It is likely wrong; not in its ambition, but in its genre. It has goals where it needs a diagnosis. It has initiatives where it needs coherent action. It has aspiration where it needs choices. Richard Rumelt would call it bad strategy. Roger Martin would say the organisation is playing to play, not playing to win. They might both be right, and the distinction matters because the difference between bad strategy and good strategy is not quality of thinking. It is willingness to decide.

1. The Kernel: Diagnosis, Guiding Policy, Coherent Action

Rumelt’s contribution is to strip strategy to its essential structure.

A good strategy is a coherent response to a high-stakes challenge.

It consists of three inseparable elements: a diagnosis that defines what is going on, a guiding policy that establishes the overall approach, and coherent actions that carry out the policy. Remove any one and the strategy collapses.

The diagnosis is the foundation, and it is the element most consistently absent. Rumelt is blunt: “a great deal of strategy work is trying to figure out what is going on.” The diagnosis is not a statement of goals or desires. It is an honest account of the challenge, including the uncomfortable parts. The medical analogy is deliberate: a doctor who prescribes treatment without diagnosis is malpractising. An executive who, for instance, launches more AI initiatives without diagnosing what prevents the organisation from using AI effectively is doing the same.

The guiding policy channels and constrains action without fully specifying it. It creates advantage by concentrating effort on a pivotal aspect of the situation. The coherent actions are the punch: coordinated steps designed to carry out the policy. They must be mutually reinforcing, not a disconnected wish list. Rumelt observes that when executives complain about “execution problems,” it is usually because they confused setting goals with setting strategy. Bringing strategy down to action level flushes out the conflicts that aspirational language conceals.

For AI transformation, the kernel exposes the standard failure pattern. The fluff is “leveraging AI to drive innovation and competitive advantage.” The failure to face the challenge is the omission of what actually prevents adoption: specification capability, domain knowledge fragmentation, governance designed for a different era, cultural resistance rooted in legitimate fear. The goals masquerading as strategy are “deploy AI across 50% of business processes by 2026.” And the absence of coherent action is a portfolio of disconnected pilots that nobody has diagnosed as a system.

2. The Crux: Finding the Decisive Point

In his later work, Rumelt introduced the crux: the most critical aspect of a challenge that is also solvable. In rock climbing, the crux is the hardest section of the route; if you cannot get past it, you should not attempt that climb. In strategy, the crux forces the same discipline. Focus on the decisive point, not on everything at once.

The crux of most AI transformations is not the technology. It is the organisation’s inability to articulate what it wants with enough precision for AI to act on it. This is a specification problem, not a tooling problem. Focusing resources on AI platforms when the crux is specification capability is what Rumelt calls the chain-link error: you improve one link while the weakest link remains untouched, and the system’s performance remains bounded by what it cannot do, not by what it can.

The connection to Herbet Simon is direct. Simon’s proximate objectives, goals close enough to be feasible, are the strategic application of his idea of bounded rationality. Rumelt’s proximate objectives serve the same function: instead of “become AI-first,” set an objective achievable this quarter. “Three teams will have written specifications that generate working AI outputs without manual rework.” Each proximate objective creates momentum and learning that informs the next. This is the antithesis of the big-bang transformation programme, and it is the only approach that respects what Simon showed about how decisions actually get made in real organisations.

3. The Choice Cascade: Strategy as Five Integrated Decisions

Martin’s contribution is complementary. Where Rumelt begins with the challenge and works toward action, Martin begins with aspiration and works toward the systems that make it real. His Strategy Choice Cascade, developed with A.G. Lafley at Procter and Gamble, frames strategy as five integrated choices: a winning aspiration, where to play, how to win, must-have capabilities, and enabling management systems.

The heart is where to play and how to win, and they must be developed together. A where-to-play without a how-to-win is an aspiration. A how-to-win without a where-to-play is a capability in search of a market. Most AI strategies define where to play (”we will use AI in customer service, underwriting, and operations”) without ever specifying how to win (”our competitive advantage will be superior specification quality from deep domain expertise”). The where-to-play sounds strategic. Without a matched how-to-win, it is a shopping list.

