Argumentum

The Argument in Brief

This disputation examines two recent instruments from the Cynefin Company. They were released almost together and belong to the same programme: an "anthro-complexity" alternative to strategy. The first is SwarmCompass, a two-page brochure from 2025 powered by SenseMaker®. It sells "human-AI swarm intelligence" built on three images: bees choosing a hive, radiologists missing a gorilla on a CT scan, and Lincoln telling Congress to "think anew and act anew." The second is Weaving Risk and Strategy (WRAS, renamed WRASSE after the cleaner wrasse), introduced on the Cynefin Co blog on 23 September 2025. It is a two-by-two matrix that crosses natural versus stimulated emergence with fine versus coarse granularity. The resulting four domains are Directed, Opportunist, Hypothesis-based and Distributed, and each is placed on a "substrate" made of technological, cultural, procedural and network layers.

I take both instruments seriously. That rules out polemic, and it also rules out applause. Snowden has been one of the most consequential and combative voices in organisational complexity for a quarter-century. Many of his provocations are still correct: against best practice everywhere, against the tyranny of the average, against confusing a recipe with a context. The argument here is not that the instruments are worthless. The argument is that both instruments contradict, in their working assumptions, the complexity theory they claim as their warrant. Each one also falls back on a figure that complexity thinking has spent forty years trying to retire: the manager who stands outside the system, reads it, and then chooses whether to let emergence happen or to stimulate it.

The critique runs at three levels. The evidential level asks whether the natural-scientific stories the brochure tells are true. Two of them are not. The gorilla study contains no episode in which the people who saw it were talked out of it. The bee swarm does not "always" choose the best site, does not seek "true consensus", and does in fact stake everything on one option. The theoretical level brings the whole complexity corpus to bear: Stacey, Griffin and Shaw's five teleologies; Mead and Elias on gesture, figuration and power; Juarrero on constraint; Kauffman on the adjacent possible; Cilliers on incompressibility; Boisot on codification; Ashby on variety; Prigogine on bifurcation; Ingold on the meshwork; Hutchins on distributed cognition; Peirce on abduction; Boyd on orientation; Mowles on tools as ritual; Flinn on recognition. The political-economic level asks what it means that a theory of distributed sensemaking should come to market as a trademarked, patented, members-only instrument whose deepest promise is authentic alignment.

The critique ends with a reconstruction. If an instrument were faithful to the complexity it invokes, what would it have to give up, and what would it have to admit? Part V proposes a Tidal Loom. It is an alternative that keeps what is valuable in both instruments: distributed narrative sensing, the attention to what is difficult to change, and the preference for many small acts over one large bet. It removes the observer's privilege, the promise of mapped unknowns, and the fantasy of alignment.

A compass assumes a north. A loom assumes a frame. Complexity offers neither, only the next move made among others who are also moving.

Theses for Disputation

Twelve Theses

Twelve theses from the first edition, with three added in the second (XIII–XV), one amendment to Thesis VII, and eight more in the fourth (XVI–XXIII). Under the Additive Covenant nothing has been withdrawn.

1. The gorilla is a fable. The brochure's founding anecdote adds a social-conformity episode that the 2013 radiology study does not contain. The study's real finding is that blindness comes from the task frame and not from the person, and that finding undercuts the instruction to "find the 17%" as individuals.
2. The swarm is mis-sold. Seeley's honeybees decide by quorum, not consensus. They use cross-inhibiting stop signals to avoid premature alignment. They share a common genetic interest that no human organisation shares. And they commit the whole swarm to a single site. Each property contradicts a brochure claim.
3. The wrasse bites back. The cleaner wrasse, the WRASSE emblem, is the best-studied case of an animal that cheats when unobserved, and it has a famous mimic that looks like a cleaner and eats its clients. The metaphor contains its own critique.
4. Both instruments keep dual causality. In Stacey's terms, the system is given a formative teleology (it unfolds what is enfolded) while the leader or consultant keeps a rationalist one (choose, deploy, steer). Transformative teleology, in which the future is under perpetual construction and nobody stands outside it, is missing.
5. "Natural versus stimulated emergence" is the observer in disguise. The WRAS vertical axis assumes someone who can decide to leave a social process alone. In Mead's account every withholding is already a gesture that calls out a response.
6. The substrate is a stratigraphy and not a meshwork. Placing "culture" as a layer beneath strategy reifies what Elias, Griffin and Mowles show to be the patterning of interaction itself. People become a growth medium.
7. Unknown unknowns cannot become mapped territory. Kauffman's adjacent possible cannot be stated in advance, and Cilliers's complex systems cannot be compressed. The brochure's central promise is a category error presented as a feature. Amended in the second edition, after the adversary's brief (Movement 11, Pro Parte Contraria): the form of the unknown (its rate, its missing mass, its shape of growth, its error distribution) can be stated in advance, and complex systems compress in their predictive structure. What cannot be stated is the identity, function and meaning of what comes. The category error lies in promising the inventory where only the form is available.
8. SenseMaker codifies and abstracts, so it moves down the I-Space. In Boisot's terms, signifier triads turn tacit, situated meaning into codified, diffusible data. That data then becomes, explicitly, training data sets for further AI-based automation. The swarm is harvested.
9. Risk is assigned to quadrants, though it lives in relations. Giving each WRAS domain "its" risk strategy treats risk as a property of a situation rather than an anxiety lived and negotiated among people, as in Shaw and Stacey's Experiencing Risk and Menzies Lyth's social defences.
10. The novelty is genealogically thin. The four WRAS domains revisit Mintzberg and Waters (1985), March (1991), McGrath's discovery-driven planning and real-options theory without acknowledging them. "The last link in the chain" names a chain that others forged.
11. Alignment is not leadership. By Flinn's recognition test, "authentic alignment (not just top-down compliance)" describes compliance more honestly than the brochure intends. Recognition, not alignment, separates followership from coercion.
12. The Way is not a compass bearing. In 道理勢 terms, both instruments seize 勢 shì (propensity) and lose 道 dào, the way that is made only by walking it together. Cook Ding does not weave the ox. He finds the spaces it already has.
13. The adversary wins half. (Editio secunda.) Tria, Loreto, Servedio and Strogatz predicted the statistics of novelty; Cortês, Kauffman, Liddle and Smolin predicted the timing of its explosion; Turing measured the mass of the unseen; Machta, Transtrum and Sethna showed why complex systems compress. Kauffman and Cilliers survive only for identity, function and meaning, which is where they always mattered.
14. The watched needle never turns. (Editio secunda.) By the quantum Zeno effect, frequent measurement freezes evolution. An instrument of constant micro-pulses risks suppressing the emergence it claims to detect. Measure weakly.
15. Harmony is recognised at the rendezvous, not broadcast. (Editio secunda.) By the no-communication theorem, entangled compasses share correlation without signal. Alignment cannot be transmitted to people. It can only be discovered when they meet and compare what each of them lived through.
16. The answerer is rotated by the question. (Editio quarta.) Question-order effects obey the QQ equality that quantum probability predicts. An assessment that opens with "blind spots" changes the strategy answers that follow, and rotating question order hides the effect from the analyst without undoing it for the respondent.
17. Every measure of people is a gesture to them. (Editio quarta.) Goodhart, Campbell, Lucas and Soros describe one loop: measure, target, respond, drift. Any instrument that publishes its findings enters that loop, and its period of good fit is the time for most vigilance.
18. Acumen cannot be installed, only exercised. (Editio quarta.) Snowden's salience, sapience and sentience are a conjunction, which a barycentric triad cannot represent. Read through Boyd, acumen is orientation that can destroy and rebuild itself; read through Deleuze, collective acumen is a refrain. Naming its observables invites Goodhart, and "creating the substrate" risks striating the smooth space it means to nurture.
19. The Mariner is the observer inside. (Editio quarta.) Kipling's Mariner works from inside the whale, with what is to hand, on a constraint rather than an outcome, and does not forget the suspenders. Design of constraints by insiders with stakes is legitimate; design of substrates by outsiders with products is what Movement 10 opposed.
20. Leverage is a conjunction. (Editio quarta.) Agency, adjacency and tendency give leverage only together, and all three work by decomposition and recombination. A flash mob and a murmuration both reach scale with no algorithm in sight, by opposite routes.
21. Leviathan is a pattern of motion. (Editio quarta.) Read as physics, Hobbes's commonwealth is a resultant of endeavours, a store of conferred power and a circulation that dies when it stops. A crowd becomes a Leviathan at the moment someone speaks for it; every convened crowd needs a dispersal rule.
22. Three books, one confession. (Editio quarta.) Book I, Liber Secundus and Liber Tertius state at three scales that no one sees the form they stand in. Their answers converge: measure lightly, declare the form, offer rather than prescribe.
23. A diagram is an instrument. (Editio quarta.) Every figure decides what stands out before the reader asks. An honest diagram marks the difference between the author's drawn assertions and the eye's inferred groupings, lets decoration be switched off, and sends the reader back into the argument rather than up to a centre of calculation.

Index Capitulorum

Contents

De Methodo

On Sources and Method

For SwarmCompass, the primary text is the two-page Interactive Brochure (Cognitive Edge Pte Ltd, ©2025; generated by InDesign on 5 August 2025). I read it in full and quote its own words where they carry weight. Its phrases are short and declarative, which makes them unusually easy to test.

WRAS has a looser documentary base, and that is partly what the critique is about. The framework was introduced in Snowden's blog post Weaving risk and strategy … (The Cynefin Co, 23 September 2025). It was followed by Wayshaping for substrate management (28 September 2025) and anticipated by The substrate of safety (3 September 2025). The Cynefin.io wiki page on risk management records the framework and carries the notice that it is still under development, do not use or share without permission. A conference abstract (devopsdays Atlanta, 2026) repeats the four substrates and the wrasse etymology. I worked from the published summary of the post, the wiki entry and these companion texts. Where the published material leaves the quadrant arrangement ambiguous, I say so and mark my reconstruction as a reconstruction.

The corpus is the body of complexity literature that has informed the Principia Humanitatis treatises. It includes the complex responsive processes school (Stacey, Griffin, Shaw, Mowles, Flinn) and its sources in Mead and Elias. It includes the complexity sciences (Prigogine, Kauffman, Holland, Bak, Cilliers, Juarrero, Boisot, Ashby), the processual turn (Chia and Holt, Ingold, Deleuze), the pragmatist and enactive traditions (Peirce, Schön, Weick, Hutchins), military theory (Boyd, Clausewitz, Sun Tzu) and the classical Chinese spine 道理勢. Attribution is kept exact. The 2003 IBM Systems Journal paper is Kurtz and Snowden's, but the Cynefin framework is Snowden's creation, and nothing here suggests otherwise.

Movement 1

SwarmCompass: An Anatomy

The brochure has two pages, and its argument is a funnel. Three warrant-images (the bee swarm, the gorilla on the X-ray, Lincoln's annual message) set up a diagnosis: traditional leadership tools leave you adrift in a sea of data without direction. The cure is a fusion. AI sees the map. Humans feel the terrain. The cure is then sold through three movements, Chart the Territory, Plot Multiple Courses and Set Your Course, and closes with a sales sequence: take the SwarmCompass Assessment, discuss your result with a consultant, launch a pilot.

