble of practices. A practice is a habitual, rule-guided sequence of actions with a purpose: shopping, hearing a lecture, holding a weekly meeting. Practices do not stand alone. They hang together in bundles, and the bundle gives each one its sense. Jaeggi describes the connection as moderately holistic. Being part of the ensemble changes what a practice is, and the ensemble is nothing over and above its practices. She adds that forms of life are variable nexuses, with some practices tightly linked and others loosely, and that they are never closed wholes.

Inert. Practices settle. They become habit, they are built into rooms and tools and timetables, and they are fixed in institutions. Jaeggi's phrase is that forms of life are at once given and made. People made them and keep remaking them, and they confront those same people as second nature. Because they are made, they can be criticised. Because they are sedimented, criticism is hard.

Normative. One can take part in a practice well or badly, and there are ways of getting it wrong. The norms are internal to the practice. Jaeggi separates this kind of norm from convention and from the rules of a game. A doctor who does not try to heal is not a doctor who plays by other rules. He has failed at what the practice is. Her discussion of professional ethos makes the point that such norms are tied to real conditions and to what the practice is for.

Problem-solving. Each form of life is an attempt to solve problems that arise in living together. The family is an answer to the problems of raising children and of intimacy. Jaeggi takes the notion of a problem from John Dewey and adds something from Hegel. Problems, like forms of life, are at once given and made. Something goes wrong and announces itself as a disturbance. What the problem is has to be worked out, and the available descriptions come from a stock that society provides.

The Hegelian addition is that the problems of a form of life are second-order. They do not strike from outside like weather. A form of life sets itself problems by the claims it makes, and it fails by its own measure. The modern family promises both freedom for its members and belonging, and its typical crises come from that promise.

Three Kinds of Criticism

How may such a thing be criticised? Jaeggi distinguishes three ways.

External criticism brings a standard from outside: a universal morality, a theory of human nature, a rival ideal. It has the advantage of independence. Its weakness is that those criticised have no reason to accept the standard, and the critic must justify it from nowhere.

Internal criticism holds a community to the values it professes. You say you believe in charity, and you turn away the stranger. Its strength is that the criticised already accept the standard. Jaeggi finds three weaknesses. The professed values need interpreting, and people disagree about what they demand. The method is conservative, since it can only ask for a return to ideals already held, and those ideals may be poor ones. And it assumes that a community has one consistent set of values, which is seldom so. It also depends on goodwill. The inconsistent party can simply mend their ways, or change their professed values.

Immanent criticism is the form she defends. It comes from Hegel and Marx, and it has a parallel in psychoanalysis. It also starts from norms that belong to what is criticised. It differs from internal criticism in seven respects, and since the next chapter uses each of them they are given in order.

  1. It finds norms in practices, and stated beliefs are secondary. The standard is what participants are already committed to by taking part, whether or not they say so or know it.
  2. The norms are constitutive. They are not optional extras that the practice happens to profess. Without them the practice would not be what it is, and could not function.
  3. The norms are effective, and effective in an inverted way. This is the key difference. Internal criticism says an ideal is not being realised. Immanent criticism says the norm is realised, and that its realisation produces results that turn against it. Jaeggi's example is the labour contract. Freedom and equality are really present in it, since both parties contract freely as equals. And that very arrangement produces dependence and inequality.
  4. It is oriented to crisis. Such contradictions are not just inconsistencies to regret. They show up as practical trouble: things do not work, people are blocked. Internal criticism relies on goodwill. Immanent criticism points to structural difficulties that cannot be fixed inside the existing arrangement.
  5. Both sides must change. The norm does not stay pure while reality is brought up to it. A norm that could only be realised in a self-undermining way is itself deficient and must be rethought. The yardstick changes in the measuring.
  6. It is transformative. It does not aim to restore a lost harmony. It aims at something new.
  7. It is a learning process. The new position contains the experience of the old one's failure. Hegel's term is determinate negation. The result is richer than what it replaces because it has been through the failure.

Jaeggi is candid about the difficulties. It can be hard to show that a contradiction is really in the object and has not been constructed by the critic. And the approach seems to assume that history moves toward something. Her reply to the second is to drop any final goal and keep only the idea of learning.

Learning and Its Blockages

If there is no ideal form of life to measure against, what makes one better than another? Jaeggi's answer is negative and formal. A form of life is going well when it can deal with the problems it meets and sets itself, and can change as a result. It is failing when it cannot. In her words, a failing form of life suffers from a collective practical reflexive deficit: a blockage to learning.

She takes the outline from Dewey. A problem-solving process succeeds when it actually resolves the difficulty and when, in doing so, it deepens and widens the capacity for further experience. The maxim, which Hilary Putnam took from Peirce, is: do not block the path of inquiry. A process fails when it narrows experience or protects itself against new experience.

Two kinds of blockage are distinguished. Some are simple: old habits of interpretation that have not caught up with changed conditions. Others are systematic. They concern the categories in which experience can be had at all, and they resemble what an older tradition called ideology. In such a case people do not merely lack information. Their way of framing things prevents the relevant experience from registering.

She gives a homely example of a learning process that fails although something appears to have been learned. A German man conscripted as a boy at the end of the war concludes, in the 1950s, that politics brings only misfortune and that a decent man keeps to his family. He calls this the lesson of experience. His children reply that he has learned the wrong thing, and more: that his "experience" was itself a distortion, a way of not facing what happened. The process, and not only its result, was deficient.

Against Abstinence

The book opens with an argument against what Jaeggi calls ethical abstinence. Liberal political philosophy, in John Rawls and Jürgen Habermas, holds that public reasoning should confine itself to questions of justice and leave conceptions of the good life to individuals. Forms of life become a black box.

Her objection is that the abstinence cannot be kept. Forms of life are always already shaped by public arrangements: infrastructure, law, markets, technology. Decisions that bear on how people can live are taken constantly, by institutions that profess neutrality. To decline to evaluate forms of life is to leave that shaping to proceed unexamined. Neutrality does not mean that nothing has been decided. It means the decisions were taken where nobody can be asked about them.

Experimental Pluralism

Jaeggi's conclusion is not that there is one right way to live. Her position is an experimental pluralism. No one yet knows a fully satisfactory form of life. Many attempts are needed, since their variety raises the chance of good solutions, and since consequences cannot be worked out in advance and must be learned from. But the attempts are comparable. Some deal with their problems and learn. Some are regressive and can be shown to be so by their inability to do either. She explicitly rejects the idea that every form of life has a right of its own that puts it beyond comparison.

The Bearing on Anthro-Complexity

Four connections should be stated before the application begins.

Anthro-complexity is an ensemble of practices in Jaeggi's sense, on a small scale. It consists of practices that hang together and give one another sense: the Cynefin session, narrative capture, signification, the pattern workshop, constraint mapping, the portfolio of probes. It has attitudes and interpretations: the Pelagian stance, suspicion of idealised futures. It is sedimented in software, licences, training and a vocabulary. And it has internal norms, so that one can do it well or badly. A professional practice is narrower than a family or a city, and Jaeggi's concept is being stretched. Her own treatment of professional ethos shows that the stretch is reasonable.

It solves problems, and they are second-order. Anthro-complexity exists as a solution to a problem: how to act in human systems where prediction fails, without treating people as parts of a machine or as ants. Its difficulties, if the eight theses are right, do not strike it from outside. They arise from its own claim, the claim that people change their world by describing it.

The disputation has been immanent criticism from the start. Its method, announced in the Preliminaries, was to grant anthro-complexity's premises and follow them. The grant-and-follow exercise does exactly that. It brought no rival ontology. Jaeggi's account explains why this is a legitimate and strong kind of criticism, and it also imposes obligations that the disputation must now meet. It must show that the contradictions are in the practice and not in the critic's construction. It must show that they produce practical trouble. And it must accept that its own standard will be changed in the process.

Snowden's approach and Jaeggi's are near relations. Both are pragmatist. Both refuse an ideal end state. Jaeggi's "growth without a fixed end" is close to the vector theory of change. Both value many experiments. A criticism of anthro-complexity from Jaeggi is therefore a criticism from within the family, which is the only kind that the trialectic's author would count as rigorous.

There is also one point of friction, and it should be marked now. Anthro-complexity declines to say what a good organisation is. It offers directions and "more stories like these". Jaeggi would recognise that as a form of abstinence, and Chapter 53 asks whether it can be kept.

A PRACTICE IS MEASURED BY ITS OWN CONSTITUTIVE NORMS —

AND BY WHETHER IT CAN LEARN FROM FAILING THEM —

DO NOT BLOCK THE PATH OF INQUIRY

Addition: Jaeggi at Greater Length

The chapter above gave Jaeggi's argument in outline. Six parts of it deserve fuller statement, because each gives the disputation a sharper tool than it had.

What a practice is

Jaeggi builds up from the smallest unit. A practice is a sequence of actions and gestures, done more than once, for which a pattern exists. Her example is paying at a supermarket: finding the end of the queue, placing goods on the belt, a word with the cashier, the purse, the change, the bags. Four features matter.

A practice is repeated, so it runs on habit and on know-how that those involved cannot fully state. It is social, even when done alone, because the pattern is shared and can be recognised by others. It has a purpose internal to it, which is how one tells doing it from failing at it. And it is understood as something: the same movements count as queueing only where queueing is understood.

A form of life is a bundle of such practices that depend on one another. The dependence is what makes criticism possible. One can ask whether the practices in a bundle fit: whether what one of them requires is undone by another.

For anthro-complexity. Its practices can be listed with the same plainness: a prompt is issued, a story is typed, a mark is placed, dots are displayed, a room reads them, a probe is designed. The question of fit can then be asked. One practice asks tellers to interpret for themselves. Another has their interpretations read elsewhere by others. Whether those two fit is the matter of Chapter 50.

How large a thing is being criticised

Jaeggi asks how stable and how deep a bundle must be before it counts as a form of life. She tests the concept against fashion. A fashion is constitutively unstable and lives by distinguishing its wearers from others. A form of life is geared to lasting, and it makes a claim to be an appropriate way of dealing with something real, in the way that weatherproof clothing must suit the weather. She calls her concept modular. Forms of life come in many sizes and nest inside one another.

For anthro-complexity. The test is useful because management thought has fashions, and Snowden has resisted becoming one. By her criteria anthro-complexity is more than a fashion. It is geared to lasting, it claims adequacy to something real, and it has norms. It is a small form of life within the larger one of consultancy, and it inherits some of the larger one's problems, which is why the pressure of the market appears in its own author's account.

Why the norms are not conventions

Three ways of justifying a norm are distinguished. A conventional justification says: we agreed on it, and could as well have agreed otherwise. Driving on the left is like that. A functional justification says: without this the practice does not work. An ethical justification says: this is what makes the practice good.

Her claim is that the norms of forms of life are never merely conventional, and that in them the functional and the ethical are joined. Her example is a doctor who insists on examining patients thoroughly although the finances press against it. She is not following a club rule. She holds that only careful examination achieves what medicine is for. A colleague who cuts corners is, in her eyes, a bad doctor, and not merely a doctor with other customs.

For anthro-complexity. This gives a way to read the norms that the trialectic and the Luhmann post announce. That the originator signifies first is no house rule of one consultancy. On Snowden's account it is what makes the practice work and what makes it right, both at once. A norm of that kind can be used to judge the practice that holds it, which is what immanent criticism needs.

Problems are not needs

Jaeggi could have said that forms of life exist to satisfy needs. She declines, because talk of needs suggests a fixed list against which arrangements can be scored. Needs are shaped by the forms of life that meet them. She quotes Marx's remark that hunger met with cooked meat and a knife and fork is a different hunger from that which tears raw flesh. Problems, in her sense, are historical and interpreted. They arise for a particular way of living, in its own terms.

For anthro-complexity. The distinction tells against one familiar way of presenting any method, as the answer to an organisation's needs. It also supports Snowden's distrust of deficit. A client's stated need is an interpretation, made with the descriptions available, and the trialectic says as much when it warns that the presenting problem is not always the diagnosis.

Two ways to fail

She distinguishes two failures. A particular instance can fail to live up to its kind, as a given family may fall short of what the modern family claims to be. Or the kind can fail, because what it demands cannot be lived. The second is the serious one. Both are failures by the thing's own measure.

For anthro-complexity. The disputation should say which it alleges. Theses IX and XI aside, most of its criticisms are of the first sort: particular arrangements fall short of the practice's norm, and better arrangements are possible. It does not claim that anthro-complexity cannot be lived. The reconstruction of Book Eight assumes that it can.

How a tradition learns, and how it stops

The second half of the book draws on Alasdair MacIntyre, read as a theorist of how traditions change. Three points are taken from him.

Two kinds of crisis. In the ordinary course a tradition meets problems and finds the means to solve them within itself, changing as it does so. Sometimes it meets a problem that cannot be solved in its own terms, and by its own standards it stops making progress. MacIntyre calls this an epistemological crisis. What is then in question is the framework by which solutions are judged.

What counts as coming through. A tradition comes through such a crisis rationally if three things hold. The new position solves what the old could not. It explains why the old could not. And it can tell the story of the change in a way that shows continuity with what went before. Jaeggi calls the third narrative reintegration.

The worst case. MacIntyre's charge against modern liberal culture is that it has made itself a tradition above traditions, which cannot be criticised from anywhere, and has thereby destroyed the means of its own repair. That is a blockage to learning of the deepest kind.

For anthro-complexity and for this book. The three tests can be applied to the trialectic as an episode. It solved what the binary could not, by supplying a third position. It explained why the binary could not, since two terms leave only a choice or a mean. And it kept continuity, by retaining the binary's diagnostic use. That is a rational passage through a small crisis.

The worst case is a warning to both parties. A school that treats every other school as a symptom of a faulty substrate has made itself a tradition above traditions. So has a critic who treats every defence as evidence of the fault he is looking for. The reconstruction proposed in this book must meet the same three tests. It must solve what the present instruments cannot. It must explain why they cannot, which Chapters 50, 52 and 56 attempt. And it must be continuous with anthro-complexity, which is why nothing in it comes from outside what Snowden already holds.

SOLVE WHAT THE OLD COULD NOT —

SAY WHY THE OLD COULD NOT —

AND TELL THE CHANGE AS ONE STORY

Chapter 50

Thesis IX: The Formative Encounter Under Immanent Criticism

2,840 words, about 12 minutes of reading

The Task

Chapter 48 ended with a narrowed question. The trialectic sets a norm for work in complex situations: the encounter forms both parties, capacity belongs to the coupling, and the practitioner must be willing to be changed. Do anthro-complexity's instruments embody that norm? Chapter 49 supplied a method for answering. This chapter applies Jaeggi's seven features of immanent criticism in order.

A word on fairness first. The trialectic is a blog post of a few thousand words, written to open a conversation. It ends by asking for commentary. It should not be read as a finished doctrine. What follows is offered as the commentary it asks for.

The thesis of this book (Thesis IX). The formative encounter is anthro-complexity's own best statement of the human difference in practice, and it is a constitutive norm of work in the complex domain. As that norm is realised in the present instruments, it is inverted. The encounter that forms is moved from the meeting of persons to the meeting of a centre with a pattern, and those who supply the pattern are not party to it. The trialectic also keeps one thing constant across all three of its positions, the mediating figure, and so does not yet pass its own triquetra test.

Feature 1: The Norm Is in the Practice

Immanent criticism does not rely on what a community says it believes. It looks for what participants are committed to by doing what they do.

Is the formative encounter such a norm, or only one author's reflection? The evidence is that it is in the practice. Anthro-complexity's methods make no sense without it. One probes because response cannot be known in advance. One collects stories continuously because the system changes as it is engaged. One refuses the ideal future state because what becomes possible depends on what happens on the way. The whole method presupposes that engagement produces what could not have been specified beforehand. The trialectic names what the practice already assumes.

So the first feature is satisfied. The standard used in this chapter is taken from the practice and is acknowledged by its author.

Feature 2: The Norm Is Constitutive

A constitutive norm is one without which the practice would not be the practice.

Take away the formative encounter and see what is left. A practitioner who is unchanged by the engagement, working on a system whose capacities are fixed, is either supplying what is missing or releasing what is there. Those are the first two positions. Snowden's own judgement on a practice confined to them is severe. The deficit mode is the error the whole scheme was built to expose. And the capacity mode alone, he says, imports the assumptions that removing constraints is enough, that the practitioner can step back, and that emergence manages itself. He adds that a stable external perspective treats the complex as merely complicated.

In Cynefin's terms, then, a practice without the third position has left the complex domain. But the complex domain is what anthro-complexity is for. The norm is constitutive.

Feature 3: The Norm Is Realised, and Inverted

This is the centre of the argument. Internal criticism would say: anthro-complexity professes the formative encounter and fails to practise it. That is not the charge. The charge is that the norm is really at work in the instruments, and that as it works it produces effects that run against its own content. Four cases follow. Each takes a real merit and shows what the merit produces.

Self-signification and the relocated frame

The norm at work. Tellers interpret their own stories. The practitioner does not impose a reading. This is a real transfer of interpretive standing, and it is the Pelagian mode honestly realised.

The inversion. For tellers to interpret at scale, they must interpret in a common form. The form is the signifier set, designed beforehand. So the practitioner's reading has not been removed. It has moved upstream, into the dimensions that all later interpretation must use, where it is harder to see and to dispute than an analyst's coding would have been. A Web of Relations calls this the relocated frame. The more faithfully self-signification is carried out, the more completely the designer's frame governs.

The trialectic is candid about the neighbouring step. It says that the choice of the area of inquiry "may be Augustinian in nature or motivation". That concedes that the first mode is present at the origin of every deployment. The signifier set belongs with it.

Disintermediation and the new intermediary

The norm at work. The deficit model is condemned because it maintains an intermediary whose authority rests on exclusive access. Anthro-complexity's instruments aim to remove interpretive layers between people's experience and those who decide.

The inversion. Removing the interpreting consultant requires an apparatus: software, a signification framework, a method for reading patterns, trained practitioners, a licence. The apparatus stands between tellers and deciders. Access to it and competence in it are not evenly held. An intermediary remains, and it is now an instrument with an owner. The channel for grace has been rebuilt in another material. This is no accusation of hypocrisy. It is what happens when disintermediation is carried through by means of a designed instrument.

The encounter and its parties

The norm at work. In the complex domain, what matters arises in the coupling of practitioner and system, and both are formed.

The inversion. Ask who meets whom. A SenseMaker deployment arranges three sets of meetings. Tellers meet a prompt and a screen. Practitioners and sponsors meet a pattern, in a workshop. Practitioners and sponsors meet each other. Of these, the second and third are encounters in which something unforeseen can form between parties who answer one another. The pattern workshop is often exactly that, and people who have attended one say so. But the tellers are not in it. Their part is the first meeting, and a screen does not answer.

So the formative encounter is real, and it takes place between the centre and a representation of the many. The many supply the material for an encounter that forms other people. The better the instrument works, the richer that central encounter becomes, and the less need there is for the practitioner to meet the tellers at all.

Legibility and the receding encounter

The norm at work. The trialectic describes the course of an engagement. Early on, the practitioner does not know what the system contains, and the formative mode is most present. As the work goes on and "the landscape becomes more legible", the capacity mode takes over, because she can now see what is there and work with constraints.

The inversion. On this account the encounter is a phase that knowledge brings to an end. The practitioner is open to being formed while she is ignorant. As the instruments do their work she becomes a reader of a legible landscape, and the relation becomes one of seeing and arranging. The word legible is James C. Scott's word for what a centre does to a population. The same paragraph of the post says that with these methods "you know where to look for emergence". An emergence whose location is known in advance by one party is not held by neither.

In each case the pattern is Jaeggi's third feature. The norm is effective. Its realisation through these instruments produces a frame, an intermediary, an exclusion and a closure that the norm itself condemns.

Feature 4: The Contradiction Shows as Crisis

Jaeggi requires that an immanent contradiction be more than a logical embarrassment. It must produce practical trouble.

The disputation has already described the trouble, as Thesis VIII, without then having the trialectic to explain it. People who supply stories to an encounter they are not part of draw conclusions about what the stories are for. They adjust. Significations converge, outliers thin, and stories increasingly concern what the last round led to. The instrument goes on producing clear patterns that mean less. That is a crisis of function, and it is produced by the exclusion identified above. An encounter that included the tellers would give them less to guess at.

There is a second symptom, which the post itself reports. Snowden says that the encounter mode is often available and not taken because it feels professionally risky, and that the market rewards the deficit mode. A practice whose constitutive norm is regularly declined by its own practitioners, for reasons built into how the practice earns its living, is in the condition Jaeggi describes: the difficulty is structural, and goodwill will not remove it.

Feature 5: The Norm Itself Must Change

Internal criticism leaves the standard untouched and asks reality to catch up. Immanent criticism holds that a norm which realises itself in inverted form is itself deficient. So the formative encounter, as stated, must be examined too. Three deficiencies appear.

It is a dyad

The formula is practitioner and system, two parties. But "the system" is a collective noun for many people who stand in different positions and in relations of unequal power to one another and to the practitioner. One cannot have an encounter with a system. One has encounters with persons, some of whom commission and some of whom are surveyed. The dyadic formula lets the commissioning few stand for the whole in the encounter while the whole supplies the data. Hobbes's point about representation applies: the unity of "the system" as a party is produced by whoever speaks for it.

There is an irony here. The post was written to cure a binary by adding a third term, and its third term is stated as a relation between two.

It is still about where capacity lives

All three positions answer one question: where does the capacity for change reside? In the practitioner's supply, in the system's store, or in neither. The third answer denies possession, which is a real advance on the grammar criticised in Thesis I. But it keeps the question, and with it the noun. "Capacity held by neither" is still a capacity, a something whose location is at issue. Mead and Elias would decline the question. There is the process of people responding to one another, and what it can do is not a stock located anywhere, including in "the coupling".

The mediating figure is constant

Snowden says that the three theologians represent distinct ontological registers in their account of where capacity resides, how change happens and "what the role of the mediating figure is". In all three there is a mediating figure. The basilica has its priest at the altar. The monastic enclosure has its monks within the wall. And the Byzantine church, the Maximian space, has an iconostasis. The post describes it as marking a threshold. It is a screen. It divides the sanctuary, where the clergy act, from the nave, where the people stand. The image chosen for the formative encounter contains, at its centre, the partition between those who officiate and those who attend.

This matters for the triquetra test. Cassian failed it because he stayed on the line between two poles. A third term must be irreducible to the frame the first two share. The frame that Augustine and Pelagius share, as Snowden uses them, is the practitioner's question: what is my role with respect to this system's capacity? The formative encounter answers that same question more subtly. It does not leave the frame. A position that did leave it would ask about the encounter of the system's members with one another, at which the practitioner is one participant or is absent. Nāgārjuna was dismissed for dissolving the frame. The members' own encounter does not dissolve it. It adds the parties that the frame left out.

Feature 6: What the Transformation Would Be

Immanent criticism aims at something new, in which both the practice and its norm have changed. In outline the change is already implied by what has been found.

The norm restated. The formative encounter is among persons. Its parties include those who tell, those who commission and those who convene. What forms belongs to none of them and is visible to all of them.

The practice changed accordingly. If tellers are parties, then the pattern returns to them before it is acted on, they read it and their reading is recorded, and the sponsors' reading is shown to them. If the frame is part of the encounter, then tellers share in designing the signifiers and can say what the set leaves out. If the practitioner's formation is part of the work, then her expectations are declared beforehand and her own stories are signified with the rest.

These are the protocols of Chapter 45: return before action, the narrators' reading, the fourth vertex, the declination table and the reading of the reader. They were derived there from the eight theses. They can now be derived from Snowden's own third position. That is some evidence that the contradiction is in the object and has not been constructed by the critic, which is the first of the difficulties Jaeggi admits.

Feature 7: A Learning Process

The last feature is that the movement is a learning process, in which the new position keeps the experience of the old one's failure.

The trialectic is itself an instance. Snowden held a binary. Readers objected. He did not defend it or abandon it. He worked out what a third term would have to be, tested candidates, chose one, and stated its limits. That is determinate negation carried out in public, and the practice shows by it that it can learn.