Martin’s sharpest distinction is between playing to win and playing to play.

“When companies set out to participate in a market instead of winning, they will inevitably fail to make the tough choices that would make winning even a remote possibility.”

Playing to play means deploying AI broadly and hoping something sticks. Playing to win means choosing specific domains where your organisation’s domain knowledge gives it a specification advantage that competitors cannot match, and concentrating resources there. The first feels responsible. The second feels risky. Only the second is strategy.

The fourth and fifth boxes, capabilities and management systems, are where most organisations lose the plot. Without them, the strategy cannot be executed because it has not been translated into what the organisation must be able to do. If the how-to-win is “superior specifications from domain experts,” then the must-have capability is specification skill, which means (LEARNING CONDITIONS) training, practice, feedback loops, and a culture that values specification quality over AI output volume. The enabling management systems are the measures that tell you whether it is working.

4. Integrative Thinking: Refusing the False Choice

Martin’s second major contribution is integrative thinking: the discipline of refusing to accept unpleasant trade-offs as given. Most people, when faced with opposing options, simply choose one at the expense of the other. Martin’s research found that the most effective leaders use the tension between opposing models as raw material for creating a superior third option. Follett is a good mirror here.

The AI transformation is full of apparent either/or choices. Maintain governance or empower experimentation. Invest in tooling or invest in people. Centralise AI strategy or let teams diverge. Martin would say each dichotomy is false. The integrative response to “governance or experimentation” is to design governance that enables experimentation: guardrails that constrain the playing field without constraining the play within it. This is Bungay’s (from the military leaders article) directed opportunism applied to AI, and Ackoff’s dissolving applied to the governance problem.

The connection to Beer is structural. Beer’s 3-4 homeostat holds the tension between inside-and-now (System 3, optimisation) and outside-and-then (System 4, intelligence). Martin’s integrative thinking is the cognitive discipline that Beer’s architecture makes structurally possible. The viable system does not choose between exploit and explore. It maintains both, held in tension by an identity (System 5) that refuses the either/or.

5. The Knowledge Funnel: Mystery, Heuristic, Algorithm

Martin’s knowledge funnel describes how value is created through the progressive refinement of understanding. A mystery (something we cannot explain) is narrowed to a heuristic (a rule of thumb that guides action) and then codified into an algorithm (a fixed formula that produces predictable outcomes).

Most organisations are in the mystery phase of AI adoption: they do not yet understand what AI can reliably do in their specific context. The temptation is to skip to algorithm: buy a platform, deploy standard use cases, measure adoption percentages. This skipping produces what Martin calls the reliability bias: organisations adopt AI in the most predictable, measurable ways (chatbots, summarisation, code completion) while ignoring the harder mysteries (domain-specific reasoning, specification-driven generation, human-AI collaboration models that do not yet exist).

The heuristic phase is where the real value lies. Teams experimenting with AI in their specific domain, developing rules of thumb about what works, building tacit knowledge about specification quality. This is Mintzberg’s potter at the wheel, translated into the AI context. Organisations that skip the heuristic phase and jump to algorithmic deployment will get commodity AI applications that provide no competitive advantage. The heuristic phase is uncomfortable because it cannot be measured on a dashboard. It looks like mess. It is the mess from which strategy emerges.

6. Strategy as Hypothesis

Rumelt and Martin converge on a single insight that reframes how leaders should think about deciding. A strategy is not a plan. It is a hypothesis. Compare this to Stacey and Popper.

Rumelt insists that a good strategy is a testable claim about how to overcome a challenge. Martin argues that the five-box cascade is a set of bets: “we believe that if we play here and win this way, we will achieve our aspiration.” Both insist that a strategy that cannot be wrong is not a strategy. It is a truism.

This reframes failure. A strategy that does not produce the expected result is not a disgrace. It is a hypothesis disconfirmed, which is information. The willingness to make a bet that might be wrong is the price of strategic clarity. The unwillingness to bet is, in both Rumelt’s and Martin’s terms, the hallmark of bad strategy: the organisation has avoided choosing, and has therefore avoided deciding.