The four verbs on the masthead, Engage · Discover · Adapt · Enact, belong to the sequential improvement cycle. They read as a slower, more reflective cousin of plan-do-check-act or of Boyd's observe-orient-decide-act. A brochure promising human-AI swarm intelligence that doesn't just analyse uncertainty—it navigates through it opens, then, with a four-stage linear sequence. That is not yet a contradiction. Many complexity practitioners use sequences as scaffolding. It is the first sign of a tension that will recur: the instrument talks of distributed emergence and behaves like a staged programme.

The claims, sorted by kind

The brochure makes roughly two dozen distinct claims. They are of four kinds. Empirical claims report facts about the world: radiologists, bees. Theoretical claims report how cognition or organisations work: AI is inductive, humans abductive. Metaphorical transfers carry a property from one domain to another, as in short-cycle feedback loops as a waggle dance. Commercial promises say what the buyer gets. A careful reader keeps these apart because each answers to a different tribunal. Empirical claims answer to the literature, theoretical claims to argument, metaphors to the fit of the analogy, and promises to delivery. The brochure blends all four in one register of confident assertion, and the blending is part of its persuasive method.

The table is weighted towards red, and the reason should be clear. The red verdicts do not cluster round the peripheral claims. They cluster round the load-bearing ones: the gorilla anecdote that names the product's central metric (the 17%), the bee biology that names the product itself (Swarm), and the consensus and alignment promise that is its deepest value proposition. The green verdicts (the 83% figure, the tempo claim) are the ones where the brochure states something either well documented or modest.

The silent premise

Under the inventory sits one premise the brochure never states and could not do without. There is a vantage point from which the swarm can be read. Someone deploys the micro-pulses. Someone identifies the 17%. Someone sees which options hold under pressure and sets the course. The swarm scouts and dances, and the leader reads the dance. This is the oldest arrangement in managerial cybernetics: a system in the middle, a sensor array at its periphery, and a governor at the top who reads the instruments. Norbert Wiener named the governor kybernetes, the steersman. The brochure presents its steersman as a swarm member, but he is still the one holding the compass.

The swarm scouts. The swarm dances. The client reads the dance.

Movement 2

Weaving Risk and Strategy: An Anatomy

WRAS has a different textual character. It is not a brochure. It is a working framework announced on a practitioner's blog, still openly under construction, still unsure of its own name. The September 2025 post offers several candidate expansions of the extra letters, all built on "Weaving Risk and Strategy…": through Sensemaking and Engagement, in Substrate Management, for Structural Evolution, from Emergent Substrates, through Substrate Entanglement. It invites readers' opinions. By then WRASSE was gaining traction, borrowed from the family of reef fishes.

The provisional status is an honest feature, and I count it in the framework's favour. A framework that admits it is unfinished is more complexity-congruent than one that arrives fully armoured. Even so, the provisional framework already makes strong claims. It is described as the last link in the chain towards a complexity-based alternative to strategy, and as taking a non-linear approach. Its core proposition, recorded on the Cynefin wiki, is that the basic principle of strategy is to respond to or manipulate the substrate before taking on the risk of a major intervention, which means starting from how things are rather than from an idealised vision.

The two axes

The framework crosses two dimensions. The vertical runs from natural emergence ("just let it happen") to stimulated emergence (acting in order to discover what is possible). The horizontal is granularity, from many fine-grained interventions to coarse-grained, massive allocations of resource. Four domains result:

The published descriptions do not say with full clarity which quadrant sits where. The figure below is my reconstruction, built from the definitions. Directed and Opportunist are both coarse-grained (they commit "significant" or concentrated resources). Directed manufactures its object through research, which puts it towards stimulation. Opportunist waits for an opportunity to appear, which puts it towards natural emergence. Hypothesis-based and Distributed are fine-grained. Hypothesis-testing deliberately stimulates in order to discriminate, and distribution lets many small things happen. Readers should hold this placement lightly. The critique in later movements does not depend on it.

The substrate

WRAS sits on a concept it takes from Estuarine Mapping: the substrate. The Cynefin wiki defines it as the properties of the underlying system and context, the space on, in and through which the organisation or individual grows. Within an organisation it is broken into four: the technological (software stack, data architecture, core technologies), the cultural (shared values, beliefs, behavioural norms), the procedural (core processes, governance, workflows) and the network (relationships with partners, suppliers, competitors). Strategy works on the substrate first, before any major intervention.

Lineage

WRAS is the newest member of a family. Cynefin (Snowden, from 1999) sorts contexts by their causal ontology. The Flexuous Curves address time and the transition of narratives. Estuarine Mapping (first edition, October 2022) maps constraints by the energy and time needed to change them. It identifies what is in practice impossible to change and what cannot be controlled, and so locates the zone where strategic action can actually happen. Estuarine draws in turn on constructor theory through the Substrate-Independence Theory paper (Turner, Snowden and Thurlow, Systems, 2022). WRAS also links to AIMS and "the three As", which the wiki mentions without fully expanding. The wiki places WRAS largely in the top half of Cynefin, across the complex, complicated and aporetic/confused zones.

Movement 3

Family Resemblance

The two instruments have different registers and the same grammar. SwarmCompass is the shop window and WRAS is the workshop. Both promise to replace linear planning. Both make distributed small acts the virtuous default. Both rely on a natural-historical emblem, the bee and the wrasse. Both separate the system that emerges from the agent who reads, nudges or stimulates it.

The right-hand column is the subject of this disputation. Each entry is an assumption the instruments share, so the critique of one carries over to the other. The fourth row, read, then act from above, is the root from which most of the others grow. Part III follows that root through the corpus.

Two instruments, one grammar. The system emerges, the client reads it, and the consultant sells the reading glasses.

Movement 4

The Gorilla and the 17%

The brochure opens under the banner Find the 17%. Radiologists were asked to look for anomalies in a set of X-rays. Eighty-three per cent did not see a gorilla, forty-eight times the size of a cancer nodule, on the final image. The brochure then adds that the seventeen per cent who did see it came to believe they were wrong when they talked to the eighty-three per cent who did not. The second sentence carries most of the commercial weight. It turns a perceptual finding into a social one, and on that social finding the product's central offer rests: identify the few who see, and protect them from the many who do not.

What the study found

The study is Drew, Võ and Wolfe, The Invisible Gorilla Strikes Again: Sustained Inattentional Blindness in Expert Observers (Psychological Science, 2013). Twenty-four radiologists performed a familiar lung-nodule detection task on five chest CT cases, scrolling through the slices while their eyes were tracked. In the last case a gorilla, more than 48 times the size of the average nodule, was inserted. Twenty of the twenty-four did not report it. Eye-tracking showed that most of those who missed it looked directly at it. They scrolled past the gorilla's layer several times on average. Afterwards, shown the image again, they could all see it.

The brochure has three of these facts right: the 83%, the 48×, and the setting. (The images were CT slices rather than plain X-rays, a pardonable simplification.) The fourth claim is the problem. The study has no conversation phase. The radiologists were not brought together. The four who saw the gorilla were not exposed to the twenty who did not, and nobody "came to believe they were wrong". The radiologists worked alone and were questioned at the end. The added sentence reads like the plot of Asch's line-judgment conformity experiments of the early 1950s grafted onto a study of attention. However it got there, it is not in the source.

Why the error matters theoretically

The embellishment is more than a slip in footnoting, because it changes where blindness lives. In the study, blindness belongs to a situation: an expert, a task, an instruction, a search template. Change the instruction and the blindness changes. The Utah follow-up work by Williams, Drew and colleagues found that radiologists missed a large breast mass and enlarged lymph nodes at high rates while hunting for nodules, yet detected the same abnormalities easily when asked to look for them. The same work found that experience gave no protection: seasoned radiologists were as likely to miss the unexpected as trainees.

Now take the brochure's instruction: Identify your 17% – the individuals who sense shifts before they're measurable… find the mavericks you need to pay attention to. This moves blindness out of the situation and into the person. It suggests that some people see and others do not, and that the task is to find the seers. On the evidence, the four radiologists who saw the gorilla in 2013 are not a stable caste of noticers. Under a different instruction on a different day, the twenty might see and the four might not. Selecting mavericks does not remove inattentional blindness. It assigns the blindness to someone else's search template.

This is where the complex responsive processes school comes in. Stacey's long argument against the monadic individual (in the fifth edition of Strategic Management and Organisational Dynamics, strategic choice theory treats the individual as prior to and primary over the group) applies directly. "The 17%" is a population of monads with a trait. In Mead's account, perception is a phase of the social act. What a radiologist sees is shaped by the generalised other of radiology, the "me" that knows what a scan is for. The gorilla was invisible because it was not a social object of the radiologists' practice. In Griffin and Stacey's sense, a social object is a generalised tendency to act that takes particular form in local interaction. Nodules were social objects; gorillas were not. The remedy is not to find the people who stand apart from the practice. It is to change the conversation in which the practice is constituted, so that the unexpected becomes something people are expected to report.

The survey that maps what cannot be seen

One more irony sits in the "Next Steps" column. The first step is the SwarmCompass Assessment – a question survey to map your decision-making blind spots. The 2013 study is about precisely what cannot be self-reported. The radiologists did not know they had missed the gorilla. Asked directly afterwards, they said they had seen none. A questionnaire about blind spots meets the problem that blindness is invisible to the person who has it. At best it maps beliefs about one's blind spots, which is a different and more flattering object. At worst it becomes what Mowles and Stacey call a social defence: a procedure that relieves the anxiety of not-knowing by producing a document that looks like knowing.

You cannot ask someone where their blind spot is. That is what makes it a blind spot.

Movement 5

The Swarm That Always Chooses Best

The product is named after a swarm. The brochure describes bee scouts seeking out a new home and returning to perform a dance that shows the site's location and quality. After a short period the entire swarm moves to a new location, and the brochure says it is always the best of those to have been investigated. It then draws three lessons: adaptive scouting, dynamic course correction through short-cycle feedback loops that act as the organisational waggle dance, rapidly aligning on optimal paths, and resilience-focused decision making on the grounds that the swarm never bets everything on one option—neither should you.

The biology is among the most elegant in behavioural ecology, mostly through Thomas Seeley's decades of work at Cornell with Kirk Visscher, Kevin Passino and others, summarised in Honeybee Democracy (2010). It is worth getting right. Getting it right contradicts the brochure at nearly every point where the brochure draws a lesson.

Five corrections

First, "always" is "almost always". In Seeley's island experiments, swarms offered one superior and several mediocre nest boxes chose the superior site in most trials, even though it was never the first site found. Most trials is not every trial. The accuracy is impressive because it is high and not perfect, and the cost of that accuracy is time.

Second, the process is slow on purpose. The deliberation takes hours and often days. Seeley describes scouts debating for two days, revisiting sites, and slowly building up support. His own lesson for human groups is to discourage rapid consensus. The brochure's rapidly aligning on optimal paths reverses the biology. The swarm is accurate because it resists rapid alignment.

Third, the swarm decides by quorum, not consensus. Once about fifteen or more scouts are present at one site at once, they switch behaviour. They begin piping, which signals the swarm to warm its flight muscles, and the swarm leaves. In the paired-site experiments the swarm took off while scouts were still dancing strongly for the rival site. A decision without consensus is the mechanism. The brochure promises to surface true consensus through participatory intelligence, which is the opposite of what its emblem does.

Fourth, the decision depends on inhibition as well as advocacy. Seeley and colleagues showed in 2011–12 that scouts deliver stop signals, brief head-butts accompanied by a vibration, to dancers advertising rival sites. This cross-inhibition, formally similar to inhibitory circuits in primate brains, prevents deadlock and lets one option win. The brochure's version is all positive feedback, dances aligning. The real swarm runs on positive and negative feedback, advocacy and inhibition. It is a contest of signals, not an alignment of perspectives.