The disputation must apply the same standard to itself. Its first thesis, as originally stated, said that anthro-complexity presumes persons who possess capacities. The trialectic shows that at the level of practice Snowden has denied possession outright. Thesis I therefore stands as a criticism of the wording of the human cluster and of the instruments' unit of collection. It does not stand as a criticism of his account of practice. Chapter 54 records that change and the others.

Grant and follow: granting the formative encounter.

Granted: in the complex domain, capacity exists only in the coupling and is held by neither party in advance.

Then it must be granted that what a deployment of an instrument makes possible was not in the practitioner, the sponsor or the tellers beforehand.

Then it must be granted that the tellers are a party to the coupling, since without them there is nothing.

Then it must be granted that a party to a coupling who cannot see what the coupling produced, or answer it, has been coupled as material and has not met anyone.

Then it must be granted that an arrangement in which the many are coupled as material and the few are formed is the deficit model with respect to the many: something is taken from them and something is later supplied to them.

Therefore the formative encounter, taken at its word, requires that the pattern return to those who made it and that their answer be heard. Without that, the third architecture is built on the plan of the first.

For the opposing side. Four replies are available, and they are strong.

First, the post never claims that a SenseMaker deployment is a formative encounter with every teller. It says that a deployment traverses the three modes. Mapping is one thing and intervention another, and the co-creative work is done in designs such as entangled trios, with real people in a room.

Second, scale. Ten thousand people cannot be met. An instrument that lets each of them interpret their own experience and be heard by those who decide is far more than they had. The critic compares it with an ideal and not with the alternatives.

Third, nothing prevents return. Good deployments do feed patterns back, and the method is revised continually.

Fourth, the mediating figure cannot be removed. Someone convenes. To pretend otherwise is the baggy error, and it hides power where the trialectic at least names it.

The disputation's answer. The first reply is accepted as a correction of emphasis. The claim here concerns the mapping stage, and it is at that stage that the dimensions are fixed and the many are involved. If the encounter begins only afterwards, with a selected few, the point about parties stands. The second is accepted in full as to merit. The comparison made here is with the norm that anthro-complexity sets for itself, which is how immanent criticism proceeds. The third is welcome, and the request is that return become a requirement of the method. At present it is a mark of the better deployments. The fourth is accepted: someone convenes. That is why the convenor should be on the map and visible to all parties.

THE NORM IS SNOWDEN'S OWN —

IT IS REALISED, AND REALISED UPSIDE DOWN —

THE ENCOUNTER FORMS THE CENTRE, AND THE MANY SUPPLY IT

Addition: Thesis IX in Cards

Chapter 51

A Web of Relations: Designer, Dweller and Formation as the Bridge

4,311 words, about 19 minutes of reading

A Second Witness

The disputation has drawn on the process schools since Chapter 11: Stacey, Griffin, Shaw and Mowles, with Mead and Elias behind them. It has not so far drawn on the fullest statement of how those schools answer anthro-complexity directly. That statement is in A Web of Relations, the web edition of Kevin Flinn's work on complexity and organisational development, which Mark Downham edits. Flinn's own book, Organisational Development in Practice: A Complexity Approach (2025), stands behind it.

The web is a very large and deliberately many-voiced work. Its claim, in its own phrase, is that relation is the real. It rests on four pillars: Elias, Mead, Dewey and the group analyst S. H. Foulkes. Three of its parts bear on this book. One sets anthro-complexity and the process schools against each other and keeps a ledger. One records the process schools' replies to the trialectic itself. One proposes formation as the answer to a difficulty that both sides share. This chapter takes them in turn and then asks what they add to the immanent criticism of Chapter 50.

The Fault Line

The web's first finding is how much the two sides share. Both reject the bounded rational individual. Both hold that "the system" is not a noun. Both reject the enlightened observer, both accept unintended consequences as a law, both refuse the management of outcomes, and both hold that meaning is embodied and material. Both claim Peirce. Both can cite Peter Checkland's rule that one should say a situation "may be described as" a system and never that it is one.

They part on one precise question. Can a practitioner stand at the edge of an interaction and design the conditions under which it senses itself? Or is that design the objectifying standpoint coming back by another door?

The web gives each side a name, and the names are useful. Snowden is the Designer of Conditions. The process schools are the Dweller in the Conversation. It then states the charge each makes against the other.

  • The Designer's charge. The process schools cannot work at scale, and their account of emergence invites an alibi: whatever happens, it emerged.
  • The Dweller's charge. To design conditions for others is to reoccupy the position above them. The privileged standpoint has not been given up. It has been moved into the apparatus.

The web's judgement is that both charges are serious and that neither has been decisively answered. It adds that Snowden half-concedes the second.

The Ledger

The web keeps accounts for each side, and this disputation should adopt them, since they are more even-handed than its own opening chapters were.

What anthro-complexity pays. It pays on standpoint, for the reason just given. It pays on the power of method, which Snowden has conceded: no method wholly avoids becoming a new site of narrative power. It pays for a residue of cybernetic vocabulary. And it pays by concentrating accountability at the moment of design.

What the process schools pay. They pay on scale. They pay on purity, since their conferences and groups are structured occasions that disavow their structure. And they pay on accountability, since it is hard to say what the practitioner contributed.

What both pay. Each has declined the encounter with the other that could test whether their difference is real. The web applies Snowden's own discipline here. His aporetic domain asks practitioners to discern whether an apparent incommensurability is true or only badly framed, and warns against accepting it too soon. By that standard, the refusal of the two schools to meet is a failure of discernment on both sides. If the difference does prove real, the prescribed course is what Snowden calls choreographed coexistence, with each side acknowledged as a legitimate pole.

Four Things Snowden Has Already Said

The web records positions of Snowden's that this disputation had not credited. They belong on the record.

Entangled but not dissolved. His phrase for the practitioner. It rejects the view from above and equally rejects the merging of practitioner into field. It is the same guardrail that the trialectic places on the formative encounter, and it matches the emergent personhood of Chapter 2's addition.

Emergence as alibi and emergence as discipline. He distinguishes two uses of the idea. As an alibi, emergence excuses: nobody could have known. As a discipline, it does the opposite. Because outcomes cannot be predicted, responsibility concentrates on the design of conditions, and the designer answers for it. The web calls this the ethical centre of his position. It answers part of what Chapter 2 said about classification as an act that dissolves blame.

Method as accountability. His reply to the Dweller's charge is that a specified method makes the practitioner's contribution visible and contestable, where an unspecified presence does not.

The conceded power of any method. Noted above.

There is also a warning, which the web calls the substrate argument. Snowden holds that rival traditions fail because of the ground they stand on and that skill cannot repair it. The computational tradition cannot tell information from meaning. The psychodynamic tradition rests the practitioner's authority on privileged access to what is hidden. The web observes that an argument of this form can protect its author from criticism, since any objection can be read as a symptom of the objector's substrate. Chapter 53 returns to that.

The Replies to the Trialectic

The web sets out four responses from the process traditions. Each bears on the third position.

The family quarrel: complex responsive processes. The formative encounter is where the two sides come closest. A process theorist would say that Snowden has arrived at what Mead and Elias described: nobody stands outside, and what happens belongs to the interaction. The quarrel that remains is whether one can hold that and still design instruments. The web's formula is that the dispute has narrowed to one question: is a self-modelling instrument legitimate?

Mowles: structure as hidden power. Chris Mowles's counter-challenge is to start with experience. Imposed structure, in his view, conceals the power of whoever imposed it and dampens the spontaneity it claims to release. He would ask of the trialectic who decides which mode is doing the work at a given moment.

Rodgers: talking instead of systems. Chris Rodgers holds that organisation is conversation and that nothing sits above the interactions. His objection to dispositional mapping is put in a phrase the web preserves: such instruments are nets, and a net cannot catch what was never stored. Dispositions do not sit in the system waiting to be collected. They are made in the asking.

Roy: what is possible from here. Bonnitta Roy revives a phrase Stacey abandoned, complex potential states, in the terms of process philosophy. Her proposal is to replace the adaptive question, what do we do now, with another: what is possible from here? That is close to the adjacent possible, and it asks the question from inside.

A fifth voice is Mowles again, in what the web calls his move from complex responsive process to complex phronetic process. The earlier work described what is happening. The later asks what should be done. It names capabilities of practical wisdom, among them judging what is good in the situation, creating shared context, exercising power well and fostering practical wisdom in others. It describes three moves: intervening obliquely, simplifying thickly so that a group can act without being paralysed, and judging without a handrail. Snowden and Mowles both put practical wisdom at the centre. The web's observation is that Mowles locates it in the immersed participant, and Snowden in distributed, disciplined sensing.

The Third Charge and the Bridge

The web brings three charges against its own tradition. The third is the one that matters here. It says that the method cannot reach the ontology. A room in which people converse reflectively is a small thing. The theory is about a whole organisation. How does what happens in the room get to the corridor?

That is Snowden's scale objection, made by the process school against itself. The web's answer is the proposal it calls formation.

Facilitation is something done to a meeting. Formation is something that happens to a practitioner over years: the re-forming of perception, bodily habit, judgement and sense of self, in the company of others and under supervision. The web's claim is that formation is the bridge. The room cannot reach the corridor. A person formed in the room walks into the corridor and perceives and acts differently there.

The web describes what is formed, drawing on its four pillars and beyond.

  • From Foulkes: the practitioner as a nodal point in a matrix of communication, with no outside to stand on. The group analyst's conductor tends the medium and does not direct it. The practitioner also has to contain the anxiety that an unstructured gathering releases, since agendas are among other things defences.
  • From Bourdieu: habitus is what formation re-forms. It is the source of a practitioner's perceptiveness and also her conservatism. It changes through the slow work he called socio-analysis of oneself.
  • From Mead: the practitioner is already in the others, and they in her, by the conversation of gestures.
  • From the Boston group of infant researchers and psychotherapists around Daniel Stern: change in a relationship happens at charged now moments which, if met authentically, become moments of meeting that reorganise it. Such meetings are improvised and untidy and cannot be scheduled.
  • From Elias: involvement and detachment as a balance achieved inside the figuration.

The web calls the resulting practice post-facilitation. It works more by the quality of the practitioner's presence than by technique, and it refuses the view from above. It has disciplines of attention, which the web names with old words: watchfulness, discernment, and the reading of how rooms and buildings carry meaning. It has a discipline of return, by which the practitioner comes back from the encounter and gives an account.

The web is honest about what this does and does not supply. It keeps a ledger for itself as well. Supplied: an object of attention, a kind of act, an orientation, and a guard against paralysis. Still owed: a demonstration that formation scales, and a test of whether a change is real. To Snowden's demand for a specification it replies that formation is no procedure. To the objection that the practitioner's "gesture" is mysticism or manipulation it offers a criterion taken from the Chinese sources: the act must lean into a propensity the field already has, and must not install one. The objection from scale it concedes as the hardest. Its three partial answers are that the formed form others, that some positions in a figuration carry more weight than others, and that themes travel.

Maximus Again

The web has its own use for the theologian of Snowden's third position, and it helps with the difficulty Snowden admits.

Snowden's worry was that Maximus's synergeia can be read as two wills, each with its own capacity, cooperating. He wants something stranger, and he wants a guardrail: both parties remain real. The web draws on a neighbouring idea from the same tradition, perichoresis, mutual indwelling. In Greek theology it names a relation in which persons are fully in one another "without confusion". The modern Orthodox theologian John Zizioulas builds on it the claim that personhood is constituted in communion, and that deeper unity yields more distinctness, not less. The web applies this to practice. The practitioner is "in everyone else" and becomes more distinctly herself by it.

Whatever one makes of the theology, the structure is the one Snowden asks for. It is not two prior capacities cooperating, and it is not dissolution. It is also plural. Perichoresis has more than two parties. That is the correction Chapter 50 asked for under its fifth feature, where the formative encounter was found to be a dyad.

What the Web Adds to the Immanent Criticism

It confirms that the contradiction is in the object. Chapter 50 argued that self-signification relocates the frame. The web reaches the same finding independently and reports that critics of SenseMaker had named it. It reports too that Snowden grants the premise, that method is a site of power.

It supplies what the reconstruction lacked. Chapters 44 and 45 are all protocol: tables, steps and tiers. The web's argument is that protocol is not enough, and Snowden's trialectic says the same when it calls the practitioner's formation part of the work. A declination table filled in by an unformed practitioner is a form. The reconstruction needs a thirteenth principle, which Chapter 55 states: the practitioner is formed, in company, and returns to give an account.

It gives the disputation its own charge to answer. The Designer's objection to the Dweller applies to this book's proposals with full force. Mutual dashboards and return before action are slower and dearer, and nobody has shown them at the scale of ten thousand. The web's honesty about what is still owed should be matched. Chapter 45 calls scale a design problem. That is a promissory note.

It shows the dialectic in its real shape. The opening books of the disputation set anthro-complexity against a relational alternative, as thesis against antithesis. The web shows something better. Each side holds what the other lacks and each pays for what it holds. The Designer has reach and accountability and pays in standpoint. The Dweller has standpoint and pays in reach and accountability. The trialectic and the web's formation are the two sides moving toward each other from opposite ends: a designer who admits he must be formed, and dwellers who admit their rooms must reach the corridor.

For the opposing side. A practitioner of anthro-complexity may answer the web as follows. Formation is the Augustinian model returned in its purest form. It requires years in a community under supervision, at the end of which a person is fit to practise. That is a priesthood with a novitiate. The many have no access to it. Method, by contrast, can be handed to anyone. A triad can be signified by a cleaner on a telephone. If the question is which approach puts capacity in the hands of ordinary people, the instrument wins.

The disputation's answer. The point is strong, and the web itself reports a version of it: Snowden's critique of the Augustinian substrate has been turned on Mowles, whose skilled facilitation can be described as grace through a finer channel. Two things may be said. First, the objection shows that neither side escapes the mediating figure, which was the finding of Chapter 50. The instrument has its designers and the room has its conductor. Second, the two remedies are complements. Method without formation produces frames nobody examines. Formation without method produces wisdom that cannot leave the room. The instrument that anyone can use should be designed and read by people who have been formed, and both the instrument and the forming should be open to those on whom they are used.

THE DESIGNER HAS REACH AND PAYS IN STANDPOINT —

THE DWELLER HAS STANDPOINT AND PAYS IN REACH —

FORMATION IS THE BRIDGE, AND RETURN IS ITS TOLL

Addition: The Web at Greater Length

The chapter above drew on three parts of A Web of Relations. This addition draws on more of it, in three sections. The first reports what the web says Snowden's 2026 series of posts contains, since that series is wider than the two posts this book has read directly. The second sets out the web's own ground: the legitimate and the shadow system, the living present, and what it takes from each of its pillars. The third describes the instruments and methods the web itself admits, since a school that criticises instruments should be asked about its own.

One caution governs the first section. The disputation has read "Not a branch of the tree", the trialectic post, the Luhmann post and "Knotty problems" in full. The other posts of the series are known to it only through the web's account. What follows is a report of a report, and is marked as such.

What the web reports of the 2026 series

The system cannot be a noun. The web reports that the series opens with a provisional definition built on a carpenter's image: one reads the grain of the wood and does not impose a shape on it. Human systems are part of nature and irreducibly different. Meaning is enacted between people, in bodies, stories and practices. So the practitioner asks what the system is doing, and does not ask what it should become. The system cannot be a noun standing outside the inquiry. The web notes that the post itself says process sociologists share this ontology.

Gell-Mann's dictum. Natural science is necessary, as an enabling constraint that keeps the account honest. It is insufficient, and the deeper reason is given as reflexivity: modelling from outside fails for a system that changes when it is described. The web reports Snowden's use of a saying attributed to Murray Gell-Mann, that the only valid model of a human system is the system itself. The conclusion drawn is that methods should let the system model itself, and SenseMaker is the instrument offered.

Three I's. The human difference is stated as intelligence, identity and intention. The web's summary of the first is the important one for this book. Intelligence is reflexive abstraction, so an external framework becomes an actor inside the system it describes. Identity is contested and relational, so every intervention moves the landscape of identity. Intention can act against incentive for the sake of principle. The web adds that Snowden admits the line may not cleanly exclude some animals and treats the difference as one of degree that still calls for new tools.

Three S's. Sentience: feeling is part of how a signal is processed, so the same content met in fear and in curiosity is not the same. Sapience: practical wisdom, which grows through situated engagement and cannot be transferred by training. Salience: what comes to the foreground for a person, shaped by identity and history, so that no single observer holds the whole.

The granularity triad. Three operations: go to a finer grain so that variety multiplies, distribute cognition across many interpreters, and remove interpretive layers between experience and those who decide. Each is aimed at an orthodoxy: aggregation, forced consensus and hierarchy.

The practitioner paradox. The web reports that the series faces the sharpest objection itself. If all perception is shaped by salience, the practitioner's reading of the grain is one more reading. The answer has three parts: distributed sensing, reflexive attention to the practitioner's own situation, and a diversity of perspectives so that none dominates. The practitioner is the one who creates conditions for collective seeing, and is not the one who sees most clearly. And the web records a striking sentence about the two closing images. The apprentice's wood differs from the estuary because it knows it is being worked.

What this does to the disputation

If the web's report is accurate, more of this book's argument was already Snowden's than Chapters 48 and 52 allowed.

Deeper water: four further credits.

Thesis II said the asymmetry principle applies to anthro-complexity's own frameworks. On the web's report, the first of the three I's says that an external framework becomes an actor inside the system it describes. That is the thesis.

Thesis I said the human marks are worded as possessions. On the web's report identity is described as contested and relational.

Thesis III said the estuary pictures a medium that does not know it is read. On the web's report Snowden himself sets the estuary beside wood that knows it is being worked, and calls the practitioner's standpoint a live problem.

Thesis VI said the method centralises reading. The granularity triad's third operation is the removal of interpretive layers.

What remains, then? The web's own judgement is the best guide, since it was made with the whole series in view. It finds three load-bearing tensions in the position, which it can live with, and three cracks. The cracks are the claim to start before the theory, when any map is made with prior commitments; the claim that there is no expert in the room, when someone designed the room; and an equivocal use of the word science. It finds the process schools' central objection only half answered: that to tune the density of interaction, or to design the conditions of collective seeing, is to take up a standpoint over the whole that the theory says nobody has.

The disputation's remaining questions are the same ones in practical form. Gell-Mann's dictum says the system should model itself. A system models itself when the model is made by its members and comes back to them. Where the prompt and the signifiers are written by others and the patterns go to a few, the system has been modelled with its own material by someone else. The dictum, taken at its word, is a statement of the return path.

The web's own ground

The legitimate and the shadow. The web uses Stacey's early distinction. The legitimate system is the formal organisation: its hierarchy, rules and approved account of itself. The shadow system is the informal web of talk and relationship in which, on the evidence it cites, most adaptation and creativity actually occur. Self-organising groups form there. A leader can encourage them and cannot design them. The web adds a warning from the group analyst Farhad Dalal. The shadow is no innocent place. It reproduces the wider patterns of inclusion and exclusion, and the legitimate system can disguise itself as the shadow, as when control is exercised through "values" and peer accountability.

This bears on instruments in an obvious way. A narrative instrument commissioned by the legitimate system is an attempt to hear the shadow. Whether the shadow speaks to it, and in what voice, is the question of the hidden transcript.

The living present. From Mead by way of Stacey: the present is the only time in which anything happens. The past is here as the effects of earlier encounters on how people now relate. The future is being made now, in the anticipations that people act on. The web treats this as a claim about what is real, and draws the practical consequence that change cannot be scheduled for later. It can only be done in the next exchange.

Formative and transformative. A formative account fixes the end state in advance and designs a route. A transformative account holds that the end is made in the going. The web's charge against the literature of agility is that it keeps a formative account in the vocabulary of complexity, and it adds Mowles's observation that a specified desired state usually serves whoever holds authority.

The pillars. From Elias the web takes figuration and power as a ratio. From Mead, the conversation of gestures and the living present. From Dewey, inquiry in which means and ends shape each other. From Foulkes, the group as a matrix in which the conductor is a node. It adds Lewin, whose force field it reads as an early account of mind extended into its surroundings, and from whom it takes a warning it calls the weight of complexity: managers who meet resistance by adding driving force raise turbulence and leave the restraint in place.

The methods the web admits

The web is not against method. It reads several familiar large-group methods through its own ontology, and its comments show what a Dweller in the Conversation thinks an acceptable instrument is.

Its governing principle is that complex facilitation is a commitment about what is real and is no set of techniques. A facilitator who steers a group toward an outcome fixed beforehand produces an imitation of an emergent process. Five principles follow: attend to interactions, accept emergence, let groups interpret their own material, use constraints with judgement, and seek diversity while attending, with Dalal, to who is structurally silenced.

Its comments on particular methods are pointed.

  • Of Open Space Technology it approves the freedom to move to where one is useful, and it criticises the founder's claim that the method always works, as itself a way of managing toward a result.
  • Of participatory narrative inquiry, in which groups find patterns in their own stories, it notes that the choice of categories is already an intervention.
  • Of methods that read patterns in human systems, it warns that pattern logic can become the gaze of an expert outside.
  • Of the World Café, that a table host must pass on what was said and must not curate it.

It also adopts an instrument of its own, a set of dimensions of sensemaking developed by Karina Solsø, which it describes as a quality of attention and no map. Among the dimensions it names are designing arenas for conversation in place of managing their content, attending to what produced a pattern and what the pattern will produce, real not-knowing, noticing what is unsaid, cohesion through ethical responsibility in place of agreement, and a closing question: what would responsible engagement be?

Two things should be said. The web applies to these methods the same criticism this book applies to Snowden's, which is to its credit. And its own instrument is an instrument. It was designed by someone, it has dimensions, and those who use it are formed to see in its terms. The difference between it and a signifier set is one of scale and of who holds it. That is a real difference, and it is smaller than the rhetoric of either school suggests.

THE ONLY VALID MODEL OF A HUMAN SYSTEM IS THE SYSTEM ITSELF —

THEN LET IT MAKE THE MODEL, AND SEE IT —

THE WOOD KNOWS IT IS BEING WORKED

Chapter 52

Irritation Is Not an Alibi: The Luhmann Post Examined

3,143 words, about 14 minutes of reading

The Post

"Irritation is not an alibi" appeared on 14 November 2026, the day after the post that occasioned this disputation. It belongs to the same series, in which Snowden defines anthro-complexity against one neighbouring body of thought at a time. Its subject is the sociology of Niklas Luhmann, and more exactly the use consultants make of it. Snowden is clear that his concern is with the use and not the man.

The post matters to this disputation for a plain reason. It contains the strongest statements yet of three positions that the eight theses said were missing: that the practitioner is inside, that power filters what is heard, and that the people whose experience is described have the first right to say what it means. A critique written against the earlier post must now meet these. This chapter states the post fairly, credits what it grants, and then examines it with its own tests.

What the Post Says

Common ground

Snowden finds four agreements with Luhmann.

  1. No instructive intervention. For Luhmann nothing in a system's environment can determine what it does. The environment can only irritate the system, which then responds by its own structures. Snowden's version speaks of modulating actants and interactions, and the practical conclusion is the same. Nobody can transfer a ready-made change into an organisation.
  2. Scepticism about changing mindsets. Luhmann holds that changing individual consciousness does not change the organisation. Snowden agrees for his own reason: behaviour is the shadow of conditions.
  3. Paradox is worked with, not resolved. Luhmann's treatment of paradox resembles the place of aporia in Cynefin. The post states the principle on which this disputation's second thesis rests: a system cannot observe the distinction it uses to observe.
  4. Premises that shape decisions. Luhmann's decision premises are the often informal assumptions that settle which decisions are even possible. Snowden treats the same things as actants, and asks of each how much energy and time it would take to change.

Different geology

Four differences follow, and Snowden calls them geological: they lie beneath the surface agreements.