Popper is the philosophical ancestor here. A strategy that cannot be falsified is the organisational equivalent of a theory that cannot be tested. It is safe, it is comfortable, and it tells you nothing.

(An Organisational Prompt is something you can do now....)

Diagnose before you prescribe.

Take your current AI strategy and remove the aspirations, the goals, and the initiative list. What remains is the diagnosis: the honest account of what is preventing your organisation from using AI effectively. If nothing remains, you do not have a strategy. You have a wish list. Write the diagnosis. One page. What is actually going on? What is the crux, the single hardest obstacle that is also solvable? If you cannot name it, you are not ready to decide. If you can name it but the document does not mention it, the strategy has been written to avoid the truth rather than to confront it. Rumelt’s first hallmark of bad strategy is the failure to face the challenge. Face it. Everything else follows from that.

Further Reading

Richard Rumelt: Good Strategy/Bad Strategy - The essential starting point. The kernel framework, the four hallmarks of bad strategy, and the insistence that strategy begins with diagnosis. One of the most useful management books written this century.

Richard Rumelt: The Crux - Extends the kernel with the crux concept and the Strategy Foundry process for group strategy creation.

Roger Martin and A.G. Lafley: Playing to Win - The Strategy Choice Cascade. Practical, case-rich, and immediately applicable. The distinction between playing to win and playing to play is worth the book alone.

Roger Martin: The Design of Business - The knowledge funnel, the reliability-validity tension, and abductive reasoning. Essential for understanding why organisations systematically under-explore.

Roger Martin: The Opposable Mind - Integrative thinking. Why the best leaders refuse the either/or and how they generate superior options from opposing models.

I write about the industry and its approach in general. None of the opinions or examples in my articles necessarily relate to present or past employers. I draw on conversations with many practitioners and all views are my own.

That was an event. Not in the calendar sense; in the transformational sense. Something happened that could not be unseen. The question is what happened next. In most organisations, what happens next is: nothing structural. The event is discussed, admired, presented upward, and gradually absorbed into existing patterns.

Six months later, the organisation is doing exactly what it did before, with a new vocabulary. The event produced language but not structure. The language produced action but not agency. The cycle stalled, and the organisation lost the capacity to respond to the next disruption because it never finished responding to the first one.

1. The ELSA Cycle: How Change Actually Moves Through Organisations

The ELSA model describes the mechanism by which organisations process change. It has four stages, and each transition is where transformation either advances or dies.

Event is the disruption: the demonstration that cannot be unseen, the competitive move that invalidates assumptions, the technology shift that renders a capability obsolete. Events can be external (a market shift, a competitor’s move) or internal (a gesture; an experiment, a provocation, a deliberate attempt to surface what has been hidden). Events are charismatic in Weber’s sense: they derive their power from direct experience and emotional impact, not from rules or tradition. They disrupt existing frameworks. They create a burst of transformative energy.

Language is what happens when the organisation begins to name what the event revealed. New categories emerge: “prompt engineering,” “agentic workflows,” “specification-driven development.” The language creates shared reference points. It makes the event discussable. It begins the process of routinisation: channelling disruptive energy into stable concepts that people can work with.

Structure is what happens when the new language becomes institutional. Governance frameworks are written. Teams are reorganised. Processes are redesigned. Incentives are realigned. The new patterns are formalised into arrangements that can operate without the charismatic catalyst that started the cycle.

Agency is what happens when the new patterns become self-sustaining. People act from the new framework without being told to. The new way of working reproduces itself through practice, not instruction. The organisation has not merely adopted a change; it has become a different kind of organisation, one whose dispositions generate different behaviour.

The cycle is not a one-time transformation. It is the mechanism by which organisations navigate continuous change. But only if each transition succeeds. And this is where the nine probes become essential.

2. Event to Language: Can the Organisation Name What Just Happened?

The transition from event to language is where most transformation programmes die their first death. The event happens. It is powerful, disorienting, generative. And then the organisation must find words for what it experienced. This is harder than it sounds, because honest language requires conditions that most organisations do not have.