Fifth, the swarm bets everything on one option. This is the most striking inversion. A swarm is the old queen and roughly half the colony. When it decides, it does not hedge across sites or move a little to each. It commits all of itself to one cavity. The breadth is in the exploration phase. The commitment is total. The swarm never bets everything on one option—neither should you gets the biology backwards. If there is a lesson, it is explore widely, commit wholly, and let the threshold, not unanimity, decide when.

The deepest disanalogy: interest

There is a sixth point, and it goes past a correction to a question of category. Visscher put it directly in commenting on the stop-signal research: all the bees have a common interest in choosing the best available site. A swarm is a superorganism of sisters with a single reproductive fate. No scout benefits from the swarm choosing a poor site. No faction of scouts profits from one site at the expense of another. No scout's bonus depends on her site winning. Information is honest because interest is shared.

Human organisations are made of exactly what the swarm lacks. Elias's figurations are webs of interdependent people bound by asymmetric and shifting power-ratios. In Complex Responsive Processes in Organizations Stacey makes power a property of every relationship and not a disturbance of some ideal neutral state. Mowles, in Rethinking Management and Complexity: A Key Idea for Business and Society, insists that any complexity-informed account of organising must take politics seriously, because people have interests, form coalitions and include and exclude one another. Scouts dance honestly because nothing is gained by dishonesty. Employees, customers and partners scanning the environment for SwarmCompass are in a different position. What they report bears on their careers, their budgets, their teams, their standing and their fears. In the anthro-complexity programme's own terms (the Multi-Ontology Sense Making paper distinguishes human from non-human complexity and coins contextual complexity for the difference), the bee is the wrong emblem precisely because it is not human.

The third column is not offered as advice to readers. It shows how much more interesting the true analogy is than the one sold. The real swarm offers organisations a genuine challenge: decide before agreement, institutionalise inhibition, commit wholly. That challenge is uncomfortable, and the brochure replaced it with something comfortable.

The bees teach that you can decide while others still disagree. The brochure sells the bees as a way to make everyone agree.

Movement 6

The Wrasse and Its Mimic

WRAS became WRASSE partly for the fish. The devopsdays abstract gives the reasoning. Wrasse are known in ecology as cleaners of the ecosystem and masters of adaptation. Like the wrasse that changes colour to thrive in shifting environments, the framework helps organisations adapt. Like the wrasse that removes parasites and keeps the reef healthy, it helps identify and reduce risk. The name, the abstract says, evokes intelligent adaptation and symbiotic relationships within a complex system.

The emblem was well chosen, though not for the reasons given. The bluestreak cleaner wrasse, Labroides dimidiatus, is one of the most closely studied cases in the evolution of cooperation, and what has been learned about it is less comforting than the abstract suggests.

What the cleaner wrasse actually teaches

The cleaner prefers to cheat. Grutter and Bshary (2003) showed that cleaner wrasse prefer client mucus and tissue to the parasites they are supposed to remove. The relationship is mutualist only on the surface. Underneath it is a standing conflict of interest, held in check by the clients.

The cleaner behaves better when watched. In Image scoring and cooperation in a cleaner fish mutualism (Nature, 2006), Bshary and Grutter showed that clients eavesdrop on cleaning interactions and prefer cleaners they have seen behaving cooperatively. Cleaners, in turn, learned to feed more cooperatively when an image-scoring bystander was present. An earlier paper by Bshary had a franker title: Biting cleaner fish use altruism to deceive image scoring clients (2002).

Clients discipline cleaners through a market. Clients that can switch cleaners get better service. Clients punish cheating cleaners by chasing them or leaving. Cleaner mutualisms are a textbook biological market. The reef stays healthy not because the cleaner is virtuous but because the clients are vigilant and have alternatives.

The cleaner has a mimic. The sabre-toothed blenny, Aspidontus taeniatus, has evolved to look like the cleaner wrasse in colour, stripe and swimming display. Clients approach it expecting a cleaning and are bitten. It exists because the cleaner's reputation is valuable enough to fake.

"Changing colour" is usually sex change. Many wrasses are protogynous hermaphrodites. In Labroides harems the dominant female can become male when the male is lost. The colour changes the abstract celebrates are mostly changes of reproductive role inside a hierarchy, not adaptive camouflage.

This table is not meant as a gotcha. Snowden is a deliberate and literate chooser of metaphors, and the wrasse may have been picked partly for its ambiguity. Even so, a framework that chooses an emblem takes on the emblem's whole natural history. Read that way, the wrasse is a more honest emblem than the abstract allowed. It says that cleaning is a contested relationship, honest only under watching eyes.

The cleaner wrasse is honest when watched. So is every framework.

Movement 7

Lincoln, 1862: Borrowed Thunder

The brochure closes with Lincoln: the dogmas of the quiet past are inadequate to the stormy present; the occasion is piled high with difficulty, and we must rise with the occasion; as our case is new, we must think anew and act anew. The quotation is accurate and correctly dated, from the Annual Message to Congress of 1 December 1862.

Context matters for the reading, as it always does. The passage belongs to Lincoln's argument, in the second winter of the Civil War, for proposed constitutional amendments on compensated, gradual emancipation. It comes a month before the Emancipation Proclamation took effect. The sentences after the quoted lines are the famous ones: we must disenthrall ourselves, and then we shall save our country. Lincoln is not praising distributed sensemaking. He is a wartime executive using all his rhetorical sovereignty to persuade a legislature to make a deliberate, coarse-grained, irreversible intervention at the level of the Constitution, on the gravest moral question of the republic.

In WRAS terms the speech is Directed, not Distributed. The brochure borrows the thunder of a sovereign act of political will to sell an instrument whose stated philosophy is the opposite: many small probes, emergence over design, the swarm over the steersman. The quotation shows, more truthfully than the brochure intends, the desire running underneath it. The client wants to be Lincoln: to see the stormy present clearly, to rise with the occasion, and to act. The swarm is the method and the sovereign is the fantasy.

The 17% who saw the gorilla, the bees who found the best site, and the President who disenthralled the nation are three images of one hero: the one who sees rightly when others do not, and is vindicated.— the brochure's implicit myth

Shaw's warning in Changing Conversations in Organizations applies here. She says we are never on top of or ahead of the process we participate in, and that the wish to be so leads to delusions of omniscience and omnipotence. The closing Lincoln is not a flourish. It shows what the instrument is for. It promises the reader the place of the one who sees, which is the place complexity says nobody occupies.

The swarm is the method. The sovereign is the fantasy.

Movement 8

Five Teleologies and a Split Causality

In Complexity and Management: Fad or Radical Challenge to Systems Thinking? (2000), Stacey, Griffin and Shaw gave organisational theory one of its sharpest diagnostic tools. They asked of any theory what kind of causality it assumes, meaning what it takes movement into the future to be for and towards. They identified five teleologies. The tool's force lies in a second finding: most management theories use two of them at once, one for the organisation and another for the manager, and never notice the split.

The transformative teleology is the CRP school's own proposal. In it the future is created in the present through the interplay of intentions, and no one, not leader, consultant or theorist, stands outside the process as its author or observer. The guard-phrase is this: each moment is a repetition of the past with the potential for transformation.

Locating the instruments

Run the instruments through the table. The swarm, as the brochure presents it, is formative. It always arrives at the best site among those investigated, so the optimum is latent in the option set and the process unfolds it. The substrate in WRAS is also formative. Its energy gradients, propensities and dispositions are properties of the system that will express themselves along the paths of least resistance. Estuarine Mapping's aim, to manage dispositions so that what you want costs less energy than what you don't, is formative teleology stated as technique.

The client, though, is rationalist. The client deploys micro-pulses, identifies the 17%, sees which options hold, sets the course, chooses the quadrant, sets the granularity, decides whether to let emergence happen or to stimulate it. The consultant is doubly rationalist. The consultant designs the signifiers, the assessment and the pilot. The novelty of the future lives in the choosers. The system's job is to unfold.

Why the split survives

Why should a programme as deliberately anti-systems-thinking as Snowden's rebuild the split? The CRP answer is that the split is not an intellectual mistake waiting for a better argument. It is a response to anxiety. Clients do not buy instruments that tell them they are one dot among many, unable to stand outside the patterns they help produce. They buy instruments that give them back the upper box. In Complexity and Organization: Readings and Conversations and the later editions of Strategic Management and Organisational Dynamics, Stacey and Mowles repeatedly observe how complexity concepts are taken up by management discourse and returned to the service of control. Emergence becomes a lever, self-organisation becomes a technique, the edge of chaos becomes a destination. The instruments are a fresh case of a familiar process.

To be fair, Snowden himself has argued against this capture for years. He objects to "complexity" sold as a fad, to the abuse of complex adaptive systems theory as metaphor, and to the claim that one can engineer outcomes in a complex system. His long insistence on managing the evolutionary potential of the present rather than designing for an idealised future is close in spirit to transformative teleology. The disagreement here is not about his values. It is about whether his instruments, once they are tools that someone deploys on someone else, can honour those values. Any tool sold to a client sets up an upper box by doing so. The real question is whether the tool acknowledges this and builds the acknowledgement into its practice.

Every chooser is chosen too. An instrument that forgets this is systems thinking under a complexity name.

Movement 9

There Is No Outside: Mead and Elias

The vertical axis of WRAS, natural emergence ("just let it happen") to stimulated emergence ("taking action to discover what is possible"), is the framework's most interesting move and its most vulnerable. It is interesting because it acknowledges that emergence goes on whether or not anyone intervenes. It is vulnerable because it treats letting it happen as a position available to an agent, a null option of non-intervention from which stimulation is a departure.

The gesture that cannot be withheld

For George Herbert Mead, social life is made of the gesture–response. A gesture is any act that calls out a response in another, and its meaning is not in the gesture but in the whole social act, gesture and response together. Meaning is always completed by the other. A raised eyebrow means something only in the response it gets. Mead adds that a gesture also calls out a response in the one who makes it, because humans can take the attitude of the other towards themselves. That is the origin of self-consciousness.

It follows that not acting is a gesture. A leader who "just lets it happen" is seen to let it happen. Their silence in the meeting, the project they do not cancel, the probe they decline to fund, the report they read and do not answer all call out responses. People infer approval, indifference, a hidden plan, a loss of nerve. Those responses then become the emergence the leader believed they were leaving alone. In a social figuration, natural emergence is not available to anyone whose position is visible. The only real distinction is between gestures one makes knowingly and gestures one makes without noticing.

Elias and the asymmetry of power

Norbert Elias adds what Mead lacks: power. In What is Sociology? Elias builds game models of increasing complexity, from two players to many players at many levels, to show that power is not a thing one holds but a balance, or ratio, between interdependent people. As figurations lengthen and differentiate, no player, however strong, can control the game's course. Elias observes that the game then increasingly takes on a life of its own, beyond what any player intended.

Elias's model adds a crucial point. In long figurations the more powerful players experience themselves as standing outside or above the game, because their moves have larger effects. That experience is real, and it is still an illusion, because their larger moves call out larger responses that reshape their position. The WRAS chooser at the crossing of the axes is the powerful player. Being able to choose between "letting it happen" and "stimulating" is itself a measure of position in the power figuration, and the framework presents it as a neutral methodological choice. A junior nurse cannot choose to "let it happen" in the same sense as the chief executive. Her inaction is invisible, his is a signal. The axis hides the power that makes it meaningful.