  1. Where the human sits. Luhmann puts people in the environment of social systems. The system consists of communications. Minds and bodies are coupled to it and are not part of it. Snowden holds that this is a testable claim about what social life is made of, and that cognitive science tells against it. Anthro-complexity puts people, with their bodies, artefacts, stories and relationships, in the assemblage. Luhmann's structural coupling links systems that are already separate. Snowden's entanglement denies that they were separate to begin with.
  2. The basic unit. For Luhmann it is the communication. For Snowden it is nearer to the story. Humans are story-telling beings, and narrative fragments stay tied to the person who told them. That is why the teller's own signification matters. He adds an image: the fragments form collages at several scales, and a collage is not a mosaic.
  3. Complexity. For Luhmann a system exists by reducing the complexity of its environment. Snowden sides with Gregory Bateson against that emphasis. Much of his method exists to resist premature reduction, and to distribute sense-making so that weak signals are not removed by, in his words, the people with the power to filter.
  4. Natural science. Luhmann built a grand theory that can describe anything put into it. Snowden holds that natural science serves anthro-complexity as an enabling constraint, because it makes the theory answer to something outside itself.

The alibi and the licence

The centre of the post is two criticisms of how Luhmann is used.

Irritation as alibi. If an intervention can only irritate, and the system processes the irritation in its own way, then the consultant is covered whatever happens. Success shows the irritation was apt. Failure shows the system did what systems do. Snowden's formulation is exact. The trouble is not that the theory fails to predict. It is that the theory removes the vocabulary in which an intervention could be held to account.

Attribution as licence. Luhmann describes how organisations constitute actions by attributing them to actors, in ways that need not match the actors' own understanding. Snowden accepts that as description. He objects when it becomes a permission for the consultant's attributions to displace those of the people concerned. He calls that the Augustinian error in sociological dress.

The commitments

Against both, the post states what anthro-complexity commits to.

  • First attribution. In SenseMaker the originator of a story says what it means first, and that signification is kept beside whatever others later make of it. Later re-attribution is not prevented. It has to answer to a persistent record.
  • Embeddedness. The sentence deserves near-quotation: nobody stands outside, the practitioner least of all. Snowden says this is the starting point of the method, which is why the mapping of constraints is distributed across many people and why originators interpret their own experience.
  • Epistemic justice. He cites Miranda Fricker. To discount an originator's account, or to impose an interpreter's categories on experience they would frame differently, is an injustice whatever theory is used to justify it.
  • Ethics in the method. The closing contrast is that Luhmann's theory has no normative standpoint, which is part of its appeal to consultants. Anthro-complexity answers the ethical question in the design of its methods, and does not leave it to the character of the practitioner.

A postscript lists the post's own limits. Snowden is a methodologist and no Luhmann scholar. The German systemic tradition has done serious work. The similarities may be closer than he allowed. And the tone may owe something to a frustrating exchange.

What Must Be Credited

The record corrected. Four of the disputation's charges need amending in the light of this post.

Thesis III said the practitioner is pictured outside. Snowden now says in terms that nobody stands outside, the practitioner least of all. The thesis survives only as a claim about images and instruments.

Thesis V said power is the missing term. The post speaks of the people with the power to filter, of the abuse of power and agency, and of epistemic injustice. Power is named.

Thesis VI said tellers are treated as sources. The post grounds self-signification in a right: people should give meaning to their own experience before anyone else does.

Thesis II said the theory exempts itself from second-order observation. The post states that a system cannot observe the distinction by which it observes.

The postscript is also to be credited. It is a small declination table: what the author does not know, what he may have got wrong, and what coloured his tone.

So the gap between the disputation and anthro-complexity has narrowed again. What remains is, once more, a question for immanent criticism. The post states norms. It also supplies tests. The fair procedure is to apply the post's tests to anthro-complexity itself.

The Alibi Test Applied at Home

Snowden's test is this. A theory is an alibi when success and failure are both redescribed in its terms, so that no outcome counts against the intervention.

Consider how outcomes are described in anthro-complexity's own vocabulary.

  • A probe succeeds: a beneficial pattern emerged, and is amplified.
  • A probe fails: it was safe to fail, the failure was the purchase of information, and it is dampened.
  • A large intervention fails: the situation was complex, coherence is retrospective, and nobody could have known.
  • Stories shift in the wanted direction: the dispositional state has changed.
  • Stories do not shift: the system's dispositions are being read accurately.

Each sentence is defensible. Together they leave it unclear what result would show that a given deployment was a mistake. That is the structure Snowden objects to in the use of Luhmann.

His reply is ready, and it is a good one. A Web of Relations records it: emergence can be used as an alibi or as a discipline, and as a discipline it concentrates accountability on the design of conditions. He adds here that natural science constrains the theory from outside.

Two questions follow. First, natural science constrains the theory. It rules out, for instance, the picture of the brain as a programme. It does not obviously constrain a deployment. What observation would tell a sponsor that this signifier set was badly designed, or that this estuarine map misled? Second, if accountability concentrates at design, then the design must be open to inspection by those affected, and there must be stated conditions under which it would be judged to have failed. Snowden's rule for probes already requires this: signs of failure are written down in advance. The disputation's request is that the same rule be applied one level up, to the instrument. Thesis VIII offered four predictions for that purpose. They are a vocabulary of accountability of exactly the kind the post says must not be removed.

The alibi test is sound, and it applies to its author's practice until failure conditions for instruments are stated

Is the Originator First?

The post's central commitment is that the originator makes the first attribution. Look at the order of events in a deployment.

  1. A sponsor and practitioners choose an area of inquiry. The trialectic allows that this may be Augustinian.
  2. They write a prompting question, which says what kind of experience is wanted.
  3. They design a signifier set, which says along what dimensions an experience can mean something.
  4. A person tells a story in answer to the prompt.
  5. The person signifies it within the set.

The originator's signification is the fifth act of attribution. Three precede it, all by others, and they fix the space within which the fifth is made. It is the first attribution of that story. It is not the first attribution in the process. What the method secures is real: the teller's reading comes before any reader's. What it does not secure is that the teller's reading comes before the designer's.

This bears on the post's own citation of Fricker. Snowden describes one form of epistemic injustice as imposing an interpreter's categories on experience that people would frame differently. A signifier set is a set of categories, composed by interpreters, within which experience must be framed. Whether tellers would have framed it differently is precisely what a fixed set cannot reveal. By the post's stated criterion, an unexamined signifier set is a candidate for the wrong it names. The remedy is at hand and cheap. Let tellers share in composing the set, and ask each of them what it leaves out. Those are the co-design and fourth-vertex steps of Chapter 45.

A note on Fricker is in order, since she is being used on both sides. In Epistemic Injustice (2007) testimonial injustice is the giving of less credibility to a speaker because of prejudice about who they are. Hermeneutical injustice is a gap in the shared resources of interpretation that leaves someone unable to make sense of their own experience, her example being the time before "sexual harassment" had a name. The imposition of an interpreter's categories is a related wrong that later writers have described under other names. The distinction matters here. The hermeneutical case is the one where people lack the terms. A signifier set supplies terms. If the terms do not fit, the instrument records an experience as something it was not, and nothing in the data shows the misfit.

First attribution of the story: holds. First attribution in the process: does not

A Record That Answers to Whom?

The post says that self-signification creates a persistent record of the originator's meaning, to which any later re-attribution must answer, and that this makes re-attribution visible and open to challenge.

Visible to whom, and open to challenge by whom? The record is held in the instrument. Those with access can compare a later interpretation with the original signification. In the usual arrangement the tellers are not among them. They do not see the patterns, the workshop's reading, or the decisions taken. A record that a re-attribution must answer to, kept where the originator cannot consult it, protects the originator only if someone at the centre chooses to act for her. That is the benevolent intermediary again.

The commitment is right, and it entails more than the present arrangement delivers. If re-attribution is to be open to challenge, the originators must be able to see it. That is return before action and the mutual dashboard. The post's own example points the same way. To show what working on conditions means, Snowden lists a meeting rhythm, a reporting line, "or who sees which information". Changing who sees which information is the reconstruction's main proposal.

The principle holds It requires the return path to be real

Tied to the Teller

Snowden makes the difference from Luhmann anthropological. People are story-telling beings, and a narrative fragment remains tied to the person who told it. That tie is the ground for asking the teller what it meant.

This is the grammar of Thesis I in its most attractive form. The story is the teller's. The disputation agrees that she has a right over it. It adds what Mead and Bakhtin would add. A story is told to someone. It is shaped by the listener the teller has in mind and completed by a response. The same event is told one way to a workmate, another way to a manager and a third way to a screen. A story is tied to its teller and equally to its address. An instrument that preserves the first tie and removes the second has kept the possessive half of narrative and dropped the relational half. Luhmann, for his part, erred in the other direction, by giving the communication to the system and leaving the persons outside. The position between the two is the one the disputation has called emergent personhood: persons are real and their stories are theirs, and both are made in address and response.

There is a practical consequence. Step 6 of the reflexive SenseMaker protocol asks tellers to whom they would usually tell this story and who else is in it. That question restores the address.

Method or Character?

The post ends by saying that anthro-complexity answers the ethical question in the design of its methods and does not leave it to the character of the practitioner. The trialectic, five months earlier, said that the practitioner's own formation is not incidental, and that the willingness to be changed is the condition of anything useful happening.

The two statements pull apart. If ethics is secured by method, the practitioner's formation is a private matter. If nothing useful happens without the practitioner's formation, method does not suffice. A Web of Relations takes the second view and makes formation the bridge. Fricker, whom the post cites, also takes it in her way. Her remedy for testimonial injustice is a virtue of the hearer, a trained sensitivity that corrects for prejudice. Later writers, Elizabeth Anderson among them, replied that virtues of individuals are too weak and that institutions must be designed for justice. The sensible conclusion is that both are needed, which is the conclusion Chapter 51 reached.

The disputation reads the two posts as two halves of one position that has not yet been joined. In the Luhmann post Snowden is the Designer of Conditions, and he distrusts character because character cannot be inspected. In the trialectic he is close to the Dweller, and he distrusts method without formation because it turns the complex into the complicated. Each is right against the other's excess.

Borrowed Authority

The post begins with a complaint that consultants reach for a major thinker to lend authority to a practice the thinker would not support. It ends by calling one such pairing, of Luhmann with Juarrero, misappropriation and not synthesis. It gives a rule that is worth keeping: a concept moved from one domain to another needs its own justification in the new one. The example is autopoiesis, which Luhmann took from biology and which Maturana declined to see extended.

The rule applies widely. Anthro-complexity moves the constructor from physics, the actant from a sociology whose symmetry it rejects, 勢 from Chinese strategy in one of its three senses, the triad's thirdness from Peirce, and now three theologians, with the candid note that the post is no contribution to patristics. Some of these transplants are well justified in their new place and some are less so. Chapters 22, 25, 32 and 35 examined four of them. The same rule binds this disputation, which has moved quantum measurement, a Kipling story and a great many thinkers into a field where they did not grow. The adversary's Article 10 said so. The honest position for both is the one the post's postscript models: say what was moved, from where, and what would show that the move was a mistake.

What the Luhmann Post Changes

The post is, with the trialectic, evidence that the disputation was written against a moving target, and that the target was moving in the disputation's direction before the disputation began. That should be said without grudging.

It also sharpens what remains. The remaining differences are no longer about whether the practitioner is inside, whether power matters or whether tellers have rights. Snowden has affirmed all three. They are these.

  1. Whether the designer's prior attributions, in prompt and signifier, are themselves subject to the originators' challenge.
  2. Whether the persistent record is one the originators can consult.
  3. Whether instruments have stated conditions of failure.
  4. Whether ethics can rest on method alone.

Each is a question the post's own principles raise. That is what immanent criticism predicts. The norms are in the practice, and they ask more of it than it yet gives.

Following a granted claim.

Granted: people have the right to give meaning to their own experience before anyone else attributes meaning to it.

Then it must be granted that the dimensions in which they give it are part of the meaning given.

Then it must be granted that dimensions fixed beforehand by others are a prior attribution.

Then it must be granted that the right is fully honoured only if the originators can shape or contest those dimensions.

And granted: later re-attribution must answer to the originator's record and be open to challenge.

Then it must be granted that the originator is among those who may challenge, and so must be able to see what was made of her account.

Therefore the two commitments of the Luhmann post, taken at their word, require tellers to share in the design of signifiers and to receive the patterns and readings drawn from their stories. Anthro-complexity's ethics, as its author states them, already contain the reconstruction.

NOBODY STANDS OUTSIDE, THE PRACTITIONER LEAST OF ALL —

THEN THE DESIGNER IS INSIDE TOO —

AND THE FIRST ATTRIBUTION IS THE ONE THAT WROTE THE QUESTION

Chapter 53

The Zen Garden: The Politics of Perspective Applied Throughout

2,804 words, about 12 minutes of reading

A Third Witness

Mark Downham's study The Stacey Matrix and Walking in the Zen Garden at the Ryōan-ji Temple, Kyoto: On the Politics of Perspective concerns another framework than Cynefin, and it does not mention Snowden. It is brought in here because it is the most developed treatment in this body of work of a question the disputation keeps reaching: what is done when a situation is placed on a diagram, who does it, and who pays. Its findings transfer. This chapter states them, applies them to anthro-complexity, and then walks them through every book of the disputation in turn.

The Garden and the Joiner

The study begins with two images.

The garden. The dry garden at Ryōan-ji in Kyoto has fifteen stones set in raked gravel. It is viewed from the veranda of the abbot's hall. The stones are so placed that from any seat on the veranda at least one is hidden behind another. Nobody seated sees all fifteen. The study takes this as a figure for every perspective: a place to look from gives a view and withholds part of the whole. It adds a sixteenth element. The garden's makers were kawaramono, riverbank labourers of low status, and names thought to be theirs are cut into the back of one stone, half buried. The people who made what is looked at are present in it and are not seen.

The joiner. In 1983 David Hockney photographed the same garden for a work he called Walking in the Zen Garden. He took many exposures while moving and assembled them into one image. The result has many vanishing points. It was a deliberate refusal of one-point perspective, which fixes a single eye in a single place. A joiner shows that it was made by someone walking. The study notes that a second joiner of the same garden, made in 2026, is another assembly and is also partial. No joiner is the whole garden either.

From these it draws its governing claims. There is no view from nowhere. One cannot draw a distinction and stand outside it. And the remedy for a limited seat is the walk. A better seat does not exist.

The Matrix and Its Lost Axis

Ralph Stacey's diagram of the early 1990s had two axes. One ran from near to far from certainty: can cause and effect be known? The other ran from near to far from agreement: do the people concerned want the same thing? The study's central observation is that these are different kinds of axis. Certainty is epistemic. Agreement is political. An organisation, on the evidence of a long literature on coalitions that the study reviews, is a set of parties with different ends, and agreement among them is a continuing achievement that someone works to produce.

The diagram had regions. Where both certainty and agreement are high, technical rationality serves. Where certainty is high and agreement low, the mode is political: negotiation and coalition. Where agreement is high and certainty low, people act on shared vision or judgement. In the broad middle lies the zone of complexity. In the far corner, low on both, the diagram wrote anarchy.

Two things then happened, which the study traces year by year. The diagram was simplified by others into a few boxes, and in 2004 its political axis was relabelled in a way that removed the politics, turning a question about whether people agree into a question about requirements. Stacey himself came to disown the diagram. He said that it collapses a paradox it was meant to hold open, and that it sustains the dominant way of talking about management, in which a manager stands over a situation, classifies it and selects a tool.

The study's verdict on the surviving four-box version is that it deleted the axis on which power was recorded.

Five Findings

1. Sorting is an exercise of power. To place a problem in a zone is to route it: to experts, to a negotiating table, to an experimenting team. The study draws on Bourdieu's work on classification struggles. People contest how the world is to be sorted, and the sorting that prevails is the one backed by standing. It asks who classifies the classifiers, and answers that a reflexive practice puts the sorter on the diagram.

2. Agreement costs work, and someone pays. This is the study's most original argument. It reads the agreement axis thermodynamically. Disagreement is treated as social entropy, the dispersal of ends. Consensus is a low-entropy state. Like any such state it is improbable and is maintained only by continuous work, with the disorder exported somewhere. The study asks where. Its answer has two parts. Outwardly, the cost falls on those outside the agreement, whose dissent moves into the shadow organisation and the hidden transcript. Inwardly, and here the study turns to group analysis and to Byung-Chul Han, the cost is borne in individual bodies as anxiety and self-reproach, when a person experiences political disagreement as their own failure to be aligned.

3. The far corner is equilibrium. The diagram labels the corner of least certainty and least agreement as anarchy or chaos, a place of danger. On the thermodynamic reading it is the opposite of turbulence. It is the state in which no work can be done because nothing holds any gradient: a social heat-death. The generative region is the middle, held far from equilibrium. The study observes that the popular notion of an "edge of chaos" rests on a result in computer science that was later shown not to hold, and that the diagrams put the edge in the wrong place, as a boundary to stay back from. The thermodynamics of open systems stands without it.

4. Two matrices. S. H. Foulkes, the founder of group analysis, also used the word matrix. His is a web: the network of communication in a group, in which every member, the analyst included, is a node. Stacey's diagram is a grid read from outside. Both men drew on Norbert Elias. The study's irony is that the grid Stacey renounced carries his name, and the web he came to hold carries another's.

5. A threshold object. The study does not conclude that the diagram should be thrown away. It calls it a threshold object. It is valid where it points beyond itself. Its best use is to bring people to the limit of the technical toolkit and then to be put down. Used past that point as a device for sorting, it becomes what the study calls the map that ate the territory. A framework held as an offer in a conversation stays alive. A framework held as an answer is dead.

One further image should be kept. The study speaks of lamplight. Clear categories light a patch of ground brightly, and attention goes where the light is. The difficult matters are in the dark beside it.

Applied to Cynefin

Where is the agreement axis?

Cynefin's domains are defined by the relation of cause and effect and by the kind of constraint that holds: whether the relation is evident, discoverable, visible only in retrospect, or absent. That is a refined form of Stacey's certainty axis. Cynefin has no second axis. Nothing in the figure records whether the parties to a situation want the same thing.

Snowden has a reason. A Web of Relations reports that he places the Stacey matrix near the systems-thinking tradition because it defines complexity by what observers agree and are certain about. On his view that makes complexity relative to beholders, where anthro-complexity holds it to be a property of the world. The point is fair against any reading of the agreement axis as a measure of how complicated things seem.

The Garden study gives the reply. The agreement axis was never a measure of how things seem. It records a fact about the situation as real as any causal relation: that the people in it have different ends and unequal means of pursuing them. To remove it for the sake of ontology is to remove a part of the ontology. The four-box matrix lost its political axis through simplification by others. Cynefin, built on better science, never had one. The outcome is the same. Thesis V said that the missing term is power. The Garden study lets that be said in the geometry of the frameworks.

The residue at the centre

There is a qualification, and it favours Snowden. Cynefin does contain a place where disagreement appears. The central domain, confused or aporetic, is the state of not knowing which domain applies, and Snowden's account of it notes that people read a situation by their own habits and so place it differently. Contextualisation exercises make much of items that participants put in different places. This is a residue of the agreement axis, kept at the centre of the figure.

But it is disagreement about classification. It is treated as something to work through on the way to an appropriate response. Disagreement about ends, between parties who understand the situation perfectly well and want different things from it, has no domain. In Stacey's original, that was the political zone, and its prescribed mode was negotiation. In Cynefin such a situation is likely to be placed in the complex domain and probed. A probe is a poor instrument for a conflict of interest.

The veranda and the map

An estuarine map is drawn from a seat. Chapter 21 argued that energy cost and time to change are relations between a constraint and a party. The garden states it without argument. From this seat, that stone cannot be seen. What a group of directors places beyond the counter-factual line is a stone hidden from their veranda. The reconstruction's mapping by position is the walk, and a set of maps laid side by side is a joiner. The study's caution applies to the reconstruction too. A joiner is still an assembly made by someone, and it is not the garden.

The hidden makers

The half-buried names have two applications. The designers of an instrument are makers whose hand is in the thing and is not seen. The reconstruction's rule that they be named, with their relation to the commissioner, is an unearthing of the stone. The second application runs the other way. The tellers make the pattern that the sponsors contemplate from the veranda. Their names are, properly, protected. Their part in making what is looked at is forgotten at the moment of looking.

A shift in a narrative landscape toward the stories a sponsor prefers is, in the study's terms, a reduction of social entropy. It is bought with work. The question the thermodynamic reading forces is who did the work and where the disorder went. It may be that conditions changed and people's experience improved. It may be that people learned what to say, and that the dissent moved to where the instrument does not reach. From the pattern alone the two cannot be told apart. The study's inward account adds a cost that Thesis V did not name. Where disagreement has no legitimate coordinate, people carry it privately. An instrument that asks individuals, one at a time, to signify their experience can privatise in just this way what was a shared political condition.

Probing and the far corner

Cynefin does not make the diagram's mistake about the far corner. It values a bounded entry into chaos as a source of novelty, and it places its cliff between complacent order and chaos. On this point anthro-complexity and the Garden study agree against the popular matrix.

Cynefin as a threshold object

Snowden insists that Cynefin is for sense-making and should emerge from the data. That is the study's living map. He has also resisted the packaging of his methods, which Chapter 5 recorded as Element 22. The Garden study's genealogy is a warning about what happens anyway. A diagram that is easy to draw is copied, simplified and attached to purposes its author rejects. Stacey could not prevent it. The study adds from Judith Williamson and Jacques Lacan an account of why the simplified version spreads. A sorting diagram returns to the manager a flattering image: the one who stands above, sees the kinds of problem there are, and chooses. Cynefin's best-known publication is titled a leader's framework for decision making. Whatever its author intends, the figure can be read from the veranda, and often is.

The practical conclusion is the study's own. Use the figure to bring a group to the edge of what analysis can do, and then put it down and talk. Date it, as Chapter 45 proposes for every instrument.

The Garden Walked Through the Whole Disputation

The request that governs this chapter is that the study be applied to every part of the book. The table gives the application book by book. Additions placed at the end of the relevant chapters carry each point further.

Book What it contains What the Garden asks of it
Before the Disputation The argument in brief and the method From which seat is this disputation written, and which stone can it not see? The author's position is given in Article 10 and should stand at the front.
Book One: The Claim Laid Open The twenty-two elements Each element is a view from the practitioner's veranda. The slow readings add the walk: who classifies, whose disposition, who sees.
Book Two: The Neighbours CAS, systems thinking, the process schools A map of neighbours is a classification struggle. Snowden objects to being placed on another's tree. The same care is owed to those this book places.
Book Three: Why Human Complexity Is Treated Differently The five reasons and the three jobs The positional reason is the politics of perspective applied to a school. A distinction is drawn from a seat.
Book Four: The Eight Theses The criticisms Thesis V gains a geometry: the missing axis. Thesis III gains an image: the veranda. Thesis VIII gains a mechanism: the cost of produced agreement.
Book Five: The Corpus Brought to Bear Mead, Elias, Juarrero and others Foulkes joins them. His matrix, in which the conductor is a node, is the group-analytic form of the observer inside.
Book Six: Through 道理勢 The Chinese terms The garden is itself a classical East Asian statement that the whole is not given to a seated viewer. It should be read with Cook Ding, who no longer sees the whole ox.
Book Seven: The Observer Inside Measurement, the mariner Looking is never free. The study's use of Maxwell's demon and Landauer's principle gives a second physical analogy beside the Zeno effect: sorting has a cost, and the sorter is inside the box.
Book Eight: Reconstruction Principles and protocols Mapping by position is the walk. The mutual dashboard is a joiner. Dated limits make every instrument a threshold object.
Book Nine: After the Coda Trialectic, Jaeggi, Luhmann The three architectures are three verandas. The triquetra's empty centre is the gravel: the place nobody may stand, around which the stones are seen.

A Dialectic and Its Limit

For the opposing side. Three replies.

First, agreement is a fact about the participants' states of mind, which change as the situation is engaged. To build it into the definition of the domains would make the framework a mirror of opinion. Cynefin is right to define domains by constraint, and power is already there, since every governing constraint is someone's power made durable.

Second, the thermodynamic reading is a metaphor. Social entropy is not measured in joules, and to speak of who pays for consensus is to dress a political claim in borrowed physics. The Luhmann post's rule applies: a concept moved between domains needs its own justification.

Third, the walk has a cost. A practitioner who must map from every seat and then show that the joiner is partial will never finish. Leaders need a figure they can hold.