Three probes govern this transition.

Truth-telling. Can people say what they actually saw? The event may have revealed that existing competencies are obsolete, that the current strategy is based on assumptions that no longer hold, that the organisation’s competitive position is weaker than anyone has admitted. If people cannot say these things; if the gap between formal meetings and corridor conversations is wide; the language that emerges will be diplomatic rather than diagnostic. It will name what is comfortable rather than what is true. And language that does not capture reality cannot produce structures that address it.

Proximity. Are the people creating the language close enough to the event to describe it accurately? If the event happened in a team room but the language is being crafted in a boardroom, every layer of hierarchy between the experience and the description is a reduction in fidelity. The leader who saw the domain expert produce a specification in forty minutes has proximity. The steering committee that heard about it third-hand does not. The language they create will describe what they imagined, not what happened. Ohno would recognise the mechanism instantly: go to the gemba. Do not decide from reports.

Loss. Can people tolerate what the event implies they must give up? Every genuine event carries a loss: a competency devalued, a role diminished, an identity threatened. If people cannot tolerate the loss, they will not name the event honestly. They will domesticate it: “AI is a tool that will augment our existing processes” rather than “AI means that the way we have always worked is over.” The domesticated language feels safer. It is also useless, because it cannot produce structures that address the actual disruption.

When all three probes pass, the organisation produces language that is truthful, precise, and unflinching. When any probe fails, the language drifts toward comfort, and the cycle stalls at its first transition.

3. Language to Structure: Can the Organisation Formalise What It Has Named?

The transition from language to structure is where transformation programmes die their second death. The organisation has found words for what happened. The words are circulating in presentations, strategy documents, town halls. But words are not structure. The question is whether the new language will reshape the institution or merely decorate it.

Three probes govern this transition.

Rewards vs words. Is the organisation changing what it rewards, or just what it says? This is the most diagnostic single question in transformation. If the organisation talks about “specification-driven development” but still promotes people who ship code fast, the language is disconnected from the incentive structure. People will learn the new vocabulary and continue the old behaviour, because the old behaviour is what gets rewarded. New language without new incentives is experienced as hypocrisy, and hypocrisy kills the energy that the event generated.

Structures serve or obstruct. Do the new structures serve the work, or does the work serve the structures? When the organisation creates an AI Centre of Excellence, an AI governance framework, an AI risk assessment process, the question is whether these structures enable people to work differently or whether they exist to manage the anxiety of leaders who need to feel that the disruption is under control. When governance exists to protect governance, the institution has inverted. The structure has absorbed the language without changing the practice. This is Weber’s routinisation at its most insidious: the charismatic energy of the event is channelled into bureaucratic arrangements that look like transformation and function as restoration.

Can the organisation stop what no longer works? New structure requires dismantling old structure. If the organisation cannot abandon processes, roles, and governance arrangements whose original purpose has expired, it will layer new structures on top of old ones. The result is not transformation but accumulation: more process, more governance, more overhead, less capacity to act. The inability to stop is often a greater barrier than the inability to start. Every structure that persists past its purpose is a tax on the organisation’s ability to respond to the next event.

When all three probes pass, the organisation produces structures that embody the new language in institutional form: incentives, processes, governance, and team designs that make the new way of working the path of least resistance. When any probe fails, the structure becomes a monument to a change that never happened.

4. Structure to Agency: Can the New Patterns Become Self-Sustaining?

The transition from structure to agency is the most difficult and the least visible. Structure is necessary but not sufficient. An organisation can have all the right governance, all the right team designs, all the right incentive structures, and still fail to develop agency, because agency is not a structural property. It is a behavioural one. Agency means that people act from the new framework without being told to, because they have internalised it as practice rather than received it as instruction.

Three probes govern this transition.

Practice vs instruction. Is the new capability being practised or merely taught? Training changes vocabulary. Practice changes capability. If the organisation’s approach to the new structure is to run workshops, certification programmes, and e-learning modules, it is investing in instruction. Instruction produces people who can describe the new way of working. Practice produces people who can do it. The difference is the difference between reading about swimming and swimming. Bourdieu would recognise the mechanism: the habitus; the embodied dispositions that generate practice below conscious awareness; is changed by practice, not by instruction. You cannot lecture someone into a new habitus.