In Complexity: A Key Idea for Business and Society Mowles describes the social game as one that influences us more than we are able to influence it. That observation is the counter-axis to WRAS. It runs from those who feel they influence to those who know they are influenced, and every organisation is distributed along it.

The reformulated axis runs from unnoticed to noticed gesture. It replaces a distinction between kinds of emergence with a distinction between degrees of reflexive awareness in the one who gestures. Emergence is not something one can choose to have or not have. What one can choose is how much attention to pay to one's own part in it. That is a discipline of the chooser, not a property of the situation.

There is no "let it happen" for anyone whose silence is watched.

Movement 10

The Substrate as an Ontology

Every metaphor smuggles in an ontology, and the word substrate carries a heavy one. In biology a substrate is the surface on which an organism lives or grows, or the molecule an enzyme acts upon. In materials science it is the base layer on which other layers are deposited. In each sense the substrate is passive relative to what acts on it or grows on it, and prior to it. WRAS says strategy should respond to or manipulate the substrate before major intervention. The substrate is the thing worked upon.

Juarrero: constraint, not substrate

Much of the Cynefin family's recent sophistication comes from Alicia Juarrero, whose Dynamics in Action (1999) and Context Changes Everything (2023) rebuild causality around constraints: enabling and governing, context-independent and context-sensitive. Snowden has drawn on this work extensively, and the vocabulary of governing, enabling and dispositional constraints runs through Estuarine. On Juarrero's account constraints are not a substrate. They are relations: they alter the probability distribution of what can happen next, and the interactions they constrain produce and reshape them in turn. Context-sensitive constraints in particular are made by the dynamics they govern. A river's banks shape the flow, and the flow reshapes the banks. Tellingly, the estuary is the image Snowden himself chose.

A faithful reading of Juarrero cannot keep a substrate under strategy. Strategy and substrate are co-constituted. The "cultural substrate" is not a layer of values and norms waiting to be manipulated. It is the ongoing patterning of what people do together, and that includes what they do with strategy documents, consultants and SenseMaker surveys. An intervention on the cultural substrate is not an act on culture. It is more culture.

Gibson: affordance belongs to no one side

Estuarine work invokes James Gibson's affordances, the parts of the substrate one can get a handle on. In The Ecological Approach to Visual Perception (1979) Gibson was emphatic that an affordance is neither objective nor subjective but both, a relation between an animal and its environment. A step affords climbing to a creature of a certain leg-length and not to another. Gibson's affordance cannot be mapped as a property of the substrate alone. It changes with the body, the skill, the history and the purpose of whoever encounters it. An affordance map of "the substrate" without the people whose bodies and histories complete it is a map of half a relation.

Griffin, Elias, Ingold: culture as the moving pattern

In Griffin and Stacey's account (especially in Experiencing Emergence in Organizations), what systems thinkers call culture is a set of social objects: generalised tendencies of large numbers of people to act in similar ways in similar situations. These exist only in being particularised in local interaction, here and now, by these people. They are not beneath interaction. They are its recurring shape. Elias's figuration makes the same point at larger scale: a society is not a layer under individuals, nor individuals a layer on top of society. "Individual" and "society" are two perspectives on one moving web.

Tim Ingold's distinction between the network and the meshwork (in Lines and Being Alive) gives the sharpest image. A network joins pre-existing points with connectors. A meshwork is made of lines of movement, entangled trails of beings who are themselves their paths. The WRAS "network substrate" (partners, suppliers, competitors) is a network in Ingold's derogatory sense: nodes and links. The organisation that lives is a meshwork, a tangle of lines still being drawn.

Constructor theory and the unchanged constructor

The deepest root of the substrate concept in the Cynefin family is constructor theory, proposed by David Deutsch and Chiara Marletto as a reformulation of physics in terms of which transformations are possible and which impossible. It was taken up in the Substrate-Independence Theory paper. In constructor theory a constructor is an entity that causes a transformation in a substrate and keeps its ability to cause it again. A catalyst is the paradigm case. The constructor is defined by being, in the relevant respect, unchanged by its work.

This is the precise point where the import fails for human affairs. On the CRP account, and on Mead's, no human participant remains unchanged by participation. The strategist who intervenes in the "cultural substrate" is changed by doing it: by the responses called out, by the reputational consequences, and by the shift in their own "me" as others now see them as the one who intervened. A catalyst that is transformed by the reaction is not a catalyst. It is a reagent. In organisations everyone is a reagent. Constructor theory may be a profound reformulation of physics. Carried into anthro-complexity, it brings with it the figure of the agent who acts without being acted upon, which is the upper box of Movement 8 restored at the level of fundamental ontology.

The final row is the argument of the movement in a single line. The four substrates leave out the one group that a complexity-faithful account must not leave out: those doing the strategising. Put them back as a fifth, reflexive "substrate" and the stratigraphy falls apart, because a layer that reads the other layers and is changed by the reading is no longer a layer. It is the weave.

A catalyst changed by the reaction is a reagent. In organisations, everyone is a reagent.

Movement 11

Against Prestatability

The brochure promises, twice, to turn "unknown unknowns" into mapped territory. It promises to map the energy gradients of the current state: what can't be changed and what is most likely to change. These are the most philosophically consequential sentences in either instrument, because they claim that the space of what may happen can be laid out in advance and populated with likelihoods.

Kauffman: the adjacent possible cannot be listed

Stuart Kauffman's Investigations (2000) and his later work with Giuseppe Longo and Maël Montévil (No entailing laws, but enablement in the evolution of the biosphere, 2012) make a precise claim. The phase space of an evolving biosphere, and of an economy or culture, cannot be prestated. The adjacent possible, the set of things one step from what exists, is not a fixed list waiting to be found. Its members come into being only as their precursors do, and they create new adjacencies in turn. Kauffman's emblem is the screwdriver: list in advance all its possible uses (open a paint tin, scrape a wall, defend oneself, prop a window, serve as a plumb-bob…) and you find the list has no algorithmic end. The uses depend on contexts that do not yet exist.

If the phase space cannot be prestated, then the unknown unknowns are not hidden regions of an existing map. They are regions that do not yet exist and that will be made by the moves, the brochure's included, taken in the present. You cannot map them, and not because the cartographer lacks skill. There is nothing there yet to map. The promise is not hard to keep. It is a category error.

Cilliers: no compression

Paul Cilliers, in Complexity and Postmodernism (1998) and the papers collected as Critical Complexity, argued that complex systems are incompressible. No model of a complex system can be simpler than the system without leaving something out, and we cannot know in advance whether what we left out matters. Cilliers drew an ethics of modesty from this. Every frame is a choice, every choice excludes, and the responsible modeller acknowledges the exclusion as theirs. A complexity practitioner faithful to Cilliers can promise a useful partial view and must say what it leaves out. They cannot promise to turn the unknown into the mapped.

Snowden against the brochure

The sharpest refutation comes from inside the Cynefin family. The Cynefin.io page on risk management argues that conventional risk assessment wrongly assumes a Gaussian distribution, whereas complex systems show power-law distributions with fat tails, so that extreme events are more likely than they first appear. It concludes that risk calls for continuous sensing, early detection and rapid response rather than detailed prediction and avoidance. That is good complexity science, in line with Bak's self-organised criticality and Taleb's critique of the bell curve. It is also incompatible with mapping what is most likely to change. Under a power law the "most likely" changes are the small, frequent ones, and the changes that matter, the tail events, are by construction not the most likely. A map of likelihoods is the Gaussian instinct coming back.

The right-hand column matters. Each claim has a defensible complexity version, and the Estuarine heritage is often near it. The brochure takes the defensible version (constraint mapping, the energy cost of change) and inflates it into the indefensible one (mapping the unknown, predicting likelihood). The inflation is commercial and not scientific, and it is the commercial inflation that Movement 17 examines.

You cannot map what does not yet exist. You can only become quicker at noticing that it now does.

Movement 11 · Pro Parte Contraria

The Adversary's Brief: The Adjacent Possible Stated, the Complex Compressed

A disputation that only argues one side is a sermon. The scholastic method requires the master to put the opposing case in its strongest form before replying, and to put it better than its own advocates did. Movement 11 argued that the adjacent possible cannot be stated in advance (Kauffman) and that complex systems cannot be compressed (Cilliers). This movement argues the contrary as forcefully as the literature allows. It then shows exactly how much of Thesis VII survives and how much has to be given up.

Videtur quod sic: it seems that the adjacent possible can be stated, and the complex compressed.

I. Six ways the adjacent possible is stated in advance

1 · It was born stateable

The first argument is genealogical, and it is the most embarrassing for the strong thesis. Kauffman first developed the adjacent possible in his work on molecular and biological evolution. There the adjacent possible of a chemical soup is the set of molecules one elementary reaction step away from those that exist. Tria, Loreto, Servedio and Strogatz restate the concept in exactly these terms: the things one step from what actually exists, reachable by incremental modification and recombination of existing material. Give a reaction grammar and a set of species, and the one-step neighbourhood is enumerable. It may be astronomically large, but it is listable in principle. In its founding sense the adjacent possible is a combinatorial neighbourhood, and combinatorial neighbourhoods can be stated in advance.

The strong thesis of unprestatability is a later and narrower claim, made by Longo, Montévil and Kauffman in 2012. What cannot be prestated is the phase space of biological evolution: the ever-new relevant observables and functions that selection reveals, such as the swim bladder that becomes a niche for a worm. The adversary's first point is therefore that the strong thesis concerns functions, not configurations. Movement 11 slid from one to the other.

2 · Intension, not extension

To state a set in advance you do not have to list it. A grammar of a few dozen rules states the infinite set of a language's sentences. A recursive definition states the integers. The objection from the screwdriver (list all its uses and the list never ends) shows that the uses cannot be listed. It does not show that they cannot be generated. Treat the screwdriver as a vector of properties (rigid, pointed, elongated, conductive, a lever-arm of known length, a mass of so many grams) and the space of physical operations those properties afford is the image of that vector under a finite vocabulary of actions: prise, pierce, wedge, conduct, weigh, prop, scrape, strike. "Prop a window" is not mysterious. It is rigidity × length × wedge. The adversary says that the uses of the screwdriver have an intension even where they lack a finite extension.

3 · The statistics of novelty can be predicted

The empirical counter-blow came in 2014. Tria, Loreto, Servedio and Strogatz built a model of the adjacent possible, a generalisation of Pólya's urn in which each novelty adds new possibilities to the urn ("triggering"). The model predicts statistical laws for the rate at which novelties occur (Heaps' law), for the distribution over what is explored (Zipf's law), and for signatures of one novelty enabling another. They tested these predictions against Wikipedia edits, annotation tags, the words of texts and online music listening, and all four datasets showed the predicted patterns. One sentence of their paper amounts to a direct rebuke: it had not been clear how to extract testable predictions from Kauffman's idea. They extracted them.

4 · The shape of the explosion can be predicted

The adversary's most ironic witness is Kauffman himself. The TAP equation, the Theory of the Adjacent Possible, worked out by Cortês, Kauffman, Liddle and Smolin, counts the ways new objects can be generated from combinations of existing ones, minus an extinction rate. Its generic behaviour is a long plateau followed by a sharp, explosive divergence, and the authors derive accurate analytic approximations for the blow-up time, validated against simulation. Which objects will appear is not stated. When the space of objects will explode is.