The disputation's answer. The first reply is half right. Constraints do carry power, and that was the argument of the asymmetric constraint capacity card. But a constraint map that does not say whose power a constraint is has drawn the stone without the hidden one behind it. The second reply is accepted as to status. The study itself says its physics is a reading, and the claim that agreement takes work and that the cost is unevenly borne can be made, and is made here, on sociological grounds alone. The third is a real cost and is conceded. The reply is the study's: the figure is for the threshold. Nobody needs a complete walk. They need to know that they are seated.

FIFTEEN STONES, AND NO SEAT SHOWS THEM ALL —

THE AXIS OF AGREEMENT WAS THE AXIS OF POWER —

WALK, AND PUT THE MAP DOWN AT THE THRESHOLD

Chapter 54

Does Anthro-Complexity Block Its Own Learning?

1,621 words, about 7 minutes of reading

The Question

Jaeggi gives no picture of the good form of life. Her test is whether a practice can deal with the problems it sets itself and learn from doing so. A failing practice has what she calls a blockage to learning. This chapter applies the test to anthro-complexity, and then to the disputation. It also takes up the point of friction marked at the end of Chapter 49: whether anthro-complexity's refusal to say what a good organisation is can be kept.

The Evidence That It Learns

The case in favour is strong and should be given first.

It revises in public. The trialectic exists because readers objected to a binary, and its author changed the scheme. The Luhmann post ends with a list of its own possible errors. Cynefin has been redrawn several times over twenty-five years, with domains renamed and the central domain rethought.

It moves toward its critics. A Web of Relations records Snowden conceding that method is a site of power, distinguishing emergence as alibi from emergence as discipline, and describing the practitioner as entangled but not dissolved. Each of these is a position a critic might have pressed on him. He reached them himself.

Its method is built for learning. Parallel probes, expected failure, amplification and dampening, and the refusal of a fixed end state are a design for Dewey's experimental inquiry. Jaeggi's own metacriterion, growth without a fixed end, could be mistaken for the vector theory of change.

It is plural by design. Many coherent hypotheses are kept alive, and premature convergence is treated as the main risk. That is Jaeggi's experimental pluralism in miniature.

By the main test, then, anthro-complexity is not a blocked practice. Anyone who has followed the argument of this book should say so plainly.

Four Places Where Learning May Be Blocked

Jaeggi distinguishes simple blockages, where old interpretations have not caught up, from systematic ones, where the categories of the practice prevent an experience from registering. Four candidates of the second kind can be identified. They are offered as hypotheses.

1. The substrate argument

Snowden holds that rival traditions fail because of the ground they stand on. The argument has force. A Web of Relations notes its side effect. If every objection from a systems thinker shows the systems substrate, every objection from a process theorist the therapeutic one, and every objection from a modeller the computational one, then no objection from outside can register as an experience of being wrong. It registers as further evidence about the objector. That is the form of an immunising strategy, whether or not it is so used.

The Luhmann post supplies the test. A grand theory, it says, can redescribe the outcome of any intervention in its own terms and so answers to nothing outside. The same may be asked of a theory that can redescribe any criticism in terms of the critic's substrate.

2. The border

Book Three argued that the human distinction does institutional work. It marks anthro-complexity off from its neighbours. Jaeggi notes that whether a form of life is pressed to change depends largely on contact with rival forms that challenge it. A firmly drawn border reduces contact. The neighbours who have worked longest on the observer's position are the critical systems thinkers and the process schools, and they are the ones kept at a distance. A Web of Relations finds the same on the other side and charges both with declining the meeting that would test their difference.

3. The commercial form

The trialectic reports that the market rewards the deficit mode and that the encounter mode is often declined as professionally risky. That is a learning blockage described by the practice's own author. The practice knows what its constitutive norm requires and is prevented, by the conditions under which it is paid, from regularly doing it. Jaeggi's fourth feature applies. A difficulty of this kind is structural and is not removed by goodwill.

4. What the instrument cannot register

This is the most exactly systematic of the four. An instrument registers experience in its own categories. If the categories do not fit, the misfit leaves no trace in the data. If tellers have learned the instrument, their learning appears as a clear pattern. The instrument's own failure modes are the experiences it is least able to have. Jaeggi describes blockages that concern the conditions under which experience is possible at all, and this is one. The remedies are devices that let the instrument register its own misfit: the question about what the signifiers leave out, the separate reading of stories about earlier rounds, and the estimate of how much has not been seen.

Against Methodological Abstinence

Anthro-complexity declines to state what a good organisation is. It rejects idealised future states. It asks for more stories like these and fewer like those, and leaves the choice of stories to the people concerned. In the Luhmann post Snowden describes the aim as letting people give meaning to their own experience without the practitioner imposing a reading.

This has the shape of the position Jaeggi calls ethical abstinence, and it has the same merit. It refuses to impose a conception of the good on people who have their own. Her objection also applies. The abstinence cannot be kept, because the decisions it declines to make are made anyway, elsewhere and out of sight.

In a deployment, the decisions that bear on what is good are made at four points. The area of inquiry is chosen. The prompt and signifiers are written. Certain stories are selected as the ones to have more of. A response is designed. The practitioner may abstain at each point. Someone does not, and it is usually the sponsor. An abstaining practitioner then lends the neutrality of a method to a judgement of value made by the party with most power. Jaeggi's sentence about the liberal state fits: where public decisions are in fact taken about how people can live, there ought to be public discussion of them.

It should be said at once that anthro-complexity is not consistently abstinent, and that its departures are to its credit. The Luhmann post is openly normative. It asserts a right, names an injustice and judges a practice wrong. The trialectic condemns the deficit mode. The distinction between alibi and discipline is an ethical one. So the practice has an ethics. What it lacks is a place in its method where the ethics is argued about by those affected.

The reconstruction's answer is neither a doctrine of the good organisation nor abstinence. It is to make the four decisions visible and contestable: who chose the question, what the signifiers were built to make sayable, who picked the stories, and by what rule the response was decided. That is Jaeggi's proposal carried into a method. The practice does not say how people should live. It refuses to let that be settled where nobody can see.

The Test Turned on the Disputation

Immanent criticism binds the critic. Three questions must be put to this book.

Has it learned? Its first thesis was amended in the addition to Chapter 2. Its opening charges were amended in Chapters 48 and 52 when two posts of Snowden's showed them overstated. Chapter 51 accepted from A Web of Relations that its own proposals carry an unpaid debt on scale. Chapter 55 revises the verdicts. So far, the book has changed when shown to be wrong.

Where might it be blocked? In the same four places. It has a substrate, the relational sociology of Mead and Elias, and could treat every defence of instruments as a symptom of the possessive self. It draws a border, since a critique needs its object to stay still. It has a form to maintain: the disputation's apparatus of theses and verdicts rewards finding against the other side. And its categories cannot register one thing in particular, which is the experience of a practitioner inside a real deployment, of which the author has none. Article 10 of the adversary's case says so.

What would count against it? The Luhmann post's rule requires an answer. The disputation would be shown wrong on its central point if repeated deployments on one population, studied independently, showed no convergence of significations, no thinning of outliers and no rising share of stories about earlier rounds. It would be shown wrong on its remedy if return before action and co-designed signifiers produced no difference in what tellers say or in what sponsors do. Those are stated here so that the book does not ask of others what it does not offer.

Deeper: do not block the path of inquiry. The maxim Jaeggi takes from Putnam is Peirce's, from an unpublished lecture of 1898. He listed four ways in which philosophers block inquiry. One is absolute assertion. A second is to declare that something can never be known. A third is to hold that some element of science is basic and beyond explanation. The fourth is to hold that a law or truth has found its final and perfect formulation.

Peirce is anthro-complexity's chosen philosopher, and each of the four gives a question to both parties here. Is any claim asserted beyond challenge? Is anything declared unknowable that could be looked into, for instance what tellers make of the instrument? Is any element treated as basic, such as the triad? And is any formulation treated as final? Snowden's record on the fourth is good, as the repeated redrawing of Cynefin shows. The disputation's contribution, if it has one, concerns the second and third.

THE PRACTICE LEARNS, AND THAT IS ITS BEST DEFENCE —

IT IS BLOCKED WHERE ITS OWN INSTRUMENT CANNOT SEE —

AND ABSTINENCE HANDS THE DECISION TO WHOEVER PAYS

Chapter 55

The Thirteen Dialectics, the Verdicts Revised, and a Fourth Architecture

2,356 words, about 10 minutes of reading

Why Gather Them

A disputation proceeds by setting positions against each other. Over fifty-four chapters this book has set many against many, and a reader may fairly ask what the oppositions come to. This chapter gathers them. It names thirteen dialectics that run through the work, says for each what either side holds and what it pays, and states what has emerged. It then revises the verdicts on the theses, adds a thirteenth principle to the reconstruction, and closes.

One caution about the word. Snowden distrusts dialectic when it means two poles and a midpoint, and his triquetra test rules out the compromise. The sense used here is Jaeggi's, taken from Hegel: a position fails by its own standard, and what replaces it keeps the experience of the failure. Nothing below is a midpoint. In several cases nothing has replaced the opposition, and that is said.

Thirteen Dialectics

1. Possession and relation

One side. Humans hold identities, act with intention and create abstractions. The capacities belong to persons, and persons can be asked. The other. Identities are called out, intentions grow from shared intention, and abstractions exist by acceptance. There is only the process. What each holds. The first holds the reality and the rights of the person. The second holds the truth about how persons come to be. What each pays. The first collects from individuals what was made between them. The second risks losing the person, and with her the one who can refuse. What emerges. Emergent personhood: persons formed in relation and real as formed. Snowden's "entangled but not dissolved" and the trialectic's "held by neither" arrive at the same place. Largely resolved. (Chapters 2, 19, 27, 28, 50.)

2. Outside and inside

One side. A system can be read: its dispositions sensed, its constraints mapped. The other. The reader is a member and the reading is an event. What emerges. Agreement in principle. Snowden now says that nobody stands outside, the practitioner least of all. The open matter is whether the designer, and the images and instruments, have come inside with the practitioner. Resolved in theory, open in instruments. (Chapters 21, 39 to 41, 52.)

3. The Designer of Conditions and the Dweller in the Conversation

One side. Method makes a practitioner's contribution visible and accountable, and it works at scale. The other. To design conditions for others is the view from above, rebuilt in the apparatus. What each pays. The Designer pays in standpoint. The Dweller pays in reach and in accountability. What emerges. No synthesis. Two movements toward each other: a designer who says he must be formed, and dwellers who say their rooms must reach the corridor. The Tidal Loom is one proposal for joining them. Open, and the most important. (Chapters 11, 45, 51.)

4. Method and formation

One side. Ethics belongs in the design of methods. Character cannot be inspected. The other. Nothing useful happens unless the practitioner is willing to be changed. Formation is the bridge. What emerges. Both positions are Snowden's own, five months apart. They are complements. Method without formation yields unexamined frames. Formation without method yields wisdom that cannot leave the room. Resolvable. The thirteenth principle below states how. (Chapters 48, 51, 52.)

5. First attribution and the relocated frame

One side. The teller says what the story means before anyone else does. The other. The prompt and the signifiers were written first. What emerges. The teller's attribution is first among readers and fifth in the process. The right Snowden asserts is real, and it extends to the dimensions. Resolvable by co-design and the question of what is left out. (Chapters 25, 45, 50, 52.)

6. Legibility and encounter

One side. A landscape made legible lets a practitioner see what is present and work with it. The other. What has been made legible to one party is no longer met by that party. What emerges. Legibility that is mutual. If the many can read the pattern, the centre and one another, legibility becomes a condition of encounter and stops being its end. Resolvable in design, unproven at scale. (Chapters 24, 31, 50.)

7. Scale and presence

One side. Ten thousand people cannot be met. Only an instrument reaches them. The other. What forms people happens in presence, in moments that cannot be scheduled. What emerges. Nothing settled. Nested circles, formed convenors and the travel of themes are partial answers. A Web of Relations lists this as still owed, and so does this book. Open. (Chapters 31, 43, 51.)

8. Abstinence and judgement

One side. The practitioner imposes no picture of the good. People choose their own stories. The other. Someone chooses, and an abstaining method lends its neutrality to that choice. What emerges. Neither a doctrine nor abstinence: the four deciding moments made visible and open to contest. The Luhmann post already departs from abstinence when it names an injustice. Resolvable. (Chapters 23, 49, 54.)

9. Alibi and discipline

One side. In a complex situation nobody could have known. The other. Because nobody could have known, the designer of conditions answers for the design. What emerges. Snowden's own distinction settles the principle. What remains is to state, for instruments as for probes, the signs by which failure would be recognised. Resolved in principle, owed in practice. (Chapters 2, 26, 52.)

10. Knowability and agreement

One side. Situations differ by whether cause and effect can be known. That is a fact about the world. The other. Situations differ also by whether the parties want the same thing. That is a fact about the world too, and it is where power is recorded. What emerges. Cynefin has the first axis in a refined form and a residue of the second at its centre. A political reading of every placement, by position, restores what the figure omits. Open. The frameworks have not met on this. (Chapters 23, 53.)

11. Structure and conversation

One side. Without structure, talk is shapeless and the loudest voice prevails. The other. Structure hides the power of whoever imposed it. What emerges. The observation of the earlier treatise holds: each takes its own form for freedom. Structure that is declared, dated and open to those inside it meets most of the second objection. Conversation that admits its own structure meets most of the first. Narrowed. (Chapters 11, 45, 51.)

12. Two terms and three

One side. Pairs trap thought. Three terms give a field. The other. A triangle whose shares sum to one is a trade-off. What matters is mediation. The count of terms is secondary. What emerges. Agreement that the third must be irreducible, which is Snowden's triquetra test. Disagreement about whether the triad as an instrument, and the formative encounter as a concept, pass it. Open on both. (Chapters 25, 32, 38, 50.)

13. Critic and practice

One side. This book: written from outside any deployment, by one author, in the form of findings against. The other. A practice of twenty-five years that revises itself and had moved toward most of these findings before the book was written. What emerges. The critic has been corrected three times by the practice's own texts. The practice is asked four questions that its own principles raise. Jaeggi's test for both is whether they go on learning. Open by its nature. (Chapters 43, 54.)

The Dialectics Displayed

Dialectic State What would settle it
1 Possession and relation Largely resolved An account in anthro-complexity of how persons are formed
2 Outside and inside Resolved in theory Instruments that place designer and sponsor in view
3 Designer and Dweller Open The two schools meeting, as each one's own discipline requires
4 Method and formation Resolvable Formation named as a requirement of practice
5 First attribution and frame Resolvable Tellers sharing in signifier design
6 Legibility and encounter Resolvable in design Mutual dashboards tried at scale
7 Scale and presence Open A demonstration
8 Abstinence and judgement Resolvable The four deciding moments declared
9 Alibi and discipline Owed in practice Failure conditions for instruments
10 Knowability and agreement Open A political reading of Cynefin placements
11 Structure and conversation Narrowed Declared, dated structure
12 Two terms and three Open Independent scales tested beside triads
13 Critic and practice Open by nature Independent study of repeated deployments

The Verdicts Revised

The Table of Verdicts in the Apparatus stands as written. Under the Additive Covenant nothing in it is removed. The revisions below are additions to it, made in the light of Book Nine.

Thesis Original verdict After the trialectic and the Luhmann post Revised verdict
I. The possessive fallacy Holds (amended) The formative encounter denies possession in practice. "Tied to the teller" keeps it in the unit of collection. Holds for wording and instrument, fails for the account of practice
II. The asymmetry applies to itself Holds (amended) Snowden grants that method is narrative power and that a system cannot observe its own distinction. Holds, and is partly conceded
III. Humans are not estuaries Holds "Nobody stands outside, the practitioner least of all." Holds for images and instruments, fails as a claim about his theory
IV. The under-socialised constraint Holds The practitioner's formation is part of the work. Decision premises are treated as actants. Holds for the typology, weakened for the constructor
V. The missing term is power Holds (amended) Power to filter, abuse of power and epistemic injustice are named. The agreement axis is still absent. Holds as "power is named and not mapped"
VI. Computational practice Holds Self-signification is grounded in a right. The return path is not required. Holds, with the return path still not required
VII. The triad is a simplex Partial Untouched by either post. The triquetra test strengthens the demand for irreducibility. Partial, unchanged
VIII. Instruments degrade Holds (amended) Untouched. Now also required by the alibi test as a statement of failure conditions. Holds as prediction, untested
IX. The inverted encounter New Stated in Chapter 50. Holds for the mapping stage, contested for intervention

Five of the eight original theses are weakened or partly conceded, and none is overturned. The direction of the change is the finding. The disagreement between the disputation and anthro-complexity has moved from theory to instrument.

A Thirteenth Principle

Chapter 44 stated twelve principles. Book Nine requires one more.

Principle 13. The practitioner is formed, in company, and returns to give an account. Method does not suffice, and neither does character. Those who design and convene inquiries undergo a continuing formation of perception and judgement among peers who can see what they cannot. They enter each engagement expecting to be changed, and say afterwards how they were. Their formation is open to those they work among, to the extent that the reading of the reader allows. Answers Thesis IX. Grounded in Chapters 48, 50 and 51. It joins Principle 7 (declare) to Principle 8 (return), and it takes from A Web of Relations the discipline of return and from the trialectic the willingness to be changed.

The principle has a safeguard, which comes from the objection in Chapter 51. Formation must not become a priesthood. So it is paired with the rule that instruments stay usable by anyone and that the formed are answerable to the unformed.

A Fourth Architecture

Snowden's banner shows three buildings: a basilica, a monastic enclosure and a domed church. Chapter 50 observed that each is a sacred building with those who officiate and those who attend, and that the third has a screen.

It would break the triquetra to propose a fourth position, and none is proposed. What may be offered is a fourth building, as a reminder of what the three leave out of view. The people who attend all three go home afterwards to a place with no altar and no officiant: a kitchen, a workshop or a street. Most of what forms them happens there, among themselves. It was happening before the practitioner arrived and goes on after she leaves. Jane Jacobs described it on the pavement, Hutchins on the bridge of a ship, and Elias in a suburb of Leicester. The Garden study's hidden makers belong there.

The formative encounter, in its corrected sense, is mostly that. The practitioner's part is to be one person in the kitchen for a while and to leave it no worse. If she brings an instrument, it should be one that people there can pick up, look into, and use to see one another. The test from the Laozi stands: afterwards, with everything in view, they say that they did it themselves.

Last Words

The book began with an image of a river, a city, the planners who argue about the river, and someone drawing the map. It has argued throughout that the map-maker is in the city. Its last addition is that Snowden says so too, and has said so with increasing force during the year in which these posts were written. What remains between his position and this one is practical: whether the people in the city can see the map, change its legend, and watch the map-maker at work.

That is a small difference to have written so much about. It is also the difference between an instrument that a population uses to read itself and one that is used to read a population.

THIRTEEN DIALECTICS, AND FEW OF THEM CLOSED —

THE QUARREL HAS MOVED FROM THE THEORY TO THE TOOL —

GIVE THE TELLERS THE MAP AND THE PEN

Addition: A Fourteenth Dialectic, and All Fourteen in Cards

Chapter 56 adds a fourteenth dialectic to the thirteen above: incompressibility and altered compressibility. One side holds that every model excludes and that the modeller answers for the exclusion. The other holds that in human systems every description acts. What emerges is an ethics of compression as action. Chapter 57 then puts the book's questions to the alternative practice it has pointed to throughout.

Chapter 56

Thesis X: Incompressibility and Altered Compressibility

3,141 words, about 14 minutes of reading

A Position the Book Has Relied On

Chapter 37 and several cards treat Paul Cilliers as the complexity theorist nearest to this disputation. His argument was used as follows. Complex systems are incompressible. Every model of one leaves something out. What is left out is chosen by the modeller. So framing is an ethical act.

Under the Additive Covenant those passages stand. This chapter challenges them. The challenge is the author's own, and it changes what the book may take from Cilliers. It is set out here in the usual form: the position, the objection, the case for the other side, and a ruling.

The Position Stated

A recent statement of the view, in the tradition of Cilliers and of Kurt Richardson, runs as follows. It is paraphrased closely, since it is a good statement.

  1. Every attempt to understand a complex system involves an act of exclusion.
  2. Complex systems are incompressible. Their behaviour arises from relations, histories and settings that no simpler model fully captures. So every model is incomplete.
  3. To make sense of complexity we must simplify: draw boundaries, select what matters, leave out what seems less relevant. Knowledge is possible only because of these limits.
  4. Because relations are non-linear, what is left out can matter enormously.
  5. The choices are never neutral.
  6. The observer is not separate from what is observed. What is seen depends on standpoint, concepts, tools, history and training.
  7. Models do not only describe. They take part in producing the realities people live in.

From this the statement derives an orientation for practice with five parts: philosophical inquiry into one's assumptions, boundary critique, critical pluralism that engages many practices without holding them all equally valid, creativity, and provisionality. It adds that framing has an ethical side, and concludes that complexity practice is the development of better judgement in worlds that always exceed our descriptions of them.

Much of this the disputation has itself said. Points 5, 6 and 7 are close to Theses II and III. The challenge concerns points 1 to 4 and the conclusion, and it begins from an inconsistency between them and point 6.

The Challenge in Six Moves

Thesis X. Incompressibility is a paradox of the external observer. It belongs to the modelling of complex adaptive systems, where a describer stands outside a fixed object and tries to make a smaller copy. In human systems there is no fixed object and no outside. There, compression is something done within the situation, and it changes how compressible the situation is. The right concept is altered compressibility. Cilliers, and those who repeat him, describe the limits of description and mistake them for the limits of knowing.

Move 1: incompressibility needs the outside it denies

The idea comes from algorithmic information theory. A string of symbols is incompressible if no programme shorter than the string can produce it. Three things are assumed. The string is given and stays the same while it is described. The describer is not part of the string. And success means reproduction: the short programme must regenerate the original.

Carry the idea to a complex system and the assumptions come with it. The system is there, the modeller is elsewhere, and the model is a would-be replica that falls short. Point 6 of the statement says that the observer is not separate from the situation. If that is so, the first two assumptions fail, and the concept in point 2 has lost its footing. The statement holds both. It denies the outside position and keeps a concept that only an outside position can define.

Move 2: two kinds of compression

The author's distinction is between two things that the word compression runs together.

Abstracted simplification is external abstraction. One stands back and makes a reduced description. The object is untouched and the description is smaller than it. Something is always lost, and Cilliers is right about this kind.

Altered compressibility is what happens when one engages with a situation. The engagement changes the situation, and among the things it changes is how much regularity the situation has.

Two sentences of Snowden's carry the argument. Both are from a short post of January 2024, "Knotty problems".

You can unravel a knot easily if you understand its structure.

All forms of diagnosis are in fact interventions, and all interventions should be designed as diagnostic instruments.

Take the knot first. A tangled cord, described from outside crossing by crossing, needs a long description. A person who understands how it was tied needs a short one: a few moves in the right order. That is already a difference in compressibility that depends on the knower. Then the person unties it. The cord is now a straight line, and almost nothing is needed to describe it. Nothing has been excluded. The object was changed by someone who understood it, and its complexity went with the change. Snowden notes the other course too: Alexander's sword. That also alters compressibility, by another means.

Take the second sentence. If every diagnosis is an intervention, then there is no description that leaves the described as it was. Each act of finding out alters what there is to find. So compressibility cannot be a fixed property that a model fails to reach. It is a variable, and practice moves it.

Human systems do this to themselves continually. A rule, a routine, a standard form, a shared category and a weekly meeting are compressions made from inside, and each makes conduct more regular than it was. Karl Weick called organising the reduction of equivocality. Luhmann made the reduction of complexity the defining act of a social system. Ian Hacking showed categories of people making people fit them. Anthro-complexity's own method is of this kind. Estuarine mapping, in Snowden's words in that same post, aims to change the energy gradients so that the system is more likely to shift in a wanted direction. That is deliberate work on how the system is disposed, which is to say on its regularities.