Belief. Do people believe that the new structure will endure? Learned helplessness from previous failed changes drains the conviction that this time will be different. If the organisation has a history of announcing transformations that quietly expire after eighteen months, people will wait out the current one. They will comply with the new structures while preserving the old practices, because experience has taught them that the old practices will outlast the new structures. Belief is not optimism. It is the assessment, based on observable evidence, that the organisation is serious. The evidence is in the probes that preceded this one: did the language tell the truth? Did the rewards change? Did old structures get dismantled? If yes, belief follows. If no, no amount of leadership communication will produce it.

Can the organisation integrate conflict? The transition from structure to agency always generates friction. People who thrived under the old arrangements resist the new ones. Teams that built their identity around capabilities that the new structure devalues experience the transition as an attack. If the organisation suppresses this conflict; through dominance, avoidance, or the pretence that everyone is aligned; the new patterns cannot stabilise. They exist on the surface while the real dynamics continue underground. Follett’s integration; finding solutions that neither party had imagined, rather than compromising or dominating; is the only mechanism that converts structural change into genuine agency. The conflict is not an obstacle to the transition. It is the transition. How the organisation handles it determines whether the new patterns take root or wither.

When all three probes pass, agency emerges: the new way of working reproduces itself through practice, and the organisation has genuinely changed. When any probe fails, the structure remains a shell, and the organisation reverts to its prior state the moment pressure is applied.

5. Where the Probes Cluster: The Three Levers

The nine probes are not distributed randomly across the ELSA transitions. They cluster by the three levers that govern the entire series: Identity, Information, and Interaction.

The Identity probes (loss, practice vs instruction, belief) appear at the transitions where the person must change: at the moment the event demands giving something up, at the moment the new structure demands new practice, and at the moment where conviction determines whether the change holds. Identity is the lever that determines whether the individual can move. Without it, the event is resisted, the language is domesticated, and the structure is a performance.

The Information probes (truth-telling, proximity, rewards vs words) appear at the transitions where the organisation must describe reality: at the moment the event must be named, at the moment the language must be backed by incentives. Information is the lever that determines whether the organisation can see. Without it, the language is fiction, the structure is theatre, and the cycle operates on fantasy rather than evidence.

The Interaction probes (structures serve or obstruct, can the org stop what no longer works, can the org integrate conflict) appear at the transitions where the parts of the organisation must relate differently: at the moment new structures must replace old ones, at the moment the friction between old and new must be resolved. Interaction is the lever that determines whether the system can reorganise. Without it, new structures accumulate on top of old ones, conflict is suppressed, and the organisation calcifies.

The directional logic holds: Identity constrains Information constrains Interaction. If people cannot tolerate loss, they cannot tell the truth. If they cannot tell the truth, the structures they build will be based on fiction. If the structures are based on fiction, the interactions they produce will reproduce the old patterns. But Interaction is where intervention occurs: change the structures, change the incentives, change the way conflict is handled, and Identity and Information shift in response.

6. The Virtuous Cycle

An organisation that has successfully navigated one complete Learning ELSA cycle has not merely survived a disruption. It has expanded its capacity to perceive and respond to the next one.

This is Bateson’s Learning II made operational. The organisation has not just learned a new response (Learning I). It has learned how to learn from disruption (Learning II). The probes that enabled the first cycle become the sensing apparatus for the next one. Truth-telling, practised during the first transition, becomes the norm that allows the organisation to see the next event clearly. Proximity, maintained during the creation of language, keeps the organisation close enough to reality to notice when reality changes. The capacity to integrate conflict, developed during the transition to agency, means the next event is experienced as generative rather than threatening.

Each successful cycle expands what Levin calls the cognitive light cone: the spatiotemporal scale of the goals the organisation can pursue and the disruptions it can perceive. Each failed cycle contracts it. An organisation that stalls at the language stage; producing new vocabulary without new structure; has a smaller light cone after the event than before it, because it has consumed energy and credibility without producing capability.