5 · The mass of the unseen can be measured

The adversary's most elegant witness was English and worked in Buckinghamshire. At Bletchley Park, Alan Turing and his assistant I. J. Good developed what Good published in 1953 as the Good–Turing estimator. Given a sample, the probability that the next observation belongs to a species never yet seen is estimated as N₁/N: the number of species seen exactly once, divided by the sample size. The estimator is largely independent of the distribution of species frequencies. It has since been shown to coincide, under Pitman–Yor priors, with the modern Bayesian nonparametric estimator of discovery probability.

Here, read carefully, is a formula for the missing mass: the total probability carried by what you have not seen and cannot name. Turing used it against Enigma's keys. Ecologists use it for unseen species, and language models use it for unseen words. It does not tell you what the unknown unknowns are. It tells you, with known error, how much of the world they occupy. The adversary presses the point. "Unknown unknowns" was always a loose phrase. Turing turned half of it into a number.

6 · Even technological surprise has a distribution

The sixth witness is economic. Farmer and Lafond (Research Policy, 2016) treated generalised Moore's laws as correlated geometric random walks with drift. Across 53 technologies they derived a closed-form approximation for the distribution of forecast errors as a function of time horizon. In hindcasting, forecast errors across many technologies and horizons collapse onto the same universal distribution. Technological futures, archetypal sources of novelty, are not stated in advance in their content. They are forecastable in their error bars, and you can estimate the probability that one technology overtakes another at a given date.

II. Six ways complex systems are compressed

1 · Why science is possible at all

Cilliers held that a complex system cannot be given an account less complex than itself without losing some of its aspects. The adversary answers from statistical physics and systems biology. In a 2013 paper in Science, Machta, Chachra, Transtrum and Sethna showed that parameter space compression underlies the emergence of effective theories. Using the eigenvalues of the Fisher Information Matrix, they found that a few "stiff" parameter combinations govern macroscopic behaviour while most directions are "sloppy" and barely matter. The pattern holds in the diffusion equation, at the Ising critical point, and in multiparameter models across the sciences. Sethna puts it plainly: if one needed to extract every detail of the true underlying theory to make a useful theory, science would be impossible. In a cell with thousands of interacting parameters, only a few combinations predict how the system behaves.

2 · The minimal optimal predictor exists

Crutchfield's computational mechanics gives the most rigorous answer to "how much can a process be compressed without losing predictive power?" The rule is simple: do not distinguish histories that lead to the same predictions of the future. The resulting causal states and their transitions form the ε-machine, the process's unique optimal predictor of minimal size, which is in effect its equations of motion. Shalizi and Crutchfield proved the causal-state representation is the minimal one consistent with accurate prediction. This is compression with a guarantee: nothing predictive is lost, and nothing redundant is kept.

Honesty requires the qualification the adversary would sooner omit. For most processes the statistical complexity, the information the causal states store, is infinite, so storing them exceeds any learner's capacity. That is why Marzen and Crutchfield developed lossy causal states and predictive rate-distortion theory. Compression is real, and it is graded. One can compute exactly how much prediction is bought with each bit of model.

3 · Macro can beat micro

Cilliers's incompressibility assumes that compression always loses. Hoel, Albantakis and Tononi showed in 2013 that this is false in a precise sense. Using effective information as a measure of causal power, they found that a coarse-grained macro model can carry more causal information than the micro model it is built from, even when the macro is completely specified by the micro. This happens when macro mechanisms are more deterministic or less degenerate, enough to outweigh the smaller state space. They call this causal emergence. A map can be more informative than the territory. Some compressions are gains.

4 · A city compresses to an exponent

The social sciences have their own striking compression. Bettencourt, Lobo, Helbing, Kühnert and West (PNAS, 2007) found that socioeconomic quantities in cities, such as wealth and new ideas, scale with population as a power law with an exponent β ≈ 1.15. The same family of exponents covers wealth, innovation and crime, and infrastructure scales sub-linearly. A city of millions, with all its uncountable interactions, compresses in these respects to a number, and the number recurs across urban systems. The 2010 follow-up found that each city's deviation from the line is itself stable over decades, a second-order compression of the city's character.

5 · Cilliers against Cilliers

The adversary's fifth point is internal. Cilliers never counselled refusing to model. His tribute literature records his insistence that reduction is necessary in order to deal with complex problems at all. What he called for was modesty: an attitude open to the provisional and ethical character of partial models. Critics have also noted that Cilliers preferred connectionist neural networks as the best way to model complex systems, a commitment that sits uneasily with his post-structural claim that no model is privileged. A neural network is, among other things, a compression of its training data. Cilliers's own favoured model of complexity was a compressor.

6 · Learning is compression

The sixth point is general. Every successful predictor of a complex system, whether a weather model, a language model or an experienced receiver's feel for a distressed asset, is a compression that discards the incompressible residue and keeps the predictive structure. It is now a commonplace of machine learning that predicting and compressing are two sides of one operation. If complex systems were incompressible in the strong sense, no organism could learn anything about its environment. Organisms do learn, so complex environments compress far enough to be lived in.

Respondeo: what survives and what is surrendered

The adversary has won a great deal, and Thesis VII has to be amended openly in his favour. The Additive Covenant governs this treatise, so the original text stands above, and the amendment is added here and in the Twelve Theses.

First concession. The adjacent possible can be stated in advance in its configurations (given rules), its generator (as intension), its rates (Heaps and Zipf), its shape (TAP), its missing mass (Good–Turing) and its error distribution (Farmer and Lafond). Movement 11's sentence, "there is nothing there yet to map", was too strong. There is something to map, namely the form of the not-yet.

Second concession. Complex systems are compressible in their predictive structure (ε-machines), their parameter geometry (sloppiness), their causal grain (causal emergence) and their aggregate scaling (urban exponents). Cilliers's incompressibility holds exactly for the complete micro-description and loosely for what is left over. It does not hold for the knowledge that action needs.

What is retained. Three things survive. First, the identity, function and meaning of particular novelties remain unstateable (rungs ⑦ to ⑨). Longo, Montévil and Kauffman's thesis concerns precisely these: new observables, new functions, new niches. Second, every compression of a human system is reflexive. Its publication is a gesture in Mead's sense and calls out responses that move what was compressed. MacKenzie's study of options pricing (An Engine, Not a Camera, 2006) showed a model reshaping the market it described. Third, Cilliers's ethical point survives untouched and is sharpened. Because compression is possible, choosing what to keep and what to discard is a real choice made by someone. Modesty is not the claim that nothing can be compressed. It is the acknowledgement that every compression has an author and an exclusion.

The table turns the earlier critique of the brochure (Movement 11, Table 8) into something more precise and more useful. The brochure's sin was not that it claimed too much knowledge of the unknown. It claimed the wrong kind. It promised inventory ("mapped territory") where the science supports only mass, rate and form. A SwarmCompass rebuilt on Good–Turing, Heaps' law and sloppy-model analysis would be a defensible instrument, more so than the one on sale.

The form of the not-yet can be stated. Its name cannot. Wisdom is knowing which is which.

Movement 12

Codification: Boisot, Ashby and the Triad

SenseMaker® is the engine of SwarmCompass. The brochure says it uses a patented approach to capture human attitudes, going beyond text analysis to tap into intuition at scale and in quantitative form. Any serious critique has to begin by granting what is genuinely original in it.

What SenseMaker gets right

In conventional survey and qualitative research the analyst interprets what respondents say. SenseMaker's central innovation is self-signification. Respondents tell a short narrative of their experience, a "micro-narrative", and then interpret it themselves by placing it on a set of prepared signifiers. The best known are triads, triangles in which the respondent places a point according to how far each of three labelled qualities is present. There are also dyads (sliders between two poles), canvases and multiple-choice items. The narrative stays attached to its signification, so analysts can move from a pattern in the aggregate (a cluster of points in one corner of a triad) back to the stories that make it up. This is an elegant answer to the expert-interpretation problem, and it is fairly credited as one of the most intelligent attempts to make narrative usable at scale without simply counting words.

Boisot's I-Space

Max Boisot's Information Space (Knowledge Assets, 1998) gives the tool for seeing what self-signification does, both its achievement and its cost. Knowledge can be placed on three dimensions. Codification runs from uncodified (tacit, embodied, hard to articulate) to codified (formalised and compressed into a few symbols). Abstraction runs from concrete (tied to a particular situation) to abstract (general categories that hold across many situations). Diffusion runs from undiffused (held by a few) to diffused (available to many). Boisot's Social Learning Cycle follows knowledge as it moves: scanning, codifying (problem-solving), abstracting, diffusing, absorbing, impacting, and back into the tacit.

A lived experience in an organisation (the nurse's near-miss, the trader's unease, the tenant's complaint the receiver heard on site) starts at the origin of the I-Space: uncodified, concrete, undiffused. Telling a micro-narrative moves it some way along codification: it becomes words. Signifying it on a triad moves it far along codification and abstraction at once. A situated, ambiguous, many-sided experience becomes a point with three coordinates against categories designed in advance. Aggregation then moves it along diffusion, though only towards those who hold the dashboard.

The triad as search template

This brings us back to the gorilla. The radiologists missed it because their search template, nodules, set what counted as a finding. A signifier set is a search template. The three words at the corners of a triad set the dimensions along which an experience can register. Whatever the narrator felt that has no vertex is not lost exactly: it stays in the story. But it is invisible to the aggregate, which is what the client sees. The person who designs the signifiers stands exactly where the brochure's own anecdote puts the radiologist: an expert looking for a known kind of thing. The 17% the instrument wants to find are, almost by definition, those whose experience does not fit the template, and the template is the instrument's own.

Ashby: requisite variety

W. Ross Ashby's law of requisite variety (An Introduction to Cybernetics, 1956) says that only variety can absorb variety. A regulator must be able to take at least as many distinct states as the disturbances it means to counter. Measured this way, a triad is a severe variety reducer. It maps experience onto a two-dimensional simplex with three named axes. The point is not that this is wrong. All sensemaking reduces variety, and that is what makes it sensemaking. The point is that the claim to "tap into intuition at scale" overstates what survives the reduction. What survives is intuition as categorised by the designer. The brochure presents a reduction as an amplification.

Two of the losses are worth pulling out. The conversation is absent because SenseMaker collects narratives addressed to an instrument, not to another person. On the CRP account meaning is made in the conversational back-and-forth, so a story told to a form is a different social act from a story told to a colleague. The later reinterpretation is lost because a signification is a snapshot. Weick's sensemaking, as distinct from the trademarked "sense-making", is retrospective, plausible rather than accurate, and endlessly revisable. The narrator may understand the experience quite differently next week. The aggregate will not.

The triad is a search template. The gorilla is whatever has no vertex.

Movement 13

Risk Without Quadrants

WRAS's most original claim is in its name: risk and strategy should be woven, not handled by separate departments with separate registers. On this the framework is right, and right in a way that conventional enterprise risk management badly needs. Its wiki entry adds a second correct claim: complex systems are fat-tailed, so risk needs continuous sensing and fast response more than prediction and avoidance. The difficulty is the third claim, that each domain has its own risk strategy, matched to the domain's characteristics and to the risk profile of a portfolio.

Risk as a property, risk as an experience

Assigning a risk strategy to each quadrant treats risk as a property of the situation. A Directed situation has Directed risks, an Opportunist situation Opportunist risks, and you read off the right posture once you know the quadrant. This is how risk registers have always worked, with the quadrant taking the place of the old likelihood-impact grid.