Move 3: matching in place of reducing

The statement says that knowledge is possible only because we simplify. That holds if knowing is modelling. There is another way of knowing, which the author calls matching complexity, by immersive participation.

W. Ross Ashby's law of requisite variety says that a regulator must have as much variety as what it regulates. One response is to cut the variety of the thing down to a model. The other is to bring enough variety to meet it. A skilled participant does the second. Cook Ding does not carry a reduced copy of the ox. He says that he no longer sees the ox whole. His knowledge is in the meeting of blade and joint. The navigation team on Hutchins's bridge has no model of the harbour in any head. A parent knows a child, and a caretaker an estate, in the same way.

Knowing of this kind does not exclude in Cilliers's sense, because it makes no copy from which things could be missing. It is limited in other ways. It is local, it cannot be handed over, and it sees from where it stands. Those are the limits of a participant. They are different from the incompleteness of a replica.

Move 4: there are no boundaries

The statement says we draw boundaries, and recommends boundary critique: asking what was included, what was left out and whose view is missing.

The author's position is stronger. There are no boundaries. An external observer has to draw one, because a model must stop somewhere. Within a situation one finds gradients, thresholds, degrees of involvement and lines that run on. Elias's figurations have no edge. Ingold's meshwork is made of lines that pass through. Juarrero speaks of constraints, which shape probabilities and enclose nothing. Snowden makes the same point against Luhmann: the boundaries that structural coupling presupposes are not there to begin with.

On this view boundary critique is the good conscience of the external observer. It asks the modeller to draw the line more carefully and to hear from those on the far side of it. That is worth doing for as long as one is modelling. It does not question the activity of standing off and bounding.

The disputation has used Werner Ulrich's boundary questions throughout, and it should say how this move affects that use. Ulrich's questions examine acts of bounding by planners and experts. They are consistent with the claim that nothing is bounded in itself. They remain the right tool wherever someone has in fact drawn a line: around who is surveyed, what a signifier set covers, who is in the room. What the present move adds is that the better course, where it is open, is not to draw the line.

Move 5: the limits of description

The statement ends with worlds that always exceed our descriptions of them. That is true. It is a truth about description.

Wittgenstein's Tractatus is the classical treatment. It sets a limit to what can be said and holds that there is also what shows itself and cannot be said. The author applies this as a distinction between knowledge, in the sense of what can be stated and modelled, and perception, in the sense of what is seen by one who is there. The reading is the author's own application of the book, and the Tractatus itself speaks of saying and showing. Its bearing here is clear enough. To demonstrate that every statement about a complex situation leaves something out is to mark the limit of saying. It does not follow that the situation cannot be known by acquaintance, acted in competently, or changed. Cilliers's argument concerns the first limit, and its conclusions are drawn as if it concerned all of them.

Move 6: the elephant

Critical pluralism, the third part of the recommended orientation, is the proposal to engage many models, each partial.

The old parable is the obvious image. Several blind men touch an elephant. One feels the trunk and says it is a snake. One feels a leg and says a tree. Each is partly right. The author calls the resulting condition a paradox of simple contradiction: a war of competing external observations, each producing a model that is an incomplete replica.

Three things about the parable are usually missed. It is told by someone who can see the elephant, and its lesson depends on that sighted narrator, whose position none of the men can occupy and which the story's teller quietly takes. All the men are outside the animal. They are in the same relation to it and differ only in where they stand. And nobody rides it, feeds it or walks with it. A mahout knows the elephant in a way that no number of reports from the blind men add up to.

A pluralism of models gathers more reports. It can be done critically and well. It remains a pluralism of outside views. The author's earlier treatise on critical systems thinking, The Connoisseur of Partial Truths, made this charge against the methodological pluralism of that school. It applies here.

The Case for the Other Side

For the opposing side. Cilliers and Richardson have strong replies, and the first four should be given in full.

First, the knot tells for them. To understand a knot's structure is to have a model of it, and a short one. So the knot was compressible. That is why it could be untied. A knot is a complicated object and an ordered one. Snowden's choice of the word knotty over wicked may itself claim more tractability than complex situations have. Human situations re-tie themselves while one works on them.

Second, Cilliers never rested on the mathematics. He used incompressibility as a way of saying that a complex system's behaviour depends on all its interactions and its history, so that any description leaves out something that may matter. He also wrote that the modeller is inside and that models have effects. The outside observer is the critic's addition.

Third, altered compressibility needs a measure and a measurer. After an intervention, who says that the situation is now more regular? That judgement is a description, made by someone, and it leaves things out. The regress returns one level up.

Fourth, immersion excludes as well. A participant attends to some things and misses others. The garden of Chapter 53 makes the point: the walker sees no more than fifteen stones at any step. Matching complexity is a way of being limited. It is not an escape from limits.

Fifth, and most seriously: to alter the compressibility of a human situation is to make people more regular. That is what a state does when it standardises names and measures, and what a manager does with a target. Cilliers's ethics was written against exactly that. A doctrine that celebrates compression by intervention has given up the warning.

The Ruling

The first reply is partly right, and it sharpens the claim. Understanding a knot is a kind of model. But it is a participant's model, a know-how of moves, and it was got by handling knots. More important, the knot shows that compressibility is relative to the knower and is changed by acting. That is the point at issue. That human situations re-tie themselves is true, and it supports the thesis: they are not fixed strings.

The second reply should be accepted as a correction of tone. Cilliers was a more careful writer than the bare word suggests, and he is no naive external observer. The criticism is of the concept he chose and of what it does in the hands of those who repeat it. A concept that defines complexity by the failure of a smaller copy keeps the copy-maker's position at the centre of the theory, whatever is said elsewhere about the observer being inside.

The third reply is correct and is not an objection. Judgements of regularity are made from within and are partial. The thesis does not claim a measure from nowhere. It claims that there is none, for compressibility or for anything else.

The fourth reply is accepted in full. Participants are limited. The thesis distinguishes kinds of limit and does not deny that there are any.

The fifth reply is the important one, and it reverses the practical lesson while leaving the conceptual one in place. If compression in human systems is an act that changes them, then Cilliers's ethics does not weaken. It becomes more severe. On his account the modeller answers for what the model leaves out. On the account defended here, whoever compresses answers for what the situation is made into. Every signifier set, dashboard and classification is a compression performed on people, and it makes them more describable in its terms. That is Thesis VIII, the degradation of instruments, seen as a general law. It is also Scott's legibility.

Incompressibility as a fixed property seen from outside: the challenge holds Against Cilliers's ethics of framing: the challenge fails, and the ethics is strengthened "There are no boundaries": holds as ontology, and boundary critique keeps its use against boundaries actually drawn

What Follows for Anthro-Complexity

Snowden's two sentences refute the doctrine of fixed incompressibility. They also bind their author.

If all diagnosis is intervention, then SenseMaker is an intervention, a Cynefin session is an intervention and an estuarine map is an intervention. Each alters the compressibility of the organisation it is used in. Stories told to a fixed set of signifiers become more alike. Problems sorted into domains are handled in more standard ways. Constraints once mapped are treated as the constraints there are. The instruments were built to read complexity without reducing it, and by Snowden's own principle they cannot merely read.

The second half of his sentence is the remedy, and the disputation adopts it as a rule. All interventions should be designed as diagnostic instruments. Apply that to the instruments themselves. A deployment should be designed so that it reveals what it is doing to the organisation: whether significations are converging, whether outliers are thinning, whether people have begun to tell stories for it. Those are the wake reader, the frame rotation and the failure conditions of Chapters 45 and 52. They follow from a sentence written in January 2024.

What Follows for the Disputation

Three amendments.

On Cilliers. The book's earlier reliance on him is narrowed. It keeps his ethics of provisionality and his insistence that framing has consequences. It gives up incompressibility as the ground for them. The ground is now that compression acts.

On the five orientations. They are kept and re-based. Philosophical inquiry, provisionality and creativity stand as they are. Boundary critique is kept for boundaries drawn. Critical pluralism is to be a pluralism of participations as well as of models: of people who have worked inside a situation in different places, and whose knowing meets. The mutual dashboard and the joiner of Chapter 53 are attempts at that.

On the conclusion. The statement says that complexity practice is the development of better judgement in worlds that exceed our descriptions. The amended sentence is that it is the development of better participation in worlds that our descriptions help to make.

A Fourteenth Dialectic

Chapter 55 listed thirteen. This is the fourteenth.

One side. Complex systems cannot be compressed. Every model excludes, and the modeller answers for the exclusion. The other. Compression in human systems is done from within and changes what is compressed. There is no fixed object to fall short of. What each holds. The first holds humility about description and an ethics of framing. The second holds the participant's position and the fact that describing acts. What each pays. The first keeps the outside observer in the definition of its central term. The second, unguarded, licenses making people regular. What emerges. An ethics of compression as action: whoever compresses is inside, changes what there is, and answers for what it becomes. Resolved in concept. The practical consequence is the reconstruction.

A MODEL THAT LEAVES THINGS OUT IS THE LESSER PROBLEM —

A DESCRIPTION THAT MAKES ITS OBJECT IS THE GREATER —

DESIGN EVERY INTERVENTION TO DIAGNOSE ITSELF

Addition: The Ruling Amended in Chapter 58

This chapter ruled that the challenge to incompressibility fails against Cilliers's ethics of framing and strengthens it. Chapter 58 examines that ethics in its fullest statement, the paper by Woermann and Cilliers, and amends the ruling. What the challenge strengthens is the responsibility of whoever frames or compresses. The ethics constructed to meet that responsibility, a provisional imperative addressed to the framer's state of mind, is found to fail. The sentence above stands in its place under the Additive Covenant, and the later ruling governs.

Addition: Thesis X in Cards

Chapter 57

Flinn: Organisational Development in Practice, and the Same Questions Put to It

4,258 words, about 19 minutes of reading

The Book Behind the Web

Chapter 51 used A Web of Relations, the web edition that Mark Downham edits. Behind it stands a printed book: Kevin Flinn's Organisational Development in Practice: A Complexity Approach, published by Routledge. Flinn teaches executive education at Liverpool John Moores University. He was formed in the Doctor of Management programme at Hertfordshire under Ralph Stacey, Douglas Griffin and Chris Mowles, and the book is in part a tribute to Stacey.

The book does not mention Snowden or Cynefin. It is used here for that reason. It shows what a practice looks like when it is built from the other side of the fault line, by a Dweller in the Conversation, with no instrument in view. A disputation that has asked anthro-complexity to change should look hard at the alternative it keeps pointing to, and should ask the same questions of it.

What Kind of Book It Is

It is an unusual book for its genre, and the form is part of the argument. Six personal essays by Flinn alternate with short accounts by other practitioners: James Traeger, Rob Warwick, Rikke Horup, Helen Charlwood with Adri van de Merwe, and Rebecca Nestor. Between them are poems and pieces of memoir about growing up in Liverpool. Flinn says plainly that it is no how-to book. The ways of working in it can be carried elsewhere, he says, as ways of being and never as recipes.

Three decisions about form deserve notice.

The word. He writes organisational development, and avoids organisation development, because the noun invites people to treat an organisation as a thing.

The cover. His earlier book on leadership had a single dandelion clock on the cover. This one has a field of them. The change is from the few at the top to everyone involved in the patterns of interaction that we call an organisation, and to the wider community.

The literature review. In place of the usual survey he reads two books closely, one of them James Joyce's Ulysses, on the ground that a novel tells more about complexity and the human condition than the standard texts on leadership together. He reports that the Hertfordshire faculty taught with fiction as a matter of course: Poe's tale of the maelstrom for Elias's involved detachment, and Orwell's "Shooting an Elephant" for power.

The Position

Flinn's statement of the theory is brief and orthodox for his school. From the perspective of complex responsive processes of relating, organisations are neither machines nor systems, and he adds that they are not complex adaptive systems either. They are patterns of interaction among people who are at once cooperating and competing to get things done. Those patterns can be influenced by anyone and controlled by nobody. Power, following Elias, is a feature of every relationship: whoever needs the other more has less of it.

Two principles are taken from Stacey.

Take experience seriously. Work with how things are. Do not be drawn into abstractions or into idealised pictures of how they ought to be. Flinn's book is about the everyday, including what is usually ignored, discounted or thrown away, and he says his aim is to notice and honour those things without idealising them either.

Practical judgement. He uses Aristotle's phronesis for what guides action when available knowledge and skill run out, which in organisational life is most of the time.

The foreword draws the line between this position and the rest of the complexity field more sharply than Flinn does, and in terms that bear on this disputation. It describes two camps. The larger holds that complexity can be put to use: with the right tool or framework, or on the understanding that it applies only in certain situations. Flinn is placed in the second camp, which holds that if the world is complex it is so all the way down. The foreword adds a sentence that could stand as the school's reply to every instrument: the most effective tool a practitioner can develop is their own practice of mature citizenship.

Four Arguments That Bear on the Disputation

1. See it, say it, sorted

Flinn takes a slogan from British railway announcements as the title of a chapter on facilitation. The assumption he attacks is that once a problem has been seen, which is to say framed, and said, which is to say named, it can be sorted, which is to say tamed for good. His reply is that very little is ever sorted for ever. Practitioners make a difference. But while the underlying structures, relations of power and ideologies stay in place, not much changes over time. He recalls how much seemed to shift in ways of working during the pandemic, and how soon it reverted.

The word sorted has a second sense that he does not press and this book may. To sort is also to classify. The assumption that a situation, once placed in its category, is on the way to being dealt with is the assumption behind every sorting framework. The Garden study made the argument at length against one of them.

2. Diagnostic and dialogic

In his chapter on how complexity has been taken up in organisational development, Flinn discusses a distinction made by Gervase Bushe and Robert Marshak. Diagnostic development collects data about an organisation, analyses it and prescribes. Dialogic development works through conversation and changed narratives. Bushe and Marshak treat the two as complementary: the first for problems of cause and effect, the second for wicked problems.

Flinn reports the criticism of Mark Cole, who calls the split a false binary. Diagnostic work, Cole argues, reinforces the prevailing order from a hidden position, and dialogic work does the same from a better disguised one, since it presents itself as more honestly engaged. Both are enmeshed in power and reproduce it. Cole's alternative is that practitioners should foster a shared appetite for inquiry that begins by bringing the workings of power in that workplace into the open.

The parallel with Cynefin is plain, and neither Flinn nor Cole draws it. A scheme that assigns analysis to one kind of problem and conversation or experiment to another is the same both-and. The foreword's remark about those who hold that complexity applies only in certain situations is aimed at such schemes.

Snowden's reply is in the sentence used in Chapter 56. All diagnosis is intervention. That sentence dissolves the diagnostic and dialogic split as thoroughly as Cole does, and from the instrument-maker's side. Here again the two schools say nearly the same thing and have not noticed.

3. Speaking truth to power is a judgement

The profession's current enthusiasm for speaking out is the subject of Flinn's second chapter. He makes two objections. The literature treats speaking truth to power as progressive in itself and to be encouraged nearly always. And it implies that those with more power are unaware of the inequities and will hurry to correct them once told. He doubts both. The consequences of speaking cannot be foreseen and may not appear for years. So speaking truth to power is a practical judgement. It cannot be turned into a moral imperative. He draws on James C. Scott's account of the poor of a Malay village, who sang one tune before the rich and another among themselves.

This bears directly on narrative instruments. An invitation to tell one's story is an invitation to speak, issued by those with more power, on the assumption that hearing will lead to remedy. Flinn's two doubts apply to it. The teller is making a practical judgement about what it is wise to say. The instrument records the result of that judgement.

He asks one of his contributors a question that the disputation should have asked earlier in its own voice: what happens when you do not like the tune that he who pays the piper is asking you to play? It is the positional reason of Chapter 17 put to a working consultant.

4. The large group

The practice Flinn describes most fully is the convening of large groups. He is a founding member of a programme called Creating Large Group Dialogue in Organisations and Society, convened by Teresa von Sommaruga Howard and based on the work of Patrick de Maré, a colleague of S. H. Foulkes and a co-founder with him of group analysis in Britain.

A large group in this tradition is some dozens of people seated together with no agenda. Flinn is candid about what that is like. It provokes anxiety and it exposes. Participants expect the convenor to call the steps of a familiar dance, and he says that without his own experience in such groups he might well have given in and done what colleagues expected. What the group offers, in a phrase he quotes from de Maré, is outsight as opposed to insight: a view of the dominant and silencing cultures as they form in the room, with the chance of repair. De Maré's word for the climate sought is koinonia, which he glosses as impersonal fellowship: a condition in which people who are not friends can speak, hear and think freely together.

Flinn has run what he calls Leadership Experience Groups on this basis, including with women leaders in Liverpool and in Ukraine.

The contributors' accounts give the texture. Warwick describes a programme for senior leaders built round questions that participants ask of their own workplaces: how power shows in small gestures such as a sideways glance, what they are not being told and not hearing, how they learn together, and how theory changes what they see. Horup describes taking over a long-established Danish programme and what happened to staff and students when the pedagogy changed. Nestor describes climate cafés, temporary groups in which citizens can bear their distress about the climate together.

The Dialectic With Anthro-Complexity

Set beside the method chain of Chapter 5, Flinn's practice is its mirror image on almost every point.

Anthro-complexity Flinn's organisational development
Unit The story, signified by its teller The exchange, in the room
Reach Thousands, at a distance Dozens, in presence
What is produced A pattern that can be shown to those who were absent A change in those who were present
The practitioner Designs conditions and reads Convenes, and bears the anxiety with the group
Where power appears In who filters and who attributes In the room, as it forms, and in who pays
What is taken away Data, maps, a portfolio of probes Formed people, and writing
Characteristic risk The frame nobody examined The room that never reaches the corridor
Account of failure Safe to fail Very little is ever sorted

Three things follow.

Each has what the other lacks. This was the finding of Chapter 51, and the book confirms it in detail. Flinn's large group is a formative encounter in Snowden's sense, with the tellers present as parties. It has no way to include the nine thousand who were not in the room. Snowden's instrument includes them, as contributors to an encounter that others have.

They agree on more than either says. Both refuse idealised end states. Both hold that people are capable. Flinn's honouring of the everyday is as Pelagian as anything in Snowden. Both centre practical wisdom. Both treat diagnosis as an act with effects. Both suspect heroic leadership: one of Flinn's chapters is titled "No more heroes", and it argues that conventional leadership education promotes self-interest and belittles leading for the common good. And both use fiction to think with. Snowden reports using gangster films to convey his estuarine material. Flinn uses Joyce.

The honest difference is about what can be handed over. Flinn says his ways of working travel as ways of being. Snowden builds things that travel as objects: a framework, a kit, a piece of software. An object can reach anyone and can be used without understanding. A way of being cannot be misused in that way, and it reaches only those who have been formed in it.

The Same Questions, Put to Flinn

Immanent criticism does not exempt the alternative. Flinn's practice has constitutive norms: take experience seriously, attend to power as it forms, honour the everyday, stay with how things are. The disputation's questions apply to it.

Who is in the room? A large group of forty is a boundary drawn, whatever Chapter 56 says about boundaries. Senior leaders, doctoral students and those with time to attend are present. The people Flinn most wishes to honour, those whose work is discounted, are the least likely to be there. The dandelion field on the cover is a just image of the theory. The rooms described hold fewer seeds.

Who convenes, and who formed the convenor? Flinn recalls his first day on the doctoral programme, when a member of faculty told the six students that half of them would not finish. That is a formation with a selective gate. Chapter 51 recorded the objection that formation is the Augustinian model returned: a novitiate, at the end of which one is fit to practise. The book does not answer it. The foreword's ideal of mature citizenship points the right way, since citizenship is open to all, and it would be good to know how it is acquired by those who cannot spend three years in a circle of chairs.

Is the convenor's structure declared? A gathering with no agenda has a structure: a time, a seating, a convenor who declines to lead, and a theory of what should happen. Snowden's charge against the process schools is that they disavow this. Flinn is better than most on the point. He names the anxiety, says what he was tempted to do, and gives the lineage of the method. That is a declination of the kind this book asks of instrument designers.

What would show that it failed? Flinn's sentence that very little is ever sorted is honest, and it is also the form of statement that Chapter 52 called an alibi when Luhmann's users make it. If little changes whatever is done, no outcome counts against the practice. The same request must be made as was made of Snowden. A practice that takes experience seriously should say what experience would lead it to do otherwise.

Does the writing return? Flinn makes writing a practice in its own right: reflective narrative, written by the practitioner about her work, read and questioned by peers. It is the school's research method and its discipline of return. The question the reconstruction asks is whether the people written about read the account and answer it. Where they do, the method is a model for what Chapter 45 calls the narrators' reading.

What the Book Gives the Reconstruction

Four additions.

  1. The convenor's anxiety as a datum. The protocols of Chapter 45 ask designers to declare what they expect and hope. Flinn's account suggests a further line: what the convenor was tempted to do under pressure, and whether she did it.
  2. Outsight. The reconstruction's mutual dashboard shows patterns. De Maré's word names what a population needs from seeing them: recognition of the silencing and dominating habits that form among themselves. It is distinct from individual insight and distinct from a sponsor's reading.
  3. Impersonal fellowship. Scale was listed in Chapter 55 as an open dialectic. De Maré's idea that people who are not friends can think together in numbers larger than a team is a resource for it. A nested set of large groups, each reading its own patterns from an instrument, is a design nobody has tried, and it would join the two schools at the point where each is weakest.
  4. Judgement about speaking. Every protocol that invites people to speak should carry Flinn's caution. To speak is a practical judgement with unforeseeable consequences, and those who invite it are not thereby committed to act.

For the opposing side. A practitioner of anthro-complexity may say: this is an attractive book and it proves the point. It describes rooms, and what a few dozen fortunate people learned in them. Meanwhile a health service of a million staff must decide what to do next year. An approach whose answer is that very little is ever sorted, and whose instrument is a circle of chairs, has left that decision to the people who will take it anyway, with spreadsheets. The Designer of Conditions is at least in the building where the decision is made.

The disputation's answer. That is the scale objection, and it has been conceded three times already. What Flinn's book adds is evidence about what the instrument cannot do and the room can: let people see power as it forms among them, in time to repair it. A health service needs both. Neither school has yet shown how to give it both at once, and each would be a better judge of the other's attempt than of its own.

A FIELD OF DANDELIONS, AND A CIRCLE OF CHAIRS —

VERY LITTLE IS EVER SORTED —

THE BEST INSTRUMENT IS A MATURE CITIZEN, AND THERE MUST BE MORE OF THEM

Addition: Flinn at Greater Length

The chapter above summarised Flinn's book. This addition goes further into six passages, because they contain evidence, in the form of a practitioner's own experience, on points where the disputation has so far only argued.

The exercise called "Sort!"

In his chapter on rethinking organisational development Flinn describes a stock classroom exercise. Students are given slips of paper naming behaviours and two sheets headed Leader and Manager. He gives a one-word instruction: sort. Most groups put the headings at opposite ends of the table and begin: leader, manager, manager, leader, where does this one go? Groups usually end with a third pile for "both". Now and then a group puts the two headings side by side and all the slips in one column beneath.

His point is that the exercise presents the behaviours and the two categories as given. The only question it allows is which pile. That leaders differ from managers, strategic from operational, transforming from transacting, is built in before anyone speaks.

The passage is a small experiment on classification, and it bears on every sorting framework. People told to sort will sort. They will argue about placement and not about the headings. The occasional group that refuses the headings shows that refusal is possible and rare. A Cynefin contextualisation is designed to avoid exactly this, by having items placed before the boundaries are drawn. The exercise shows what the design is up against, and what happens when the framework is used as most people use it, with the four headings already on the wall.

The car park

Flinn says that he often asks groups how many meetings they have attended at which the real conversation took place in the car park or the staff room the next day. He then asks what practitioners think they are doing when they encourage people to speak out. He recalls Stacey putting the question of how it would be if everyone did say to one another exactly what they thought. Stacey's answer was that this was his idea of hell. And Flinn records that many students on the doctoral programme who did speak out at work found themselves excluded, and that this was often what led them to change employer or career.

This is testimony on the hidden transcript from inside a school that prizes candour. It says three things. The offstage conversation is normal and not a defect to be designed away. Total candour is not a coherent aim. And speaking has costs, which fell on the people Flinn knew. An instrument that invites people to tell their stories is asking for a piece of the car-park conversation. The honest version of the invitation tells them what happened to others who gave it.