The corpus offers a different ontology. In Experiencing Risk, Spontaneity and Improvisation in Organizational Change (2006), Patricia Shaw and Ralph Stacey's collection, risk is not something situations possess. It is something people live. It is felt as anxiety, negotiated in conversation, displaced onto others, defended against. The collection's argument about improvisation applies directly: improvisation is not a skill located in an autonomous individual, it is a quality of interaction, and it holds together only when driven by a theme the participants recognise. Risk in this sense lives in the relations, and so does the capacity to meet it.

Mary Douglas and Aaron Wildavsky (Risk and Culture, 1982) made the anthropological version of the same point. Societies select the risks they attend to according to their forms of social organisation. Which dangers count as risks is a social choice reflecting who holds power and what order they are protecting. A quadrant that assigns risks cannot ask whose risks they are, or who chose to see them.

Risk management as a social defence

Isabel Menzies Lyth's study of a London teaching hospital (The Functioning of Social Systems as a Defence against Anxiety, 1960) is foundational for the CRP school's view of procedure. Nursing routines (task lists, rotation, depersonalised categorisation of patients) existed largely to defend nurses against the anxiety of intimate contact with suffering and death. They worked as defences and failed as care. Stacey's The Tools and Techniques of Leadership and Management (2012) and his earlier "wiggly world" writing extend the diagnosis to strategic planning and risk registers. These instruments are performed alongside the real work as rituals that relieve the anxiety of not knowing. Stacey's line that muddling through is what managers do anyway, and is no second-best process, is the deflationary counterpart.

The question for WRAS is whether a four-quadrant risk posture is a better way of attending to risk or a more sophisticated defence against the anxiety of being unable to. The answer depends on use. A framework that prompts the conversation "whose anxiety are we managing, and what are we not looking at because of it?" attends to risk. A framework that tells a board which risk posture belongs to which quadrant is a defence.

The second column is a reconstruction. The published WRAS material says each domain has a risk strategy and does not spell them out. The reconstruction follows the logic of the definitions. The fourth row carries the most important point: "distributed" strategy does not just spread risk evenly. It moves risk down the power gradient. A hundred small probes run by front-line teams put a hundred small exposures on the people least able to say no.

Risk does not live in quadrants. It lives in whoever is left holding it.

Movement 14

The Human/AI Hybrid and the Harvest

The brochure's second page is headed Human-AI Hybrid: The Only Way to Navigate Foggy Futures. It separates two kinds of "superpower". Humans offer intuition for weak signals, ethical judgment, analogical thinking ("this feels like the 2008 crisis…"), and the capacity to think in stories and weave patterns in narrative over time to scaffold meaning. AI offers processing speed, pattern detection at scale ("if properly trained") and data aggregation. Earlier it claims humans reason in abductive, analogical ways, whereas AI is digital, inductive and dependent on its training datasets.

Peirce on abduction

The appeal to abduction is welcome, and Peirce deserves a better hearing than the brochure gives him. For Peirce, abduction is the only form of inference that introduces a new idea. It is the leap to a hypothesis that would, if true, make a surprising fact a matter of course. It is fallible, instinctive, guided by what he called il lume naturale, and it is the first stage of inquiry, followed by deduction (working out consequences) and induction (testing them). Abduction is not a human "superpower" set against machine induction. It is one phase of a cycle, and the cycle is social. For Peirce the community of inquirers, not the individual mind, converges on truth over the long run.

The clean split is also now empirically doubtful. Contemporary generative models produce candidate hypotheses, analogies and explanations that look like abductive leaps. Whether they "really" abduce is a live philosophical question, but the line between a human who guesses and a machine that only counts no longer holds without argument. The more interesting point is Peircean: neither humans nor machines abduce well alone. Abduction is disciplined by the community that tests it.

Hutchins: cognition in the wild

Edwin Hutchins's Cognition in the Wild (1995) studied navigation on a US Navy ship and showed that the cognition needed to fix the ship's position was not located in any one head. It was distributed across people, charts, alidades, plotting tools, speech and the division of labour. The system navigated. The brochure's image (AI sees the map. Humans feel the terrain.) divides cognition into two possessors with two kinds of property. Hutchins shows that the interesting cognition is in the arrangement, in who passes what to whom, in what form, at what moment. The question is not which superpowers to fuse. It is how to arrange the passing.

The harvest

The most revealing line in the brochure is a deliverable listed under Chart the Territory: Create epistemically balanced & multi-perceptive training data sets for further AI-based automation.

Read it closely. The human superpowers the brochure praises (weak-signal intuition, ethical judgment, narrative pattern-weaving) are to be captured through SenseMaker and turned into training data so that further activity can be automated. The 17% are found, and their seeing is recorded, so that it can be reproduced without them. If the human contribution really were irreplaceable, there would be nothing to automate. If it can be automated, the human/AI hybrid is a transitional stage: humans are the scaffolding the machine is built on and then dismantles.

They praise the human superpower, and then they train on it.

Movement 15

Compass, Star-Path and Orientation

The product is a compass. The metaphor is apt in ways its authors may not have meant. A magnetic compass gives a direction relative to a field that is invisible, global and independent of the traveller. It does not tell you where you are or where to go. It tells you which way is north.

Declination

A magnetic compass does not point to true north. It points to magnetic north, and the angle between the two, declination, varies with place and drifts over time as the Earth's field changes. A navigator who trusts the needle without correcting for local declination goes systematically wrong, and the error grows with distance. So a compass is honest only when used with local knowledge of its own error. A swarm compass would need a declination table: a running account of how its signifiers, its sample, its designers' assumptions and its clients' wishes bend the needle away from the situation. The brochure offers no such table. Its needle points to authentic alignment, which is the magnetic north of every management instrument ever sold.

The star compass

The Pacific wayfinding traditions studied by Ben Finney, David Lewis and others, and in other treatises of this imprint, used a very different instrument: the star compass, a mental division of the horizon into rising and setting points of stars, together with etak. In etak the canoe is held still and the islands move past it along the star bearings. The navigator does not stand outside the voyage reading a fixed field. The navigator reads swell, cloud, birds, the colour of the water and the feel of the hull, and continuously re-imagines the canoe's place in a world that is moving with it. Ingold's distinction in Lines between wayfaring (knowing as you go, along a path) and navigation by map (knowing from above, between points) is the philosophical form of the same contrast.

Boyd: orientation is the Schwerpunkt

John Boyd, whose OODA loop is probably the most misdrawn diagram in strategy, put Orientation at the centre. In his final sketch of the loop, Orientation is shaped by genetic heritage, cultural traditions, previous experiences, new information, and the analysis and synthesis that bring these together. It sends implicit guidance and control directly to observation and to action, bypassing explicit decision. Boyd's point was that the decisive advantage lies in the quality of one's orientation, and that orientation is built through immersion, varied experience and the destruction and recreation of one's own mental models (Destruction and Creation, 1976). It is not consulted.

Against this, an external "compass" that delivers orientation as a product is a regression. It locates the decisive faculty outside the organisation that needs it. The brochure's one undeniably Boydian promise, Course-correct in days, not quarters, is about tempo, and tempo without orientation is just agitation. Boyd's advantage came from getting inside the opponent's loop, operating at a tempo that disoriented them. That requires an orientation already rich enough to read the situation faster than it changes, and no survey supplies one.

Orientation cannot be bought. It is grown in a crew that has been at sea together.

Movement 16

A Genealogy the Framework Omits

WRAS is introduced as the final link in a decade-long chain towards a complexity-based alternative to strategy, the step that completes the anthro-complexity approach. It deserves a genealogy, because its four domains have strong ancestors in mainstream strategy scholarship, and the absence of those ancestors from the framing gives an inflated sense of rupture.

Mintzberg and Waters, 1985

Forty years ago, in Of Strategies, Deliberate and Emergent (Strategic Management Journal, 1985), Henry Mintzberg and James Waters set out a continuum from deliberate to emergent strategy, with eight types along it: planned, entrepreneurial, ideological, umbrella, process, unconnected, consensus and imposed. Mintzberg's Crafting Strategy (1987) added the image of strategies that grow like weeds in a garden and called on managers to recognise emergent patterns rather than impose designs. WRAS's vertical axis, natural to stimulated emergence, is a reworking of the deliberate–emergent continuum with emergence on both ends. Its quadrants have clear Mintzbergian counterparts.

None of this makes WRAS derivative in a damaging sense. Good frameworks recombine. The objection is to the claim of rupture. A framework presented as the last link in a chain towards an alternative to strategy suggests that mainstream strategy never conceived of emergence, hypothesis-testing or distributed initiative. It conceived of all three, debated them for decades, and, in Burgelman and Quinn, documented them empirically in large firms. Acknowledging this would strengthen WRAS. It could then say exactly what its complexity lens adds: granularity, substrate and narrative sensing, which are real contributions.

The fad cycle

There is a further irony. Snowden has long and rightly mocked the management fad cycle, the repackaging of old ideas under new acronyms for a new round of consulting revenue. The WRAS naming history (several candidate expansions published, readers invited to vote, a reef fish adopted for its marketable metaphor) shows a framework that knows it will be branded and is shopping for a brand. In MacIntosh and colleagues' SAGE classification of complexity work, this is the drift from the critical-pluralist register towards the metaphorician one: complexity concepts used for their evocative power and not their explanatory force. The bee and the wrasse are metaphorician moves. The fat tail and the constraint are scientific ones. The instruments switch between the two registers, and the switching is the problem.

The chain was forged before its last link was announced.

Movement 17

The Political Economy of the Framework

The brochure ends with two lines of small print: copyright ©2025 Cognitive Edge Pte Ltd, and the SenseMaker® logos and name as trademarks of Cognitive Edge Pte Ltd (misprinted Pte td.). The final panel describes a patented approach. The WRAS wiki page says the framework is still under development and not to be used or shared without permission, and invites premium members to take part in its development. These are not incidental. They are the institutional form complexity takes when it comes to market.

Enclosure

Elinor Ostrom's Governing the Commons (1990) showed that common-pool resources can be governed sustainably by the communities that use them, without either privatisation or central control, given certain design principles: clear boundaries, rules matched to local conditions, participation in rule-making, monitoring by users, graduated sanctions, conflict resolution. The intellectual commons of complexity thinking (concepts like emergence, constraint, attractor, safe-to-fail, distributed cognition) was built by a vast, mostly unpaid community of scientists and practitioners. A framework built on that commons and then fenced with trademarks, patents and permission requirements is an enclosure in the precise historical sense: the conversion of a commons into property.

The conversion is legal and, commercially, perhaps necessary. Firms must earn revenue and research costs money. But it sits uneasily with the epistemology the instruments claim. A theory that knowledge emerges from distributed interaction, that the edge is where novelty lives, and that the 17% at the periphery see what the centre misses is a theory that argues against enclosure. Enclosure stops exactly the distributed recombination and peripheral challenge that complexity says produces new knowledge.

Alignment and recognition

The brochure's deepest promise is to navigate uncertainty as One Swarm: achieve authentic alignment (not just top-down compliance). Kevin Flinn's work on leadership development offers a test that cuts through this. Working within the complex responsive processes tradition, Flinn argues that leadership exists only where it is recognised by others, granted from below in the ongoing interaction and not imposed from above. Without recognition what remains is authoritarianism or coercion, which produce compliance and not followership. Leadership, in Flinn's account, is not about people adopting your story.