What the piper asks for

Flinn asked a contributor, James Traeger, what one does when one dislikes the tune that the payer calls. Traeger's reply names the tune. Clients have been schooled for a century in a theory of change descended from scientific management, and they ask for what he calls instrumental change. He lists its forms: fix it quickly, fix them and not me, fix that part and leave the whole alone, and fix it in a course or an away-day, since that is what learning is taken to look like. He says he is sympathetic. The people commissioning the work hope for, and were trained to expect, a system that can be perfected. He calls organisations an infinite game: something to be played out, with no solution.

This is the positional reason of Chapter 17 in a practitioner's words, and it matches Snowden's remark that the market rewards the deficit mode. The four forms of the request are worth keeping as a checklist. A deployment of any method that is quick, aimed at others, confined to a part and held off-site has been shaped by the piper, whatever the method's theory.

An event judged a disaster

Flinn describes convening a large group for a leadership network. There were long silences while everyone sat together. Colleagues wanted to break the group into small ones for a familiar activity. Afterwards they judged the event a disaster, and the senior sponsors distanced themselves from it and from him. He says that conducting a large group often feels like dancing on one's own, with everyone watching, and that without his experience in the large-group community he might have given the participants what colleagues later called more structure.

Three things follow for this book. The formative encounter that Snowden describes as professionally risky is here seen being risky: a real person lost standing with real sponsors. The account is a declaration of the kind the reconstruction asks for, including what the convenor was tempted to do. And it is honest evidence against the practice as well as for it. A method whose sponsors walk away has a problem of answerability to them, and Flinn does not hide it.

No more heroes

The chapter on leadership sets two views against each other. One is represented by Jeffrey Pfeffer's rules of power, which treat the pursuit of self-interest as the realistic path and include the rule that success excuses almost everything done to obtain it. Flinn reads this as the claim that working for the common good is an impractical road and that the end justifies the means. He argues that leadership education built on such premises promotes self-interest and belittles leading for the collective good.

The bearing on anthro-complexity is indirect and real. Snowden's methods are meant to reduce dependence on the heroic leader by distributing sensing. Flinn's chapter is a reminder that the people who commission such methods have often been educated by the other school. A dashboard of dispositions in the hands of someone trained on Pfeffer's rules is an instrument of a different kind.

Rooms, rituals and reading

Three of the contributors' accounts and one of Flinn's own add detail on how a room is made safe enough to think in.

Rebecca Nestor describes climate cafés: simple, hospitable gatherings in which people can speak of their distress about the climate. They are not therapy. Nobody is asked to commit to anything, and participants are asked not to persuade one another, so that those who feel they know or do too little are not overwhelmed. Each café opens with a round in which people choose an object from the natural world and say something in response to it. The small ritual lowers the fear of talking about feelings.

Helen Charlwood writes of working with a colleague who sees the world of work differently, and reaches for Gadamer's fusion of horizons, which she connects to de Maré's small culture formed in a larger group and to Foulkes's matrix.

Flinn's chapter on writing holds that reflective narrative, written about one's own practice and read by peers, is both the research method of his school and a discipline in itself. Its first lesson is reading. He cites writers for whom one becomes a person through what one reads.

Two points for the disputation. The café's opening round is an enabling constraint in exactly Snowden's sense, used by a school that distrusts designed conditions. Nobody escapes design. And the rule against persuading one another is a small case of the principle the Tidal Loom calls recognition, not alignment.

What the longer reading changes

The chapter above set Flinn's practice beside Snowden's as mirror images. The passages here soften the contrast in one way and sharpen it in another.

They soften it because Flinn's practice is full of small designed things: a one-word instruction, an object chosen from a table, a circle of chairs, a rule against persuasion. The Dweller designs conditions too. The difference is that the designer is in the room and can be seen and answered.

They sharpen it because the costs are shown. In Flinn's pages the people who bear the risk of the encounter are named or nearly so: the convenor whose sponsors withdrew, the students who spoke and lost their place. In an instrument, the risks are spread among thousands who are not named, and the practitioner's own exposure is small. Snowden says the encounter mode is risky for the practitioner. Flinn shows what the risk looks like when it is taken.

TOLD TO SORT, PEOPLE SORT —

THE REAL CONVERSATION IS IN THE CAR PARK —

AND THE CONVENOR WHO STAYS IN THE SILENCE PAYS FOR IT

Chapter 58

Thesis XI: Cilliers and the Ethics of Framing, Interrogated in Full

4,043 words, about 18 minutes of reading

Why Another Chapter on Cilliers

Chapter 56 ruled that the author's challenge held against incompressibility as a concept and failed against Cilliers's ethics of framing, which it was said to strengthen. That left the refutation partial. The ruling rested on a short account of the ethics. This chapter goes to the fullest statement of it, a paper Cilliers wrote with Minka Woermann and which appeared after his death: "The ethics of complexity and the complexity of ethics" (South African Journal of Philosophy, 2012). It sets the paper out, interrogates it step by step, gives its authors their strongest replies, and rules again.

The result can be stated at the start. The earlier ruling ran two things together. One is the responsibility of whoever frames or compresses, which is real and grows heavier on the argument of Chapter 56. The other is the ethics that Woermann and Cilliers build, which is a specific construction with a ground, an imperative and four supporting attitudes. The responsibility stands. The construction does not.

The Paper Stated

The ground

The paper follows Edgar Morin in separating two views of complexity. Restricted complexity, which it associates with the Santa Fe Institute, studies complex systems in order to find the rules that govern them, and hopes for a general theory. General complexity holds that complex phenomena are irreducible. The authors take the second view. They add, candidly, that they cannot say conclusively whether complexity is a feature of the world or a result of the limits of knowledge. For irreducibility they cite Robert Rosen: a system is complex to the extent that it admits encodings that cannot be reduced to one another.

From this the ethics is derived in four steps.

  1. We cannot have complete knowledge of complex things.
  2. So we must model, which means reducing complexity and leaving things out.
  3. What to leave out cannot be computed or justified objectively. It is chosen, and the choice rests on judgements about what matters.
  4. So an ethical element is built into all knowledge of complex phenomena. Ethics is constitutive of knowledge and of ourselves. It is no system of rules for right action added afterwards.

The practical upshot is a self-critical rationality that claims no objectivity and no special standing for its own grounds, and a duty of vigilance. The authors say plainly what this achieves and what it does not. In their words the ethics of complexity cannot do more than generate awareness that we are always in trouble. It cannot give guidance of substance.

The provisional imperative

To go further they turn to Kant. The categorical imperative has no content of its own. It is a test applied to principles one already has. They propose a counterpart, the provisional imperative: when acting, always remain cognisant of other ways of acting.

They note that the phrase is a contradiction in terms, since an imperative is absolute and this one declares itself revisable, and they accept the contradiction as the point. They also say what kind of command it is. Kant's tells us something about rules for action. Theirs concerns a state of mind or attitude in choosing rules. Their reason, taken from Derrida, is that concepts and rules have an all-or-nothing logic, so that we cannot do other than model and exclude. What can vary is our awareness of doing so. They hold that the awareness makes an ethical difference: someone who keeps it is more likely to be self-critical, to respect diversity and to revise.

Four operations

Four attitudes support the imperative.

Provisionality. Meaning shifts with context and no context can be fully specified, so every position, the good included, is open to revision. Positions should not be made to look natural.

Transgressivity. The critical stance must cross accepted limits and cannot merely confirm what is current. Here the paper turns to those a model leaves out. Borrowing from Alain Badiou, it calls them the non-existent: stakeholders who are not counted in our conceptual scheme. We owe them justice. But the moment we account for their interests we absorb their otherness into our scheme. So justice is an endless pursuit, and one must act and take a position while knowing one cannot get it right.

Irony. Irony affirms a position and undercuts its claim to be absolute. The authors use Lévi-Strauss's bricoleur, who makes do with what is to hand, and Derrida's remark that the engineer is a myth produced by the bricoleur. They say there can be no prior basis for arguing the merits of one strategy of life over another. Irony lets one own one's choices lightly. It "lightens the burden" of self-awareness, and involves not taking oneself or one's ideas too seriously.

Imagination. To be cognisant of other ways one must be able to imagine them. The arts are recommended as the means: novels and films bring us to sympathise with people we would otherwise find foreign.

The paper ends on trust. Since none of this guarantees ethical action, the complexity of ethics can be fostered only through trusting relations and active engagement with difference.

The Interrogation

Thesis XI. The ethics of framing is an ethics of the framer's conscience. It derives ethics from the limits of a model, addresses its one command to the modeller's state of mind, and assigns to the modeller every virtue it names. Those the model leaves out appear in it only as objects of the modeller's vigilance. Cognisance is not answerability. An ethics that asks for the first and provides no means for the second fails by its own aim, and it fails the test that Snowden sets for any theory of practice: nothing counts against it.

First: ethics is not downstream of modelling

The derivation makes ethics arise at step 3, when a modeller must choose what to leave out. Before that there is a knower facing a complex object and failing to capture it. The picture is of one mind, a world and an encoding. Rosen's criterion of complexity is stated in terms of encodings, and the paper's admission that it cannot tell whether complexity is in the world or in our limits shows how far the whole construction stays inside the relation of describer to described.

Two objections follow.

The first is that of Chapter 56. Much knowing of complex situations is done by taking part, and makes no copy from which things are missing. A ward sister, a caretaker and a ship's navigator do not stand before their situations as modellers. For them the founding scene of the paper does not occur.

The second goes deeper. People are answerable to one another before anyone models anything. A person who walks into a hospital ward is at once in relations of dependence, expectation and trust. Mead's conversation of gestures, Elias's interdependence and Jaeggi's practices with their internal norms all say this in different ways. Ethics is a feature of that condition. It does not wait for a failure of representation in order to begin. To derive it from the incompleteness of models is to make the modeller's predicament the origin of obligation, and that puts the origin in the wrong place. The paper's opening claim is that ethics is constitutive and no add-on. Its derivation adds ethics to epistemology, as a consequence of what cannot be known.

Second: the imperative commands an attitude

"Always remain cognisant of other ways of acting" is addressed, by the authors' account, to a state of mind. Four difficulties arise.

It cannot be checked. Whether a planner was cognisant of alternatives when she drew the map is known to her alone. Two planners can make the same exclusion, one with exquisite awareness and one with none, and the excluded are in the same position.

It costs the framer nothing and gives the excluded nothing. Awareness of other ways of acting is compatible with never taking one. The imperative is satisfied by a decision-maker who notes that others would have chosen differently and proceeds.

It is empty in a way Kant's is not. The categorical imperative, whatever its faults, rules things out. A maxim that cannot be willed universally fails. The provisional imperative rules out no action at all, only a frame of mind. The authors say it is substantively empty, like Kant's. It is also empty as a test.

It is offered as a contradiction and defended as one. The authors grant that a provisional imperative is a contradiction in terms and say that this is exactly the point. A critic need not accept that reply. The contradiction arises from a premise, Derrida's claim that concepts and rules are all-or-nothing, so that a rule is either followed or not. That premise is contestable. The later Wittgenstein argued at length that following a rule is a practice among people, sustained by training, agreement in judgement and correction, and that no rule fixes its own application. On that account rules are provisional in their working without any paradox, since they live in the practice of those who apply and dispute them. Aristotle's phronesis and Dewey's inquiry say something similar. The "impossible position" the authors embrace is produced by a picture of rules they did not need.

Third: the alibi

Snowden's post on Luhmann supplies a test that the disputation has applied to anthro-complexity and to Flinn. A theory is an alibi when success and failure are both redescribed in its terms, so that nothing counts against the practitioner.

Apply it here. A framing that turns out well was made with due cognisance. A framing that harms was made in the knowledge that we cannot get it right, that we are always in trouble, and that every model excludes. The authors say that responsibility on their account is not something one can fail to live up to. They quote with approval a description of it as no basis of guilt, a recursive modality and an always renewable openness. That is a responsibility with no condition of failure. Irony is then recommended because it lightens the burden.

It should be asked whose burden is lightened. The framer's is. The burden of those the frame left out is whatever the frame did to them, and irony on the framer's part does not touch it.

So the ethics of framing, as constructed, meets the conditions of an alibi more fully than the uses of Luhmann that Snowden criticises. This is a hard conclusion about a paper whose intention is the opposite, and the intention is not in doubt. The form of the argument is what is at issue.

Fourth: the excluded never speak

The paper's concern for those a model leaves out is real and is its best feature. Consider how they appear.

They are the non-existent: those not counted in our scheme. They are compared, through Badiou's image, to an animal in flight that the hunter must not pin down. We must do them justice, and we cannot know them in their otherness. If we account for their interests we assimilate them. What we can do is remain vigilant, cross our own boundaries, and enlarge our sympathy by reading novels and watching films.

At every point the agent is the modeller. The excluded are imagined, pursued, done justice to, and protected from being absorbed. They do not describe themselves, contest the frame, see the model or answer it. The remedies proposed are improvements to the modeller's sensibility.

Three bodies of work already used in this book show what is missing.

Ulrich gives the affected a standing to question an expert's boundary judgements, and a set of questions with which to do it. The excluded become interrogators.

Fricker, whom Snowden cites, locates the wrong in not hearing people's own testimony and in leaving them without terms for their experience. The remedy is to hear and to share the making of terms. It is not to imagine them better.

Snowden himself says that people have the right to give meaning to their own experience before anyone else does.

Against these, the paper's claim that to account for the other's interests is to assimilate them reads as a reason given in advance for not asking. The difficulty it names is real when a modeller represents others inside her model. It lessens when the others speak for themselves and can correct what is made of what they said.

The Garden study has an image for this. The makers of the garden are present in what is looked at, and their names are on the back of a stone. Woermann and Cilliers ask the viewer on the veranda to remember that such people exist. They do not suggest turning the stone over or asking them in.

Fifth: omission and commission

The ethics is an ethics of what is left out. Its key words are exclude, reduce, limit and leave out. Chapter 56 argued that in human affairs a description does more than omit. It acts on what it describes and makes it more regular in its own terms.

The paper touches this twice and does not develop it. It notes, citing Luhmann, that models create further complexity, and it says that we become the stories we tell. Had these been followed, the ethics would have had to change. A framer who has omitted something can be asked to stay aware of the omission. A framer who has remade a workplace, a diagnosis or a population by describing it cannot discharge that by awareness. She has done something to people, and they can ask her to account for it.

Sixth: no basis, or a different basis

The authors write that there can be no prior basis from which to argue for the merits of one life strategy over another, and that the good is always open to deconstruction. From this they conclude that an ethics of substance is impossible and that only a formal attitude remains.

The choice offered is between a foundation fixed in advance and no substantive judgement. Jaeggi's book, set out in Chapter 49, exists to refuse that choice. Forms of life can be compared without a prior foundation, by whether they deal with the problems they set themselves and whether they can learn. That is a criterion with content, reached from inside. Peirce's maxim not to block the path of inquiry is another.

The paper does not keep to its own restriction. It says that no one can contest the urgent need to move toward a more sustainable future. It deplores Western ideals passing as universal. It sides with the disenfranchised. These are substantive commitments, and good ones. None follows from the provisional imperative, which is equally satisfied by someone who remains cognisant of other ways and chooses the unsustainable one. The ethics of the paper is better than its meta-ethics, and has no visible source in it.

Seventh: general complexity is still the modeller's complexity

Morin's distinction is meant to separate the Santa Fe programme from a more thoughtful position. Both sides of it concern the status of our models. Restricted complexity hopes the model will one day be complete. General complexity knows it never will be. Neither considers that for most people in most complex situations there is no model at all, and they are in the thing.

The author's term for the paper's position is therefore the same as for incompressibility: a paradox of the external observer. General complexity is the external observer grown modest.

The Case for the Other Side

For the opposing side. Woermann and Cilliers have replies, and they are not weak.

First, the critic demands a substantive ethics and the paper explains why none can be had. To complain that the imperative is empty is to complain that it is what it says it is. Kant's was empty too, and the authors do not claim more than he did.

Second, attitude matters. A surgeon who knows she may be wrong operates differently from one who does not. To dismiss cognisance as costless ignores how much harm is done by people certain of their frames. The paper says so: awareness makes one more likely to revise.

Third, the paper is not only about a state of mind. Transgressivity demands bold action and a position taken. It speaks of proactive attempts to give voice to the disenfranchised. It ends on trust and engagement with difference. The critic has read the first half and ignored the second.

Fourth, participation solves nothing. Someone decides who takes part, what is asked and what is done with the answers. Those are framings. The participants' own descriptions are models and exclude in their turn. The critic's dashboards and return paths are one more scheme, and the provisional imperative applies to them.

Fifth, "there are no boundaries" is false in the only sense that matters. To act is to cut. A budget goes to one ward and not another. Whoever decides has excluded, whatever her ontology.

Sixth, the charge of alibi is unjust. The paper insists that we take responsibility for our positions even when the results are bad and unforeseen. That is the opposite of an excuse.

The Ruling

On the first reply. The complaint is not that the imperative lacks content. It is that it tests nothing. Kant's imperative, applied to a maxim, gives a result. Applied to any proposed action, the provisional imperative gives none. And the claim that no ethics of substance can be had is answered by Jaeggi, who has one without foundations.

On the second. Accepted in part. The disposition is a virtue, and a practitioner without it is dangerous. The reconstruction asks for it under other names: dated limits, declared expectations, a willingness to be changed. What is denied is that the disposition is an ethics, or that it meets what the paper itself sets as the aim, which is justice to those left out. A virtue of the framer is necessary for that. It is far from enough.

On the third. Accepted as a correction to emphasis. The second half of the paper does reach toward action, voice and trust. But these are named and not built into the argument. The imperative, the four operations and the account of responsibility are all in the first person of the modeller. "Giving voice" is a phrase in one sentence. And the turn to trust at the end is the paper pointing beyond its own construction, toward the relations in which, on this chapter's first argument, ethics was located all along.

On the fourth. This is the strongest reply and is accepted without reserve. Participation is framed. The reconstruction's devices are schemes and are provisional. The difference that remains is the one on which the thesis turns. A scheme in which those affected can see the frame, say what it leaves out and answer what is made of them contains a means of correction that does not depend on the framer's state of mind. That is answerability. It is provisional too. It is still a different kind of thing from cognisance.

On the fifth. Accepted. Chapter 56 already ruled that the claim about boundaries holds as ontology and that acts of bounding are everywhere. Every decision cuts. The question is to whom the one who cuts must answer.

On the sixth. The intention is granted. The paper says we must take responsibility. It also says that this responsibility cannot be failed, is no ground of guilt, and should be borne with irony. A responsibility that cannot be failed and is owed to no named party is not the opposite of an excuse. It has the same form.

The derivation of ethics from the incompleteness of models: fails The provisional imperative as an ethics: fails. It commands an attitude, tests nothing and has no condition of failure The four operations: hold as virtues of a practitioner, fail as a substitute for the voice of the excluded That whoever frames or compresses is responsible for it: holds, and is heavier than the paper allows

The ruling of Chapter 56 is amended accordingly. That chapter said the challenge strengthened Cilliers's ethics. What it strengthens is the responsibility. The ethics constructed to meet that responsibility is refuted. Under the Additive Covenant the earlier sentence stands in its place, and this one governs.

What Should Replace It

A refutation should say what it would put in the place of what it removes. Four principles follow from the argument, and each has already appeared in the reconstruction.

  1. Ethics is prior. Obligation belongs to the relations people are already in. A practitioner does not acquire responsibilities when her model proves incomplete. She has them on arrival.
  2. Answerability over cognisance. The test of a framing is whether those it affects can see it, question it and be answered. The framer's awareness is welcome and is no test.
  3. The excluded speak. Nobody is to be done justice to in imagination who could be asked. Where they cannot be asked, as with the unborn or the absent, someone is named to speak for them, and that appointment is open to challenge.
  4. Conditions of failure. A framing, like a probe, states beforehand what would show it to have been wrong, and to whom the framer will then answer.

The four operations can be kept as habits within this. Provisionality becomes the dating of every limit. Transgressivity becomes the standing invitation to those inside a frame to break it. Imagination becomes what one uses while waiting to hear. Irony is kept for one's own theories and withheld from other people's injuries.

What This Does to Anthro-Complexity and to This Book

For anthro-complexity the result is favourable on one side and demanding on the other. Snowden's ethics, as stated in the post on Luhmann, is of the second kind. It names a right, a wrong and a party to whom something is owed. It is an ethics of answerability in intention. The questions of Chapter 52 ask whether the instruments deliver it. Had anthro-complexity taken its ethics from Cilliers, it would have a framer's conscience and nothing more. It took it from elsewhere, and is better for it.

For the disputation the result is a debt. Chapter 37 and several cards leaned on Cilliers's argument from incompressibility to provisionality and thence to ethics. Two of those three steps are now withdrawn. The book's own demands, for declination, return and contest, were sometimes presented as applications of his ethics. They are better described as its replacement.

A fifteenth dialectic can be added to the fourteen of Chapters 55 and 56.

One side. Every frame excludes, and the framer must stay aware of it. The other. Every frame acts on people, and they must be able to answer it. What each holds. The first holds a real virtue and a just suspicion of certainty. The second holds a means of correction that does not depend on anyone's virtue. What each pays. The first leaves the excluded where they were. The second is slower, and is itself a frame. What emerges. Cognisance kept as a habit of practitioners, inside arrangements of answerability. Resolved against the ethics of framing as an ethics, in favour of it as a virtue.

Grant and follow.

Granted, with Woermann and Cilliers: every model leaves out some whose interests it affects, and we owe them justice.

Then it must be granted that justice is owed to them, and so that they are the ones entitled to say whether it has been done.

Then it must be granted that a framer's own awareness of having excluded them cannot settle that question, since she is the other party.

Then it must be granted that an imperative addressed only to the framer's state of mind leaves the question of justice in the hands of the party against whom the claim lies.

Therefore the aim of the ethics of framing, justice to the excluded, requires what the ethics of framing does not provide: a way for the excluded to see the frame and to answer it.

TO KNOW THAT ONE EXCLUDES IS A VIRTUE —

TO ANSWER TO THE EXCLUDED IS AN ETHICS —

THE FIRST CANNOT STAND IN FOR THE SECOND

Addition: Thesis XI in Cards

Apparatus

Glossary

3,279 words, about 14 minutes of reading

How to Use the Glossary

The glossary gives short definitions of the principal terms used in the disputation, with the chapter in which each is developed. Terms from anthro-complexity are given as the disputation reads them. Terms the author has coined or adapted are marked †. Chinese terms are given with characters and pinyin.

A–C

Abduction. Peirce's inference to the best explanation, the only form of inference that introduces new ideas. In anthro-complexity, the logic of safe-to-fail probing. The disputation adds that abduction is social, disciplined by a community of inquirers. Chs 5, 32.

Actant. Latour's term for anything, human or non-human, that makes a difference to a course of action. Anthro-complexity adopts it with an asymmetric qualification (only humans narrate). The disputation proposes actants as knots in a reflexive meshwork. Chs 2, 22, 30.

Additive Covenant. The rule of Principia Humanitatis: nothing is deleted, only inserted; amendments are made in place, dated, with earlier text kept visible. Preliminaries; Ch 47.

Address (addressivity). Bakhtin's principle that every utterance is directed to someone and shaped by the anticipated response. Applied to micro-narratives, which are addressed to the instrument and its imagined readers. Ch 19.

Adjacent possible. Kauffman's term for the configurations reachable in one step from the present, which expand as they are explored and cannot be listed in advance. Ch 7.

Agency, adjacency, tendency. The Compass and the Loom's three levers of collective leverage. Each alone fails. Ch 42.

Deeper.* Boxes in the text giving technical excursions into the deeper ground of physics, biology or mathematics. Preliminaries.

Anthro-complexity. Snowden's name for the study of complexity in human systems, distinguished from complexity in general and from computational complexity. The disputation's subject. Chs 1–6.