Measured by this, "authentic alignment" is a confusion. Alignment describes the outcome in which many vectors point the same way. It says nothing about how they came to point that way. Under Flinn's test, alignment produced by an instrument that surveys, ranks, identifies mavericks and sets a course is compliance with a more sophisticated mechanism. What separates followership from compliance is whether the people involved recognise the direction as theirs. They can recognise it only if they were part of the conversation that made it, not merely data for the instrument that found it.

Tools that control bodies

The darkest and most precise sentence in the corpus on this subject is Stacey's, from The Tools and Techniques of Leadership and Management: no one can control outcomes, but the powerful can control human bodies to a considerable degree. He names the mechanisms as power constraints, ideology, socialisation and disciplinary power. Put next to SwarmCompass, the sentence clarifies. The instrument cannot control the swarm's outcome, and it admits as much. It can arrange what people attend to (the signifiers), what they are asked to report (the micro-pulses), who is singled out (the 17%) and what their reports become (training data). That is not control of outcomes. It is the shaping of attention and conduct, which is what Foucault meant by disciplinary power. The phrase continuous scanning via human agents on the brochure's last panel names it.

A theory of the periphery, sold from the centre.

Movement 18

Way, Grain, Propensity

Credit comes first, and generously. Of all the Western complexity frameworks, the Cynefin family's recent work is the closest to classical Chinese strategic thought. Estuarine Mapping asks practitioners to start journeys with a sense of direction rather than abstract goals, to work with propensities and dispositions so that the desired costs less energy than the undesired, and to begin from how things are rather than an idealised vision. This is very close to what François Jullien, in The Propensity of Things (1992) and A Treatise on Efficacy (1996), identified as the core of Chinese strategy. One does not project an ideal model and impose it on reality by force of will, which is the Western means–end schema Jullien traced from Aristotle to Clausewitz. One reads the potential of the situation and lets effect come from it.

Because the convergence is real, the divergence is instructive. Read through 道理勢, the instruments seize one term of the triad and lose the other two.

勢 Shì: propensity seized

勢 shì is the configuration's potential: the slope that gives the boulder its force, the bow drawn and ready, the position that makes victory follow. Sunzi's fifth chapter, 勢篇, says the skilled commander seeks victory in 勢 and does not demand it of his men. He chooses people and lets the configuration work. The Estuarine "energy gradient" and the brochure's map the energy gradients of the current state are 勢 rendered in thermodynamics. The instruments grasp 勢 well. They locate it in the substrate, read it, and propose to act along it.

Jullien insists on something the instruments leave out: 勢 includes the strategist. The commander does not stand outside the configuration reading its potential. Their own disposition, timing and restraint are part of the configuration, and the opponent reads it too. 勢 is not a property of the terrain. It belongs to the whole situation, both armies, ground, season and morale, and it changes with every move. To map 勢 is to turn it into a landscape, a substrate, and to lose its essential mobility. Jullien's term for the mobility is transformation: 化 huà, silent and continuous, which no one sees happen and everyone sees has happened.

理 Lǐ: grain lost

理 lǐ first meant the veins in jade, the grain along which the carver must work. The great image is Zhuangzi's Cook Ding (庖丁解牛, chapter three). After nineteen years his blade is as sharp as the day it was ground. He does not hack at the ox. He relies on the heavenly pattern, 依乎天理, and slips the edge, which has no thickness, into the spaces that are already there. 以無厚入有閒: insert what has no thickness into what has room. When he comes to a difficult knot he slows, his gaze settles, and the blade barely moves.

Ding's skill is a perception that comes through the hands, over years, in contact with this ox. He says he no longer sees with his eyes but meets the ox with his spirit. That is 理. It cannot be surveyed, and it cannot be found with a triad. It is the grain as known by someone who has cut along it a thousand times, the radiologist's real expertise without its search template, the navigator's feel for the swell. The instruments' substitute for 理 is pattern detection at scale, the aggregation of many points. Aggregated pattern and grain are opposites. The first is seen from above across many oxen. The second is felt from inside one.

道 Dào: the way that is walked

Zhuangzi's second chapter contains one of the most compressed statements of process philosophy in any language: 道行之而成, the Way is made by walking it. The Dào is not a road laid down in advance and then travelled, still less a bearing given by a compass. It comes into being in the going, and it is made by those who go together. This is transformative teleology in six characters: a future under perpetual construction, made in the present by participants who cannot stand outside it.

A compass offers a bearing, and a bearing is a Dào laid down in advance. This is where the instruments come closest to Western means–end thinking, despite their Jullienesque vocabulary. "Set Your Course" is not 道行之而成. It is the course set before the walking, with the walking reduced to correction.

無爲 Wúwéi: neither natural nor stimulated

The WRAS vertical axis, natural versus stimulated emergence, meets its most elegant dissolution in 無爲 wúwéi, usually translated "non-action" and better understood as action that does not force. Wúwéi is not "just let it happen". The sage is not idle. Nor is it "stimulated emergence", the deliberate perturbation of a system to see what it does. It is a third thing the axis has no room for: action so closely fitted to the propensity of the situation that it does not register as intervention. Cook Ding's blade does not let the ox fall apart by itself, and it does not stimulate the ox. It moves where the ox already opens. The Laozi says the best rulers are those the people barely know exist, and that when the work is done the people say 我自然: we did it ourselves.

We did it ourselves is Flinn's recognition test stated two and a half millennia early. Direction is followed when it is experienced as one's own. The brochure's "authentic alignment" wants that result while keeping the steersman visible. The Laozi suggests the steersman has to become nearly invisible, and not through technique. It happens through a disposition of restraint that cannot be faked, because the people, like the wrasse's clients, are watching.

經 Jīng: the warp that is also management

One more character, given to us by WRAS's own first word. 經 jīng is the warp, the long threads fixed on the loom through which the weft passes. It also means a classic or canonical text (the threads that hold a tradition), and as a verb it means to pass through, to regulate, to lay out. Paired with 營 it gives 經營 jīngyíng, the ordinary modern word for management. Management in Chinese is literally warping and encamping, laying out the long threads and setting up the camp.

The compound has a classical source that could be a parable for this disputation. The Book of Odes describes King Wen laying out the Spirit Tower: 經始靈臺，經之營之, he began the Spirit Tower, he measured it and laid it out. The common people worked on it and it was finished within days. The king had said not to hurry, and the people came as children come to a parent: 經始勿亟，庶民子來. Mencius cites the ode (Mencius 1A2) to make one point. The people came because the king shared his joy with them. Warp was laid, and the weft wove itself, because those who carried it recognised the work as theirs.

The table ends where the instruments began, with weaving. The classical image of 經營 is a warp laid by a king and a weft woven by people who came because they wanted to, not because a survey had found them, aligned them or harvested them. If WRAS is to deserve its first word, it has to accept that the weft cannot be woven by the one who lays the warp.

The warp can be laid. The weft comes, or it does not.

Movement 19

The Tidal Loom

Much is worth keeping. From SenseMaker: self-signification, narrative at scale, and the link between pattern and story. From Estuarine: the energy cost of change, the honest identification of what is, for now, too costly to move, and the discipline of fine granularity. From WRAS: the weaving of risk with strategy, against the departmental split, and the recognition that complex domains are fat-tailed. From SwarmCompass: the insistence on tempo, days not quarters, and on looking to the periphery. These are real contributions and the reconstruction keeps them all.

What goes is the observer's privilege, the promise of mapped unknowns, the fantasy of alignment, and the assignment of risk to quadrants. I call the result the Tidal Loom. Tidal because, as in Snowden's estuary, the flow runs both ways: what goes in from the centre comes back changed, and the centre is changed by its return. Loom because, as 經營 suggests, a warp can be laid but the weft is the participants' own act.

Ten principles

1. No reader without a reading of the reader. Every deployment begins and ends with the convenors signifying their own experience on the same instrument. The sponsor's anxieties, hopes and blind spots are data, and they are published alongside everyone else's.
2. Declination before direction. Every signifier set comes with a declination table: what its designers expected to find, what they hoped to find, what the sponsor would prefer not to find, and which dimensions have no vertex. The table is revised after every round.
3. Frames, not seers. Do not identify "the 17%". Rotate the search template: change what people are asked to notice, so that different people see different gorillas. Blindness lives in the task, so change the task.
4. Quorum with inhibition. Take the bee's real lesson. Decide by threshold and not consensus, institutionalise stop signals (the formal right to say "wait, consider this"), then commit wholly. Do not seek alignment. Seek a decision people can live with while they still disagree.
5. Dated limits. Record every "cannot be changed" with a date and a condition, as in Prigogine. Revisit it whenever conditions move far from where they were when it was written.
6. Noticing, not mapping. Do not promise to map unknown unknowns. Measure and improve the latency of noticing: the time between something new becoming actual and someone in a position to respond recognising it.
7. Risk follows anxiety. For every probe, ask whose anxiety it creates, whose it relieves, and where the exposure lands on the power gradient. Make failure socially safe and not just financially small.
8. Return as a design requirement. What narrators give comes back to them as patterns, as decisions and as changed conditions, and it comes back before it goes anywhere else. Ostrom's users monitor. Boisot's cycle closes.
9. Consent to the harvest. No narrative becomes training data without the narrator's informed consent and a share in what is built from it. If the instrument automates, it says so, says what, and says who benefits.
10. Recognition, not alignment. The success test is Flinn's and the Laozi's, 我自然: do the people say we did this ourselves? If they say it was done to us, however well, the loom has become a compass.

Each row could be put into practice tomorrow, and none needs a new framework, a new patent or a new acronym. That is deliberate. On the account given here, the problem with the instruments is not a missing method. It is a stance that keeps rebuilding the outside. Changing the stance makes the existing methods, SenseMaker included, more faithful to the science they draw on.

You cannot sell the weft. You can only lay a warp worth weaving on.

Movement 20 · Propositio

The Instrument Proposed

Here is the proposal. A quantum uncertainty compass is a conceptual wayfinding instrument that replaces the fixed needle with a system governed by quantum mechanics and probability. It does not point to a single, rigid destination such as magnetic north. It moves through superposition, uncertainty and observer interaction to guide a traveller through physical or metaphorical space. It can be read three ways: as a real device in physics, as a design philosophy, and as a speculative or fictional artefact.

The proposal deserves to be taken seriously because it is the exact inverse of the SwarmCompass. The brochure's compass promises collapse: alignment, a set course, unknowns turned into mapped territory. The quantum compass promises to delay collapse, and to give up the pretence of knowing the destination in exchange for knowledge of the field. Where the first sells certainty about direction, the second offers disciplined uncertainty about it. Part VI asks how much of this is physics, how much is good metaphor, how much is bad metaphor, and how much is worth turning into a playable relic.

The fourth column is this movement's contribution. Where the proposal is right, it is kept. Where it is wrong, it is replaced with the true physics, which in each case is a stranger and richer image than the error. The no-signalling correction in particular, discussed in Movement 23, turns a telepathic gimmick into a parable of meaning made at the meeting.

Movement 21 · Physica

How Quantum Navigation Works Without GPS

This part is not speculative. Satellite navigation is now routinely attacked, and quantum navigation has left the laboratory. A joint submission to the International Civil Aviation Organization by Sweden and five neighbouring states reported that nearly 123,000 flights by 365 airlines suffered GNSS interference over the Baltic region between January and April 2025, more than 27% of flights in April alone. In the June 2025 Iran–Israel war an estimated 970 ships a day experienced GPS jamming in the Strait of Hormuz region. When the sky's signal can be jammed or spoofed, navigation has to turn to things the adversary cannot fake: inertia, the Earth's magnetic fingerprint, and its gravity.