Asymmetry principle. Snowden's claim that, while river and planners are both actants, only the planners can change the story they tell about the other; "where anthro-complexity starts". The disputation applies it to anthro-complexity's own descriptions (Thesis II) and re-enters it among humans. Chs 4, 18, 20.

Augustinian / Pelagian. Snowden's analogical terms for interventions that supply what a system lacks from outside (Augustinian) and those that help a system use what it has (Pelagian). The disputation proposes Pelagian content, Augustinian form† for instruments that help a system see itself but must be supplied from outside. Chs 5, 17.

Barnesian performativity. MacKenzie's term, after Barry Barnes, for the use of a model making reality more like the model. Ch 34.

Boundary critique. Ulrich's and Midgley's practice of examining the boundary judgements that define a systems inquiry, especially who is "affected but not involved". Ch 12.

Bricolage. Lévi-Strauss's making-do with what is to hand. With exaptation and détournement, the Mariner's method. Ch 41.

Children's party. Snowden's early image of complexity management: boundaries, attractors, amplify and dampen. A human image, with a reflexive medium and the reader inside, set against the later estuary. Ch 21.

Closure (constraint closure). Montévil and Mossio's concept of a set of constraints each maintained by others in the set. The disputation applies it to human communities, where energy cost becomes a property of position. Chs 22, 29.

Commensuration. Espeland and Sauder's term for making unlike things comparable on a single scale, so that what the scale does not capture becomes invisible. Ch 26.

Complementarity (Path or Place). Bohr's principle of mutually exclusive experimental arrangements. In the Pyxis, a reading chooses between knowing where and knowing how. Ch 40.

Complex adaptive system (CAS). The Santa Fe programme's model of systems of adaptive, rule-following agents. Anthro-complexity's principal neighbour. Chs 7, 8.

Complex responsive processes (CRP). The Hertfordshire perspective of Stacey, Griffin, Shaw, Mowles and Flinn, grounding human complexity in Mead and Elias. The neighbour not named. Ch 11.

Constitutive constraint. The constraint that makes a community what it is, reproduced in compliance and in mapping. The "missing third" beside governing and enabling constraints (The Compass and the Loom). Chs 22, 29.

Constructor. Deutsch and Marletto's entity that causes a transformation and retains the capacity to do so again, unchanged. The disputation argues that the import cancels the human distinction, because in human affairs everyone is a reagent. Ch 22.

Correspondence. Ingold's term for lines of life answering one another over time. The disputation proposes it as a model for instruments: answering, not capturing. Ch 30.

Counter-performativity. MacKenzie's term for the use of a model making reality less like the model. Ch 34.

Crew. Snowden's term for a temporary identity aligned to a role, formed around a task. Chs 4, 19.

Crowd crystal. Canetti's small, rigid, constant group around which crowds precipitate. Read in The Compass and the Loom as the seed of representation. Ch 42.

Cynefin. Snowden's sense-making framework distinguishing clear, complicated, complex and chaotic domains, with the aporetic at the centre. Chs 5, 6, 44, 45.

D–G

Dated limits. Tidal Loom principle: record each "cannot be changed" with a date and condition. Chs 22, 45.

Declination before direction. Tidal Loom principle: each signifier set states what its designers expected, hoped for and would prefer not to find. Chs 20, 45.

Decoherence (into cliché). In physics, the loss of interference through interaction with an environment. In the Pyxis, saturating expectation selects safe, reportable clichés. Remedy: protect pockets of low monitoring. Chs 26, 40.

Descriptive reflexivity†. The disputation's term for the human difference of kind: human systems are changed by public descriptions of themselves, which members take up and act on. Ch 14.

Dispositional state. A complex system's readiness to move in some directions more than others, given its configuration. With vector, its direction and speed. Ch 2.

Dispositif. Foucault's heterogeneous apparatus in which power is moved into the arrangement of positions. Ch 23.

Doxa. Bourdieu's domain of what goes without saying in a field. Ch 33.

Double hermeneutic. Giddens's term for the social sciences interpreting a world of subjects who already interpret it, whose self-interpretations then absorb the social sciences' concepts. Ch 14.

Duality of structure. Giddens's principle that structure is both medium and outcome of the practices it organises. Chs 22, 33.

Entangled knowledge. In Snowden's tacit–explicit–entangled trialectic, knowledge that exists only in the interaction of people, artefacts and context. Distinguished from Barad's entanglement. Ch 4.

Escape, conquest, grating. The Compass and the Loom's three stances towards automated systems: from outside, from above, from inside with stakes. Ch 41.

Established and outsiders. Elias and Scotson's figuration in which a more cohesive group stigmatises a less cohesive one through praise and blame gossip. Chs 28, 45.

Estuarine mapping. Snowden's method of placing constraints on axes of energy cost and timescale. The disputation reads its root metaphor (Thesis III) and proposes a reflexive protocol. Chs 5, 21, 45.

Exaptation. Gould and Vrba's co-optation of a trait for a new function. Central to Snowden's practice and the Mariner's method. Ch 41.

Felicity conditions. Austin's conditions for a performative to succeed, including appropriate persons and uptake. Ch 34.

Figuration. Elias's web of interdependent people, shot through with power ratios. Ch 28.

Fisherman in the maelstrom. Elias's image, from Poe, of detachment achieved within involvement. Proposed as the practitioner's figure against the estuary's sailor. Ch 28.

Frames, not seers. Tidal Loom principle: rotate the search template, so different people see different gorillas. Chs 9, 26.

Generalised other. Mead's organised attitudes of the community, taken by the individual towards herself. Ch 27.

God trick. Haraway's term for the claim to see everything from nowhere, which disguises a particular powerful position. Ch 39.

Grating. In The Compass and the Loom, a checkpoint where a human must sense before a machine decides; tied by "suspenders" (trust, informal networks, people who notice). Mesh size matters. Chs 24, 41.

H–M

Habitus. Bourdieu's system of durable, transposable, embodied dispositions. Ch 33.

Harvest loop. The Compass and the Loom's term for the conversion of human noticing into training data for automation. Ch 24.

the closed person.* Elias's "closed person", the self-image of the sealed individual facing a world. Against it, open people. Chs 19, 28.

Horizon of observation. Hutchins's term for the portion of a distributed task each member can see. Overlapping horizons enable error detection, learning and emergent reorganisation. Ch 31.

Hysteresis. Bourdieu's lag of habitus behind changed conditions (the Don Quixote effect). Proposed as a third axis for estuarine mapping. Chs 33, 45.

Interactive kind. Hacking's term for a kind whose members are affected by being classified, as against indifferent kinds. Chs 14, 33.

Interpretant. Peirce's third term of the sign relation: the effect of a sign on an interpreter. The disputation argues it should belong to the community, not the analyst alone. Chs 25, 32.

Kairos. Opportune, qualitative time, as against chronos. In the November post, the turn of the tide. The disputation adds that in human affairs waiting is a gesture. Chs 3, 21.

Looping effect. Hacking's term for classifications that change the classified, whose change changes the classification. Chs 14, 26, 33.

Mariner inside the whale. Kipling's figure, from The Compass and the Loom, of the observer inside with stakes who changes the system's constraint by bricolage. Ch 41.

Meshwork. Ingold's tangle of lines of life, with knots, not nodes. Against network and stratigraphy. Ch 30.

Mesh size. The grating's design parameter: too wide and the whale swallows everything, too fine and it starves. Ch 41.

Mētis. Scott's practical, local, experiential knowledge, destroyed by schemes of legibility. Added by Snowden to the definition of a complex system. Chs 3, 24.

Model collapse. The loss of a distribution's tails under recursive training on generated data. Applied to weak-signal instruments. Ch 24.

N–R

Neither (the claim to). The disputation's term for the gesture of positioning anthro-complexity as neither of two errors, read through Wigner as a claim to stand outside. Chs 20, 39.

No-signalling. In physics, the impossibility of using entanglement to send signals. In the Pyxis, harmony is recognised at the rendezvous, not transmitted. Ch 40.

Noticing, not mapping. Tidal Loom principle: measure the latency between something new becoming actual and someone recognising it. Chs 26, 45.

Grant-and-follow exercise.* Scholastic exercise in following what a granted proposition commits one to. Boxes in the text. Preliminaries.

Participants dimension. CST's distinction of unitary, pluralist and coercive relations among participants. Proposed as an addition to Cynefin. Chs 12, 44.

Participant objectivation. Bourdieu's reflexive practice of objectifying the objectifying subject. Ch 33.

Participatory sense-making. De Jaegher and Di Paolo's account of social meaning emerging in interaction. Ch 15.

Performativity. The capacity of descriptions to change what they describe. Austin, Butler, Barnes, Callon, MacKenzie. Ch 34.

Possessive fallacy†. The disputation's Thesis I: placing human distinctiveness in what persons have, not in the relations through which they are formed. Ch 19.

Practitioner inside†. The disputation's figure of the reflexive practitioner, assembled from the fisherman, Mariner, general, calligrapher, cook, wayfarer, navigator, friend and barely known ruler. Ch 44.

The Case for the Other Side.* "For the opposing side": boxes conceding the adversary's case where it holds. The adversary wins half. Preliminaries; Ch 43.

Uncertain Compass. The Compass and the Loom*'s compass that measures its own uncertainty, with four laws (Zeno, complementarity, no-signalling, decoherence) and a training game. Ch 40.

Quorum with inhibition. Tidal Loom principle from the bee swarm: decide by threshold, institutionalise stop signals, commit wholly; seek what people can live with. Chs 9, 46.

Reading of the reader. Tidal Loom principle: convenors signify their own experience, published alongside everyone else's. Chs 20, 45.

Recognition, not alignment. Tidal Loom principle: the test is whether people say we did this ourselves (我自然). Chs 11, 37.

Re-entry. Spencer-Brown's re-entry of a distinction into the space it distinguishes. The human/non-human distinction re-enters as describer/described. Ch 18.

Representation. Hobbes's making of a multitude into one person by one who represents it. The asymmetry principle at the scale of the collective. Ch 42.

Requisite friction†. By analogy with Ashby's requisite variety: a system must contain enough friction to prevent its appetite consuming what it should not. Ch 41.

Return as design requirement. Tidal Loom principle: narrators receive patterns and changed conditions before anything else. Chs 32, 45.

Risk follows anxiety. Tidal Loom principle: track whose anxiety a probe creates and where exposure lands on the power gradient. Chs 16, 23.

S–Z

Scholastic fallacy. Bourdieu's projection of the observer's leisure (skholè) onto the practitioners observed. Ch 33.

Self-signification. SenseMaker's design in which narrators interpret their own stories on signifiers. A real advance, partial in its reach (Thesis 14). Chs 5, 19.

SenseMaker®. The Cynefin Company's narrative-sensing instrument. Chs 5, 6, 45.

Shadow principle. Snowden's behaviour is the shadow of conditions. The disputation adds an ethical supplement: change the conditions with the knowledge and participation of those whose conditions they are. Chs 5, 16.

Simplex. The geometry of the barycentric triad: three values summing to one. Thesis VII. Ch 25.

Single source of truth. The data-management aspiration to one authoritative source, read as Wigner's error institutionalised for narrative data. Ch 39.

Situated knowledge. Haraway's accountable, located knowledge, as against the god trick. Ch 39.

Split causality. Stacey's term for combining a formative teleology for the system and a rationalist teleology for the manager. Chs 11, 20.

Status function. Searle's status assigned by collective acceptance, carrying deontic powers. X counts as Y in C. Ch 13.

Steersman (kybernetes). The Compass and the Loom's figure of the hidden governor who reads the dance and sets the course. Chs 6, 28.

Stratigraphy. Layered representation (the substrate), set against meshwork and constraint closure. Chs 22, 30.

Suspenders. In Kipling and The Compass and the Loom, what ties the grating in place: trust, informal networks, people who notice. Ch 41.

Teleologies (five). Stacey, Griffin and Shaw's secular natural law, rationalist, formative, adaptionist and transformative teleologies. Ch 11.

Thirdness. Peirce's category of mediation, habit and law, irreducible to pairs. Chs 5, 25, 32, 38.

Tidal Loom. The Compass and the Loom's reconstruction: a revisable warp laid by convenors, woven by everyone, with no position outside the cloth. Chs 37, 45.

Triquetra. Snowden's visualisation of the trialectic. Ch 38.

Weak measurement. In physics, a measurement that extracts little information with little disturbance. In the Pyxis, wúwéi: the leader who measures weakly learns without freezing. Chs 37, 40.

Zeno effect (social). By analogy with the quantum Zeno effect: over-frequent monitoring freezes change. Monitor less often than the change you hope to see. Chs 26, 40.

Chinese Terms

道 dào. The way; made by walking (道行之而成). Ch 37.

理 lǐ. Grain, pattern; known by the knife (Cook Ding). Ch 36.

勢 shì. Propensity of a configuration, including the strategist; positional power (Shen Dao, Han Feizi). Ch 35.

地形 dìxíng. Terrain; Sunzi's ground, distinct from 勢. Ch 35.

形 xíng. Disposition, form; the visible arrangement of forces. Ch 35.

虛實 xū shí. Empty and full; the configuration as the enemy reads it. Ch 35.

無為 wúwéi. Action that does not force; the third option between natural and stimulated emergence. Ch 37.

我自然 wǒ zìrán. "We did it ourselves" (Laozi 17). Ch 37.

經營 jīngyíng. Warp and weft; to plan and build; the loom. Ch 37.

三才 sāncái. The three powers: heaven, earth, humanity. The human as mediator. Ch 38.

沖氣 chōngqì. The blending breath that makes harmony between yin and yang (Laozi 42). Ch 38.

和 hé / 同 tóng. Harmony / sameness. The exemplary person harmonises but does not conform. Ch 38.

中 zhōng. The centre; a dynamic focus kept by an agent. Ch 38.

仁 rén. Humaneness; glossed as 人 and 二, the human as between. Chs 19, 38.

知行合一 zhī xíng hé yī. Wang Yangming's unity of knowing and acting. Ch 36.

Addition: Terms Added With the Additiones

Action research. Lewin's practice of research conducted with those studied, in cycles of planning, action, observation and reflection, with the researcher inside. Addition to Ch 3.

Analytical dualism. Archer's insistence that structure and agency are distinct strata that interact over time, against the central conflation of structuration theory. Addition to Ch 2.

Blind spot. In Luhmann, the distinction an observation uses, which it cannot see while using it. Addition to Ch 14.

Emergent personhood. Archer's position, adopted in the amended Principle 1: persons are formed in relation and have irreducible powers once formed. Addition to Ch 2.

Enactment. Weick's term for the way people's actions create the environments they then make sense of. Addition to Ch 13.

Enclosure effect†. A pattern produced by the conditions of observation, as with the alpha wolf observed in captivity. Ask what enclosure you are observing in. Addition to Ch 9.

Field theory. Lewin's account of behaviour as a function of person and environment in a single life space; the forces of a quasi-stationary equilibrium. Addition to Ch 3.

Latency gauge†. A proposed instrument measuring the time between something new becoming actual and its recognition. Ch 45.

Ledger of performatives†. A table of each description's constraint changed, power created, and positions strengthened and weakened. Addition to Ch 20.

Nested bridges†. Hutchins's overlapping horizons applied at every level of a large organisation. Addition to Ch 31.

Relational dashboard†. A dashboard showing patterns for each position, divergences, readings and the instrument's own position. Addition to Ch 39.

Rendezvous instrument†. A proposed occasion for parties who have acted independently to compare and recognise, or fail to recognise, a shared direction. Ch 45.

Second-order observation. Luhmann's observation of observations, which sees another's blind spot but has its own. Addition to Ch 14.

Sensemaking (Weick). Social, retrospective, enactive, ongoing, cue-based, plausibility-driven and grounded in identity construction. Addition to Ch 13.

Wake reader†. A proposed analysis separating stories about an instrument's previous effects from stories about everything else. Ch 45.

Addition: Terms Introduced in Book Nine

Each term below has a card in the Lexicon, where it is explained at length with its links.

Altered compressibility. The author's term for what compression does in human systems: it is performed from inside and changes how regular the situation is. Chapter 56.

Alibi test. Snowden's criterion that a theory is an alibi when it redescribes both success and failure in its own terms. Chapter 52.

Assumed deficit, innate capacity, formative encounter. The three positions of Snowden's trialectic of practice. Chapter 48.

Designer of Conditions and Dweller in the Conversation. The two poles of the dispute between anthro-complexity and the process schools. Chapter 51.

First attribution. Snowden's commitment that the teller says what a story means before anyone else does. Chapter 52.

Form of life. Jaeggi's term for a durable ensemble of practices with its own standards. Chapter 49.

Formation. The re-forming of a practitioner's perception and judgement in company, proposed as the bridge from a room to an organisation. Chapter 51.

Immanent criticism. Criticism that measures a practice by its own constitutive norms and shows them working against themselves. Chapter 49.

Inverted encounter. Thesis IX: the formative encounter as the instruments realise it. Chapter 50.

Joiner. Hockney's collage made while walking, used for any assembly of views from several seats. Chapter 53.

Learning blockage. Jaeggi's mark of a failing practice. Chapter 54.

Matching complexity. Knowing a situation by taking part with enough variety to meet it, as against making a reduced copy. Chapter 56.

Outsight. De Maré's term for a group's shared view of what is forming in its midst. Chapter 57.

Relocated frame. The movement of the expert's reading into the design of signifiers. Chapters 50 and 51.

Threshold object. A framework that is valid where it points beyond itself. Chapter 53.

Triquetra test. Snowden's standard for a real third term. Chapter 48.

Cognisance and answerability. The distinction of Thesis XI between a framer's awareness of what she excludes and an arrangement in which those excluded can answer. Chapter 58.

Provisional imperative. Woermann and Cilliers's command to remain cognisant of other ways of acting. Chapter 58.

Apparatus

Table of Verdicts

1,519 words, about 7 minutes of reading

The Twenty-Four Theses: Verdicts

Verdicts are given as reached in the body of the disputation and as amended in Chapter 43 under the Additive Covenant. Where a verdict was amended, the earlier verdict is kept and the amendment noted.

# Thesis (abridged) Where tested Verdict Note
1 Anthro-complexity is a body of thought with its own ontology Chs 2–6 holds Twenty-two elements in four clusters
2 Complexity is not a kind of systems thinking Ch 10 holds Origins, scope and Pepper's maxims; the mirror noted
3 Human systems differ in kind from other complex systems Chs 9, 13 holds Confirmed by self-categorisation, shared intentionality, social ontology
4 The reason given for the difference is misplaced Chs 8, 13 holds The science grounds each mark relationally
5 Having refused to presume wholes, it presumes parts Ch 19 holds → amended Holds of the self-definition, not of the whole practice (Ch 43, Art. 1)
6 The asymmetry principle applies to the theory that states it Ch 20 holds Applies to this disputation too (Ch 43, Art. 10)
7 Anthro-complexity exempts its own narrative-making (split causality) Chs 11, 20 holds Personal reflexivity of the author acknowledged
8 Its governing practice metaphors are non-reflexive and place the reader outside Ch 21 holds The heuristic's known limits acknowledged (Ch 43, Art. 4)
9 Jullien's 勢 includes the strategist; propensity read from outside is landscape Chs 21, 35 holds 地形 is not 勢
10 Constitutive constraints are reproduced in compliance and in mapping Chs 22, 29, 33 holds Supplied by Searle, Giddens, Bourdieu, Juarrero
11 Constructor theory's import cancels the human distinction Ch 22 holds Everyone is a reagent; use constructor theory for possibility, not actants
12 A theory of human complexity without power is CAS with better adjectives Ch 23 holds → amended Holds of the theory; practice substantially power-aware (Ch 43, Art. 5)
13 Snowden's critique of eurhythmia applies to signifiers, dashboards, residual risk Ch 23 holds Lukes, Foucault, Bourdieu, status characteristics
14 Self-signification is a real but partial advance Chs 19, 23 holds The act distributed; the frame not
15 The ontology is anti-computational and the practice increasingly computational Ch 24 holds Direction of travel, not current practice (Ch 24 PPC)
16 Recursive learning erases the tails where weak signals live Ch 24 holds As mechanism; applies to coupled human-model systems
17 The barycentric triad builds in a zero-sum trade-off and conflates at the centre Ch 25 partial Geometry holds; "more dichotomising" holds only re closure
18 The interpretant sits with analyst and designer, not narrator Chs 25, 32 holds In typical deployment
19 Reflexivity predicts the degradation of stable instruments Chs 26, 40 holds → amended As prediction; testing needs data inside the practice (Ch 43, Art. 10)
20 In human affairs the tide turns partly because someone was seen to wait Chs 3, 21, 27 holds Withholding is a gesture (Mead)
21 The distinction does three jobs, and the first undermines the other two Ch 18 holds The jobs can be made consistent (Ch 18 PPC)
22 The drift from ontology to product is the path along which reflexivity is lost Chs 6, 17, 18 holds The direction of evolution matters (Ch 43, Art. 9)
23 A reflexive anthro-complexity is possible Chs 44–46 partial Proposed; cost and protocol-bureaucracy objections partly win
24 Its test is recognition, not alignment Chs 37, 44 holds Provided the light touch is declared, visible and contestable

The Eight Theses of Book Four

Thesis Chapter Verdict Adversary's half Remedy (Chapter 44 principle)
I. The possessive fallacy 19 holds (amended) Scale requires individual sampling; practice largely relational 1, 2, 5
II. The asymmetry principle applies to itself 20 holds (amended) Full self-accounting impossible; probe-sense-respond is partial reflexivity 2, 4, 7
III. Humans are not estuaries 21 holds The estuary is a known heuristic 4
IV. The under-socialised constraint 22 holds Good facilitation surfaces constitutive constraints 3
V. The missing term is power 23 holds (amended) Practice is power-aware; power as master concept is a danger 7, 8, 11, 12
VI. Anti-computational ontology, computational practice 24 holds Current data practice is careful; not surveillance capitalism 8, 9, 12
VII. The triad is a simplex 25 partial Other signifiers compensate; visual exploration mitigates 5
VIII. Reflexivity predicts degradation 26 holds (amended) Defences against gaming are real 6, 9, 10

Verdicts on the Neighbours and Sources

Claim Chapter Verdict
The genealogical thesis: theory and instruments belong to different moments 6 partial (intent vs structure)
CAS: the difference of kind is relational, not agent-located 7, 8 partial (CAS knows much of it already)
Cynefin's placement of systems thinking mirrors Ing's taxonomic move 10 partial (Cynefin gives a domain of validity)
Anthro-complexity's sources are eclectic in Pepper's sense 10 partial (integrated by ontology; modifications where the human is lodged)
CRP supplies the relational ground anthro-complexity needs 11 partial
Boundary critique supplies the power vocabulary 12 partial (meta-methodology risks the connoisseur)
The scientific reason 13 holds, relationally grounded
The reflexive reason 14 holds; the deepest
The anti-computational reason, stated with care 15 holds
The ethical reason; the shadow principle needs a supplement 16 partial
The positional reason, as explanation 17 partial
Mead supplies the social psychology of the person 27 partial
Elias supplies power, history and the observer 28 partial
Juarrero is the bridge 29 partial (a reading she may reject)
Ingold supplies design principles, not instruments 30 partial
Hutchins: overlapping horizons are achievable at scale 31 partial
Peirce taken whole is anti-possessive 32 partial
Bourdieu, Giddens, Hacking can be approximated in practice 33 partial
Instruments are engines as well as cameras 34 partial (accuracy remains a question)
勢 taken whole includes strategist and position 35 partial
理 is lost in the instruments 36 partial
Wúwéi and wǒ zìrán need the Western critique of power 37 partial
The Chinese sources handle mediation; the human is the mediator 38 partial
Observer-relativity crosses from physics; superposition does not 39 partial
The physics of asking organises the social findings 40 partial
The Mariner's figure; its limits 41 partial
The dashboard stands at the threshold of representation 42 partial
The receiver's case: the reflexive reading's core 46 partial (the composite flatters it)

The Adversary's Twelve Articles

Article Verdict for the adversary
1. The theory is relational already partial
2. The grammar is shorthand partial
3. The instruments were designed against the outside position partial
4. The estuary is a heuristic partial
5. Power is everywhere in the practice partial
6. Scale requires the individual holds
7. Your alternative is CRP fails
8. Reflexivity is a regress partial
9. The author is reflexive, and the instruments evolve partial
10. You are standing outside holds
11. Instruments must be sold partial
12. The human distinction does its job fails

A Note on the Pattern of Verdicts

The pattern is worth stating. The disputation's negative theses, those that criticise, mostly hold. Its positive claims, those that propose sources and remedies, mostly hold in part. This is what a disputation should expect. It is easier to show that a practice's premises lead further than it goes than to show exactly how far they lead and what the practice would look like there. The criticisms are the firmer half of the book, and the reconstruction the more tentative. Both are offered under the Additive Covenant, to be amended as the inside view arrives.