Family one: the cold-atom inertial sensor

A classical inertial navigator integrates acceleration twice to get position. Every small bias in its accelerometers grows into a large position error over time, which is why inertial systems drift. A quantum accelerometer replaces the mechanical proof-mass with a cloud of atoms laser-cooled to near absolute zero. The atoms are in free fall, so they are perfect test masses. The laser pulses that probe them are a ruler marked off in wavelengths of light.

The working core is a light-pulse Mach–Zehnder atom interferometer. A first laser pulse (a "π/2 pulse") puts each atom into a superposition of two momentum states, so that each atom's wave-function travels two paths at once. After time T a "π pulse" swaps the paths like a mirror. After another T a final π/2 pulse recombines them. The two paths interfere, and how many atoms end in each output state depends on a phase φ ≈ k_eff·a·T², where a is the acceleration of the apparatus relative to the falling atoms. Rotation enters through the Sagnac effect on the enclosed area. The sensitivity grows with T², which is why laboratory fountains are tall and field devices are a compromise.

The UK has been unusually active here. Imperial College London's team demonstrated their first "quantum compass" prototype in 2018 and tested it aboard the Royal Navy research ship XV Patrick Blackett in May 2023. In February 2024 the system went to sea again in a container aboard MV Hurst Point, and in December 2025 a revised Quantum Inertial Navigation Sensing unit was trialled in the Arctic aboard MV Anvil Point, measuring accelerations and rotations from the wave-like behaviour of cold atoms. The practical advantage is not some magical knowledge of position. It is long-term stability: the atoms do not drift the way a phone's accelerometer does.

Family two: the Earth's magnetic fingerprint

The second family navigates by map-matching. Beneath the Earth's main magnetic field lies a crustal anomaly field of the order of hundreds of nanotesla, produced by the rocks below. It is unique from place to place, like a fingerprint. A sensitive magnetometer measures the total field, software subtracts the vehicle's own magnetic noise, and a filter fuses the result with inertial data, comparing the measured profile with a pre-surveyed anomaly map. The vehicle finds where it is by finding where the world looks like this.

The leading field results come from Q-CTRL's Ironstone Opal. In February 2025 the company flew more than 6,700 km in a Cessna 208B Grand Caravan near Griffith, Australia, with magnetometers mounted inside the aircraft, where avionics interference is more than ten times larger than at typical external mounts. The system outperformed a strategic-grade inertial navigation system by at least 11× in the air and up to 50× on the ground. Later reports cite accuracy up to 111× better than the best conventional alternative and positioning down to 4 metres over flights of up to 700 km. A companion gravimetric system ran for more than 144 hours aboard an Australian Navy vessel. The system is passive and emits nothing, so there is nothing to jam or spoof.

The magnetometers can themselves be quantum. In a nitrogen-vacancy (NV) centre, a nitrogen atom sits next to a missing carbon in a diamond lattice. Lit by a laser, the defect fluoresces more or less brightly depending on the surrounding magnetic field. Because the NV axis can lie along four crystallographic orientations in the tetrahedral lattice, one diamond yields vector magnetometry. A Warwick and Element Six team operated such a magnetometer in a moving van in 2024.

Corrections the physics requires

A quantum accelerometer does not "calculate exact position". It measures acceleration and rotation. Position is still integrated, and still drifts, only far more slowly. Map-matching systems infer position relative to a map. Neither family gives absolute position by quantum fiat.

The operative limit is not Heisenberg's in the popular sense. In atom interferometers the fundamental noise floor is quantum projection noise, the binomial noise of counting N uncorrelated atoms, which gives the standard quantum limit Δφ ≈ 1/√N. The Heisenberg limit, Δφ ≈ 1/N, is an ideal that can be approached only by entangling the atoms, for example in spin-squeezed states. The record of about 20 dB of squeezing in an ensemble of 10⁶ atoms produced the most precise relative phase measurement ever made with cold matter-waves. In fielded sensors the daily enemies are more mundane: vibration, dead time between measurements, and aliasing. Heisenberg is a horizon, not today's wall.

The "observer effect" is not the uncertainty principle. Popular accounts merge two ideas: that measurement disturbs, and that conjugate quantities cannot both be sharp. Barad devotes an appendix of Meeting the Universe Halfway to showing that the uncertainty principle is not the basis of Bohr's complementarity. For Bohr the point is not that looking disturbs a pre-existing state. It is that a property is defined only within a specific experimental arrangement. The difference is decisive for the metaphor in Movement 23.

The robin's eye: the first quantum compass

Long before the Royal Navy, the night-migratory songbird had a quantum compass. The leading hypothesis for avian magnetoreception is the radical-pair mechanism. Blue light striking a cryptochrome protein in the retina creates pairs of radicals, molecules with an unpaired electron, in a coherent spin state whose chemistry is sensitive to the direction of the Earth's field. In 2021 Xu, Mouritsen, Hore and colleagues showed in Nature that cryptochrome 4 from the European robin is magnetically sensitive in vitro, more so than the same protein from non-migratory chicken and pigeon, through four successive flavin–tryptophan radical pairs. The authors were careful: this was isolated protein in a laboratory, not proof that CRY4 is the sensor in the living bird.

Three features of the robin's compass bear on our instrument. It needs light: it is a compass that works only when it can see. It works by keeping a fragile quantum coherence alive in a warm, wet, noisy cell, so it resists decoherence well enough to be useful. And it gives the bird an orientation, not a destination. The robin does not know where Africa is. It knows how the field lines are inclined, and it migrates by staying faithful to that inclination night after night.

The oldest quantum compass is a bird's eye. It points nowhere, and it carries the robin to Africa.

Movement 22 · Deriva

The Drift Instrument: Psychogeography and the Probability Cloud

The second register of the proposal is speculative design and psychogeography, which Debord in 1955 called the study of the precise laws and specific effects of the geographical environment, consciously organised or not, on the emotions and behaviour of individuals. Here the compass is not meant to bring you home. It is meant to make you lost well.

Debord's dérive

Guy Debord's Theory of the Dérive (1956) defines the drift as a technique of rapid passage through varied ambiances. Two of its sentences could serve as a specification for the quantum compass. First: chance is a less important factor in this activity than one might think, because from a dérive point of view cities have psychogeographical contours, with constant currents, fixed points and vortexes that strongly discourage entry into or exit from certain zones. Second, the dérive includes both a letting-go and its necessary contradiction: the domination of psychogeographical variations by the knowledge and calculation of their possibilities. Debord also gives the drift's sociometry. One can drift alone, but the most fruitful arrangement is several small groups of two or three, and with more than four or five participants the specifically drifting character quickly fades. The average dérive lasts a day.

The proposed compass is an instrument for exactly this double movement. Its probability cloud is the letting-go: no bearing is imposed. Its lobes are the calculation: the currents and vortexes of the city given weights. Read through Debord, the "glowing haze indicating where meaningful paths are likely to be" is not mysticism. It is a psychogeographical contour map rendered as probability.

A real quantum drift-device, and its aura

The speculative instrument already has a crude real ancestor. Randonautica, launched in February 2020, claimed inspiration from chaos theory and Debord's dérive. It generated coordinates near the user from a quantum random-number source at the Australian National University and offered "attractors" (dense clusters of quantum points), "voids" (sparse regions) and "anomalies". The app invited users to "set an intention" before generating a point. Its own developers' material describes Attractor generation as resting on the claimed ability of the user's intention to influence random quantum processes.

The critique is straightforward. One sceptical review notes that only one of the app's entropy sources is quantum at all, and that a quantum random-number server simply produces a number like any other. No evidence supports intention bending vacuum fluctuations. The "quantum" in Randonautica is aura in Walter Benjamin's sense: a halo of authenticity attached to a mechanism that a die could reproduce. And yet the design works as Debord meant it to. It breaks habitual routes, sends people to places their routines avoid, and produces encounters. The lesson for the quantum compass is to keep the drift and discard the aura. Its value lies in disrupting habit, not in claiming that randomness is sacred.

Barad: the walker, the compass and the city as one phenomenon

The most serious philosophical ally for the drift-instrument is Karen Barad's agential realism. Barad is a theoretical physicist and develops it from Bohr's philosophy-physics in Meeting the Universe Halfway (2007). For Barad the basic units of reality are phenomena, the "ontological inseparability of intra-acting agencies". Her neologism intra-action, as opposed to interaction, means that the agencies do not exist as separate things before they meet. Their distinctness is made in the meeting, through what she calls an agential cut. Applied to the drift, the walker, the compass and the city are not three things, one of which reads the others. They are one phenomenon, and "direction" is something made within it.

This is the philosophical opposite of the SwarmCompass. There the client stands outside and reads the swarm. Barad places the reader inside the reading and makes the reading part of what is read. It is a quantum-flavoured statement of Mead's gesture–response and Elias's figuration from Movement 9.

The table does the work Movement 6 asked of the WRASSE emblem: it takes the source domain's whole natural history and checks what actually crosses. Six of the nine images pass as structural analogy, each testable in social terms without borrowing physics. Two fail. One, Wendt's identity thesis, is set aside as a question of ontology that this disputation does not need to settle.

Movement 23 · Quattuor Leges

Four Laws of the Uncertain Needle

The proposal names four mechanics: observer collapse, the Heisenberg trade-off, entangled wayfinding and decoherence. Each is examined here against the physics. Each is either kept, corrected or replaced, and each is translated into a law of practice that bears on the instruments this treatise has criticised.

First law: the watched needle never turns (Zeno)

The proposal says the needle locks only when the walker stops and looks, and otherwise stays in superposition. The physics offers something stronger. In 1977 Misra and Sudarshan showed that a quantum system measured frequently enough will, in the limit, stay in its initial state. They named this after Zeno's arrow, which cannot move because at every instant it is where it is. Itano, Heinzen, Bollinger and Wineland demonstrated the effect experimentally in 1990 with trapped beryllium ions, and it is the experiment most often cited as the demonstration. Frequent observation does not just collapse the needle. It freezes the evolution of what is observed. Physicists sometimes summarise it as "a system cannot change while you are watching it".

Second law: path or place (complementarity)

The proposal's "Heisenberg trade-off" says that moving fast blurs the destination, while standing still reveals the exact location and erases the path. As literal physics this is loose, because position–momentum uncertainty concerns conjugate variables of a quantum object, not a hiker's speed. As Bohrian complementarity it is admissible and beautiful. Some descriptions require mutually exclusive experimental arrangements, so one cannot have both in a single reading. Ingold gives the human version: the wayfarer knows along a path and the navigator knows from above between points. The law of practice: every reading chooses between knowing where and knowing how. A strategy instrument that claims both in one dashboard has confused two arrangements.

Third law: harmony appears only at the rendezvous (no-signalling)

The proposal's most attractive mechanic, two entangled compasses that let travellers move "in subtle, non-verbal harmony" because one's turn instantly shifts the other's weighting, is the one physics forbids. The no-communication theorem states that during measurement of an entangled state, one observer cannot transmit information to another by measuring their subsystem, regardless of distance. Each party's local results look random. Bob cannot tell whether Alice did anything at all. The correlation becomes visible only when the two compare their records over an ordinary, light-limited channel.