Addition: Verdicts of Book Nine

The table above stands as written. These rulings are added.

Matter Chapter Ruling
Thesis IX: the formative encounter is realised in inverted form 50 holds for the mapping stage, contested for intervention
The formative encounter passes the triquetra test 50 doubtful: the mediating figure is constant
The alibi test applies to anthro-complexity's own vocabulary 52 holds until instruments carry signs of failure
The originator makes the first attribution 52 first of the story, fifth in the process
A persistent record makes re-attribution open to challenge 52 holds in principle, needs the return path
Ethics can rest on method without formation 52 fails: Snowden's trialectic says otherwise
Cynefin lacks an axis of agreement 53 holds, with a residue at the centre
Anthro-complexity is a practice that learns 54 holds, with four possible blockages
Methodological abstinence can be kept 54 fails: someone decides at four points
Thesis X: incompressibility is a concept of the outside observer 56 holds
Thesis X refutes Cilliers's ethics of framing 56 fails: the ethics is strengthened (amended in Chapter 58)
There are no boundaries 56 holds as ontology; boundary critique keeps its use
Thesis XI: the ethics of framing is an ethics of the framer's conscience 58 holds
The provisional imperative as an ethics 58 fails: it commands an attitude and has no condition of failure
The four operations 58 hold as virtues, fail as a substitute for the voice of the excluded
The responsibility of whoever frames or compresses 56, 58 holds, heavier than Cilliers allows
Flinn's practice escapes the questions put to Snowden's 57 fails: the same questions apply
The fifteen dialectics 55, 56, 58 three resolved or largely so, five resolvable, seven open or owed

The revised verdicts on Theses I to VIII are in the table of Chapter 55.

Apparatus

Bibliography

2,696 words, about 12 minutes of reading

Primary Texts Under Disputation

Snowden, Dave. "Not a branch of the tree." Blog post, anthro-complexity series, 13 November 2026.

Downham, Mark. The Compass and the Loom. Principia Humanitatis, London, November 2026. markdownham88-crypto.github.io/the-compass-and-the-loom.

Downham, Mark. To the Strait-Jacketed, the Other Always Seems Baggy. Principia Humanitatis, London, November 2026. markdownham88-crypto.github.io/the-straitjacketed-and-the-baggy.

Works Discussed in Snowden's Post

Cummings, Stephen, Todd Bridgman and Kenneth G. Brown. "Unfreezing change as three steps: Rethinking Kurt Lewin's legacy for change management." Human Relations 69, no. 1 (2016).

Ing, David. Knowing Better via Systems Thinking. Lecture, Barcelona, 2022.

Ing, David, and Gary Metcalf. "Reframing Systems Thinking for Systems Changes: Sciencing and Philosophizing from Pragmatism towards Processes as Rhythms." Journal of the International Society for the Systems Sciences, 2023.

Jullien, François. The Propensity of Things: Toward a History of Efficacy in China. Trans. Janet Lloyd. New York: Zone Books, 1995.

Jullien, François. A Treatise on Efficacy: Between Western and Chinese Thinking. Trans. Janet Lloyd. Honolulu: University of Hawai'i Press, 2004.

Lee, Keekok. Works on the philosophical foundations of Classical Chinese Medicine, as drawn on by Ing and Metcalf (see their 2023 paper for references).

Ramage, Magnus, and Karen Shipp. Systems Thinkers. London: Springer, 2009; 2nd edn 2020.

Rosen, Robert. Life Itself. New York: Columbia University Press, 1991.

Scott, James C. Seeing Like a State. New Haven: Yale University Press, 1998.

Cynefin and Anthro-Complexity

Juarrero, Alicia. Dynamics in Action: Intentional Behavior as a Complex System. Cambridge, MA: MIT Press, 1999.

Juarrero, Alicia. Context Changes Everything: How Constraints Create Coherence. Cambridge, MA: MIT Press, 2023.

Kurtz, Cynthia F., and David J. Snowden. "The new dynamics of strategy: Sense-making in a complex and complicated world." IBM Systems Journal 42, no. 3 (2003).

Snowden, David J., and Mary E. Boone. "A Leader's Framework for Decision Making." Harvard Business Review, November 2007.

Complex Adaptive Systems and Complexity Science

Arthur, W. Brian. "Inductive Reasoning and Bounded Rationality." American Economic Review 84, no. 2 (1994).

Ballerini, Michele, and others "Interaction ruling animal collective behavior depends on topological rather than metric distance." PNAS 105, no. 4 (2008).

Cavagna, Andrea, and others "Scale-free correlations in starling flocks." PNAS 107, no. 26 (2010).

Cilliers, Paul. Complexity and Postmodernism. London: Routledge, 1998.

Epstein, Joshua M., and Robert Axtell. Growing Artificial Societies. Cambridge, MA: MIT Press, 1996.

Gell-Mann, Murray. The Quark and the Jaguar. New York: W. H. Freeman, 1994.

Gordon, Deborah M. Ant Encounters: Interaction Networks and Colony Behavior. Princeton: Princeton University Press, 2010.

Holland, John H. Hidden Order: How Adaptation Builds Complexity. Reading, MA: Addison-Wesley, 1995.

Kauffman, Stuart. The Origins of Order. New York: Oxford University Press, 1993.

Kauffman, Stuart. At Home in the Universe. New York: Oxford University Press, 1995.

Montévil, Maël, and Matteo Mossio. "Biological organisation as closure of constraints." Journal of Theoretical Biology 372 (2015).

Plsek, Paul E., and Trisha Greenhalgh. "The challenge of complexity in health care." BMJ 323 (2001).

Reynolds, Craig W. "Flocks, herds and schools: A distributed behavioral model." Computer Graphics 21, no. 4 (1987).

Schelling, Thomas C. "Dynamic models of segregation." Journal of Mathematical Sociology 1 (1971).

Seeley, Thomas D. Honeybee Democracy. Princeton: Princeton University Press, 2010.

Seeley, Thomas D., and others "Stop signals provide cross inhibition in collective decision-making by honeybee swarms." Science 335 (2012).

Waldrop, M. Mitchell. Complexity. New York: Simon & Schuster, 1992.

Weaver, Warren. "Science and Complexity." American Scientist 36 (1948).

Systems Thinking and Critical Systems Thinking

Ackoff, Russell L. Redesigning the Future. New York: Wiley, 1974.

Churchman, C. West. The Design of Inquiring Systems. New York: Basic Books, 1971.

Emery, F. E., and E. L. Trist. "The causal texture of organizational environments." Human Relations 18 (1965).

Flood, Robert L., and Michael C. Jackson. Creative Problem Solving: Total Systems Intervention. Chichester: Wiley, 1991.

Jackson, Michael C. Critical Systems Thinking and the Management of Complexity. Chichester: Wiley, 2019.

Jackson, Michael C., and Paul Keys. "Towards a system of systems methodologies." Journal of the Operational Research Society 35 (1984).

Midgley, Gerald. Systemic Intervention. New York: Kluwer/Plenum, 2000.

Pepper, Stephen C. World Hypotheses. Berkeley: University of California Press, 1942.

Ulrich, Werner. Critical Heuristics of Social Planning. Bern: Haupt, 1983.

Complex Responsive Processes

Flinn, Kevin. Leadership Development: A Complexity Approach. London: Routledge, 2018.

Griffin, Douglas. The Emergence of Leadership. London: Routledge, 2002.

Mowles, Chris. Rethinking Management. Farnham: Gower, 2011.

Mowles, Chris. Managing in Uncertainty. London: Routledge, 2015.

Shaw, Patricia. Changing Conversations in Organizations. London: Routledge, 2002.

Stacey, Ralph D. Complexity and Creativity in Organizations. San Francisco: Berrett-Koehler, 1996.

Stacey, Ralph D. Complex Responsive Processes in Organizations. London: Routledge, 2001.

Stacey, Ralph D., Douglas Griffin and Patricia Shaw. Complexity and Management: Fad or Radical Challenge to Systems Thinking? London: Routledge, 2000.

Social Theory

Abbott, Andrew. The System of Professions. Chicago: University of Chicago Press, 1988.

Abrahamson, Eric. "Management Fashion." Academy of Management Review 21 (1996).

Bachrach, Peter, and Morton S. Baratz. "Two Faces of Power." American Political Science Review 56 (1962).

Berger, Joseph, Bernard P. Cohen and Morris Zelditch Jr. "Status Characteristics and Social Interaction." American Sociological Review 37 (1972).

Bourdieu, Pierre. The Logic of Practice. Cambridge: Polity, 1990 [1980].

Bourdieu, Pierre. Language and Symbolic Power. Cambridge: Polity, 1991.

Bourdieu, Pierre. Pascalian Meditations. Cambridge: Polity, 2000 [1997].

Bourdieu, Pierre, and Loïc Wacquant. An Invitation to Reflexive Sociology. Chicago: University of Chicago Press, 1992.

Canetti, Elias. Crowds and Power. Trans. Carol Stewart. London: Gollancz, 1962 [1960].

Clastres, Pierre. Society Against the State. New York: Zone Books, 1987 [1974].

Deleuze, Gilles, and Félix Guattari. A Thousand Plateaus. Minneapolis: University of Minnesota Press, 1987 [1980].

Douglas, Mary, and Aaron Wildavsky. Risk and Culture. Berkeley: University of California Press, 1982.

Elias, Norbert. The Civilizing Process. Oxford: Blackwell, 2000 [1939].

Elias, Norbert. What Is Sociology? London: Hutchinson, 1978 [1970].

Elias, Norbert. Involvement and Detachment. Oxford: Blackwell, 1987.

Elias, Norbert. The Society of Individuals. Oxford: Blackwell, 1991 [1987].

Elias, Norbert, and John L. Scotson. The Established and the Outsiders. London: Frank Cass, 1965.

Espeland, Wendy Nelson, and Michael Sauder. "Rankings and Reactivity." American Journal of Sociology 113 (2007).

Espeland, Wendy Nelson, and Michael Sauder. Engines of Anxiety. New York: Russell Sage, 2016.

Foucault, Michel. Discipline and Punish. New York: Pantheon, 1977 [1975].

Freeman, Jo. "The Tyranny of Structurelessness." Berkeley Journal of Sociology 17 (1972–73).

Giddens, Anthony. New Rules of Sociological Method. London: Hutchinson, 1976.

Giddens, Anthony. The Constitution of Society. Cambridge: Polity, 1984.

Giddens, Anthony. The Consequences of Modernity. Cambridge: Polity, 1990.

Gieryn, Thomas F. "Boundary-Work and the Demarcation of Science from Non-Science." American Sociological Review 48 (1983).

Hacking, Ian. "Making Up People." In Reconstructing Individualism, ed. T. Heller and others Stanford: Stanford University Press, 1986.

Hacking, Ian. "The looping effects of human kinds." In Causal Cognition, ed. D. Sperber and others Oxford: Clarendon, 1995.

Hacking, Ian. The Social Construction of What? Cambridge, MA: Harvard University Press, 1999.

Hirschman, Albert O. Exit, Voice, and Loyalty. Cambridge, MA: Harvard University Press, 1970.

Hobbes, Thomas. Leviathan. London, 1651.

Jacobs, Jane. The Death and Life of Great American Cities. New York: Random House, 1961.

Lukes, Steven. Power: A Radical View. London: Macmillan, 1974; 2nd edn 2005.

Macpherson, C. B. The Political Theory of Possessive Individualism. Oxford: Clarendon, 1962.

Menzies Lyth, Isabel. "A Case-Study in the Functioning of Social Systems as a Defence against Anxiety." Human Relations 13 (1960).

Merton, Robert K. "The Self-Fulfilling Prophecy." Antioch Review 8 (1948).

Simmel, Georg. Soziologie. Leipzig: Duncker & Humblot, 1908.

Turner, Victor. The Ritual Process. Chicago: Aldine, 1969.

Zuboff, Shoshana. The Age of Surveillance Capitalism. London: Profile, 2019.

Pragmatism, Mind and Cognition

Bainbridge, Lisanne. "Ironies of Automation." Automatica 19 (1983).

Clark, Andy. Surfing Uncertainty. New York: Oxford University Press, 2016.

Clark, Andy, and David Chalmers. "The Extended Mind." Analysis 58 (1998).

De Jaegher, Hanne, and Ezequiel Di Paolo. "Participatory sense-making." Phenomenology and the Cognitive Sciences 6 (2007).

Dreyfus, Hubert L. What Computers Can't Do. New York: Harper & Row, 1972.

Dreyfus, Hubert L., and Stuart E. Dreyfus. Mind over Machine. New York: Free Press, 1986.

Hutchins, Edwin. Cognition in the Wild. Cambridge, MA: MIT Press, 1995.

Hutchins, Edwin. "How a Cockpit Remembers Its Speeds." Cognitive Science 19 (1995).

Lave, Jean, and Etienne Wenger. Situated Learning. Cambridge: Cambridge University Press, 1991.

Mead, George Herbert. The Philosophy of the Present. Chicago: Open Court, 1932.

Mead, George Herbert. Mind, Self, and Society. Chicago: University of Chicago Press, 1934.

Oyama, Susan. The Ontogeny of Information. Cambridge: Cambridge University Press, 1985.

Peirce, Charles S. "Some Consequences of Four Incapacities." Journal of Speculative Philosophy 2 (1868).

Peirce, Charles S. "The Fixation of Belief." Popular Science Monthly 12 (1877).

Peirce, Charles S. "How to Make Our Ideas Clear." Popular Science Monthly 12 (1878).

Peirce, Charles S. "Evolutionary Love." The Monist 3 (1893).

Peirce, Charles S. "What Pragmatism Is." The Monist 15 (1905).

Polanyi, Michael. The Tacit Dimension. London: Routledge, 1966.

Searle, John R. "Minds, Brains, and Programs." Behavioral and Brain Sciences 3 (1980).

Searle, John R. Speech Acts. Cambridge: Cambridge University Press, 1969.

Searle, John R. The Construction of Social Reality. New York: Free Press, 1995.

Searle, John R. Making the Social World. Oxford: Oxford University Press, 2010.

Tajfel, Henri, and John Turner. "An Integrative Theory of Intergroup Conflict." In The Social Psychology of Intergroup Relations, ed. W. Austin and S. Worchel, 1979.

Tomasello, Michael. Origins of Human Communication. Cambridge, MA: MIT Press, 2008.

Tomasello, Michael. A Natural History of Human Thinking. Cambridge, MA: Harvard University Press, 2014.

Turner, John C., and others Rediscovering the Social Group: A Self-Categorization Theory. Oxford: Blackwell, 1987.

Varela, Francisco J., Evan Thompson and Eleanor Rosch. The Embodied Mind. Cambridge, MA: MIT Press, 1991.

Language, Performativity, Epistemology

Austin, J. L. How to Do Things with Words. Oxford: Clarendon, 1962.

Bakhtin, M. M. Speech Genres and Other Late Essays. Austin: University of Texas Press, 1986.

Barnes, Barry. "Social life as bootstrapped induction." Sociology 17 (1983).

Bruner, Jerome. Acts of Meaning. Cambridge, MA: Harvard University Press, 1990.

Butler, Judith. Gender Trouble. New York: Routledge, 1990.

Callon, Michel, ed. The Laws of the Markets. Oxford: Blackwell, 1998.

Haraway, Donna. "Situated Knowledges." Feminist Studies 14 (1988).

Haraway, Donna. Staying with the Trouble. Durham: Duke University Press, 2016.

MacKenzie, Donald. An Engine, Not a Camera. Cambridge, MA: MIT Press, 2006.

Nagel, Thomas. The View from Nowhere. New York: Oxford University Press, 1986.

Spencer-Brown, G. Laws of Form. London: Allen & Unwin, 1969.

Ecology, Anthropology, Lines

Gibson, James J. The Ecological Approach to Visual Perception. Boston: Houghton Mifflin, 1979.

Ingold, Tim. The Perception of the Environment. London: Routledge, 2000.

Ingold, Tim. Lines: A Brief History. London: Routledge, 2007.

Ingold, Tim. Being Alive. London: Routledge, 2011.

Ingold, Tim. Making. London: Routledge, 2013.

Ingold, Tim. Correspondences. Cambridge: Polity, 2020.

Lévi-Strauss, Claude. The Savage Mind. Chicago: University of Chicago Press, 1966 [1962].

Gould, Stephen Jay, and Elisabeth S. Vrba. "Exaptation — a missing term in the science of form." Paleobiology 8 (1982).

Physics and Measurement

Aharonov, Yakir, David Z. Albert and Lev Vaidman. "How the result of a measurement of a component of the spin of a spin-1/2 particle can turn out to be 100." Physical Review Letters 60 (1988).

Bong, Kok-Wei, and others "A strong no-go theorem on the Wigner's friend paradox." Nature Physics 16 (2020).

Busemeyer, Jerome R., and Peter D. Bruza. Quantum Models of Cognition and Decision. Cambridge: Cambridge University Press, 2012.

Frauchiger, Daniela, and Renato Renner. "Quantum theory cannot consistently describe the use of itself." Nature Communications 9 (2018).

Fuchs, Christopher A., N. David Mermin and Rüdiger Schack. "An introduction to QBism." American Journal of Physics 82 (2014).

Itano, Wayne M., and others "Quantum Zeno effect." Physical Review A 41 (1990).

Misra, B., and E. C. G. Sudarshan. "The Zeno's paradox in quantum theory." Journal of Mathematical Physics 18 (1977).

Rovelli, Carlo. "Relational quantum mechanics." International Journal of Theoretical Physics 35 (1996).

Shumailov, Ilia, and others "AI models collapse when trained on recursively generated data." Nature 631 (2024).

Wang, Zheng, Tyler Solloway, Richard M. Shiffrin and Jerome R. Busemeyer. "Context effects produced by question orders reveal quantum nature of human judgments." PNAS 111 (2014).

Wigner, Eugene P. "Remarks on the Mind-Body Question." In The Scientist Speculates, ed. I. J. Good. London: Heinemann, 1961.

Zurek, Wojciech H. "Decoherence, einselection, and the quantum origins of the classical." Reviews of Modern Physics 75 (2003).

Classical Chinese Sources

Sunzi bingfa (孫子兵法). Trans. Lionel Giles, The Art of War, London, 1910.

Zhuangzi (莊子), chapters 2 (齊物論) and 3 (養生主).

Laozi / Daodejing (道德經), chapters 17 and 42.

Lunyu / Analects (論語), 13.23.

Mengzi / Mencius (孟子), 1A2.

Zuozhuan (左傳), Duke Zhao 20 (Yan Ying on harmony).

Zhongyong (中庸).

Yijing, Xici zhuan (繫辭傳).

Han Feizi (韓非子), chapter on 勢 (Shen Dao).

Ames, Roger T. Confucian Role Ethics. Honolulu: University of Hawai'i Press, 2011.

Ames, Roger T., and David L. Hall. Focusing the Familiar. Honolulu: University of Hawai'i Press, 2001.

Fingarette, Herbert. Confucius: The Secular as Sacred. New York: Harper & Row, 1972.

Law, John, and Wen-yuan Lin. Papers on a shi-inflected and postcolonial social science, 2010s.

Slingerland, Edward. Effortless Action. New York: Oxford University Press, 2003.

Slingerland, Edward. Trying Not to Try. New York: Crown, 2014.

Literature

Kipling, Rudyard. "How the Whale Got His Throat." In Just So Stories. London: Macmillan, 1902.

Machado, Antonio. Campos de Castilla. Madrid, 1912.

Poe, Edgar Allan. "A Descent into the Maelström." Graham's Magazine, 1841.

A Note on the Apparatus

The bibliography lists works discussed or drawn on. Page references are not given where the author has not checked them. Two entries are given in general form because the author's knowledge of their exact bibliographic details is incomplete: Keekok Lee's work on Classical Chinese Medicine as drawn on by Ing and Metcalf, and Law and Lin's papers on a shi-inflected social science. Readers are referred to Ing and Metcalf's 2023 paper for the former and to Law and Lin's published papers for the latter. Corrections are welcome under the Additive Covenant.

Addition: Sources Met at Second Hand in the Fuller Readings

The additions to Chapters 49, 51 and 57 name several writers whom this book knows only through the three sources it was reading. They are listed here so that no reader takes them for works consulted directly.

  • Alasdair MacIntyre, on traditions and epistemological crises: as discussed by Jaeggi in the second half of Critique of Forms of Life.
  • Karl Marx, the remark on hunger: as quoted by Jaeggi.
  • Murray Gell-Mann, the saying about the only valid model of a human system: as attributed to him in Snowden's series, on the report of A Web of Relations. No original source has been checked.
  • Karina Solsø, the dimensions of sensemaking: as presented in A Web of Relations.
  • Farhad Dalal, on the shadow system and on who is structurally silenced: as presented in A Web of Relations.
  • James Traeger, Rebecca Nestor and Helen Charlwood: contributors to Flinn's Organisational Development in Practice, cited from that book.
  • Jeffrey Pfeffer, the rules of power: as discussed by Flinn.
  • Hans-Georg Gadamer, the fusion of horizons: as used by Charlwood in Flinn's book.
  • Dave Snowden, the posts of the 2026 series other than the four read in full: known here only through the account given in A Web of Relations, and marked so wherever used.

Addition: Sources for Book Nine

  • Flinn, Kevin. Organisational Development in Practice: A Complexity Approach. Routledge.
  • Flinn, Kevin, edited by Mark Downham. A Web of Relations. Web edition, Principia Humanitatis.
  • Downham, Mark. The Stacey Matrix and Walking in the Zen Garden at the Ryōan-ji Temple, Kyoto: On the Politics of Perspective. Web edition, Principia Humanitatis.
  • Fricker, Miranda. Epistemic Injustice: Power and the Ethics of Knowing. Oxford University Press, 2007.
  • Jaeggi, Rahel. Critique of Forms of Life. Translated by Ciaran Cronin. Belknap Press of Harvard University Press, 2018.
  • Snowden, Dave. "Knotty problems". The Cynefin Company, 8 January 2024.
  • Snowden, Dave. "A Trialectic for organisational practice & coaching". The Cynefin Company, 31 May 2026.
  • Snowden, Dave. "Irritation is not an alibi". The Cynefin Company, 14 November 2026.
  • Wittgenstein, Ludwig. Tractatus Logico-Philosophicus. 1921.
  • de Maré, Patrick, Robin Piper and Sheila Thompson. Koinonia: From Hate, through Dialogue, to Culture in the Larger Group. 1991.
  • Stern, Daniel N. The Present Moment in Psychotherapy and Everyday Life. Norton, 2004.
  • Woermann, Minka, and Paul Cilliers. "The ethics of complexity and the complexity of ethics". South African Journal of Philosophy 31(2), 2012.
  • Morin, Edgar. "Restricted complexity, general complexity". 2007.
Apparatus

Lexicon of Explainers

120 words, about 1 minutes of reading

How to use the Lexicon

The Lexicon holds every explainer in the book as a turnable card: 440 cards, about 108,297 words. The front of each card names the term and gives a one-line hook. The back gives the full explainer: what or who the term names, where it comes from, how the disputation uses it and where the argument turns on it. The same explainers open when a blue term is hovered or tapped in the text, and each chapter ends with a deck of the cards for the terms it uses.

Cards are grouped by kind. Select a card to turn it over, and select it again to turn it back. On the back, longer explainers scroll.

Humans Are Not Estuaries. Principia Humanitatis. 253,279 words. Composed under the Additive Covenant: nothing deleted, only inserted. Built 14 November 2026.