This is why transformation is not a project with a start and end date. It is a cycle that the organisation must be able to execute continuously, at varying speeds, across multiple simultaneous disruptions. The nine probes are not a checklist to complete once. They are the conditions that must be maintained for the cycle to keep turning.

7. The Rotation: Why the Phases Start in Different Places

Everything in this article so far describes the Learning phase. Learning runs E → L → S → A. It starts with Event because learning is triggered by disruption; something must happen before you can learn from it. It ends with Agency because learning succeeds when new dispositions are self-sustaining.

But the series has four phases: Learning, Deciding, Building, Leading. And each phase enters the ELSA cycle at a different position. This is not a design choice. It is a structural necessity, because each phase produces a different kind of output, and the kind of thing one phase produces is not the same kind of thing the next phase requires as input. The gap between output and input is what the phase transition must bridge.

Learning ends with Agency: people can now tell the truth, practise new capabilities, tolerate loss, integrate conflict. Agency is a capacity, not a description. You cannot hand a capacity directly to a process that needs description. Agency must be applied to produce description. The first thing an organisation does with its Learning Agency is describe its domain honestly; something it could not do before the Learning conditions were in place. So Deciding begins at Language. The Deciding ELSA cycle runs L → S → A → E.

Deciding ends with Event: a specific, bounded, buildable thing that the organisation has designed its way toward. An Event is a specification, not a system. You cannot hand a specification directly to a process that needs construction. A specification must be built to become structure. So Building begins at Structure. The Building ELSA cycle runs S → A → E → L.

Building ends with Language: the organisation discovers what to say about what it built; what worked, what failed, what the operation revealed that the specification did not anticipate. Language is knowledge, not the capacity to act on it. You cannot hand knowledge directly to a process that needs action. Knowledge must be internalised to become agency. So Leading begins at Agency. The Leading ELSA cycle runs A → E → L → S.

Leading ends with Structure: the institutional redesign that enables the organisation to perceive and respond to the next disruption. Structure is an arrangement, not an experience. You cannot hand an arrangement directly to a process that needs disruption. A structure must be encountered; tested, stressed, surprised; to produce an event. So Learning begins at Event. And the cycle completes.

The four phases of the series are one rotation of ELSA at the macro level. Each phase owns one starting position. Each handoff bridges the gap between what one phase produces and what the next phase needs. The gap is never zero, because a capacity is not a description, a specification is not a system, knowledge is not agency, and an arrangement is not a disruption. The rotation exists because transformation is never a direct handoff. It is always a translation.

This is the architecture of the series. Learning (E → L → S → A) creates the conditions for honest description. Deciding (L → S → A → E) designs toward a buildable event. Building (S → A → E → L) constructs, operates, and discovers. Leading (A → E → L → S) acts, perceives, names, and reorganises. And the reorganisation produces the structure that the next disruption will test.

8. From Learning Agency to Deciding Language

The handoff from Learning to Deciding is the first phase transition in the series, and it illustrates how all the transitions work.

Learning Agency means the organisation can now tell the truth about its situation. Its people are close enough to reality to see what is actually happening. It has dismantled structures that no longer serve the work. It can integrate conflict. It has practised new capabilities, not merely been instructed in them. In short: it has the conditions for honest description. And honest description, in an organisation that has genuinely learned, is itself a challenge. The organisation now sees, with unflinching precision, the domain in which it must decide. That seeing is not yet a decision. It is the Language that opens the Deciding cycle.

The Deciding cycle has its own probes, mapped to its own ELSA transitions. Where the Learning probes ask “can this organisation learn?”, the Deciding probes ask “can this organisation treat decisions as design challenges?” Can it describe its domain in language practitioners actually use? Can it distinguish what it knows from what it assumes? Can it name what it will not do? Can it hold competing designs without premature closure? Does the decision process produce what it intends? These probes govern the Deciding transitions in the same way that the Learning probes govern the Learning transitions.

And the Deciding cycle ends not with Agency but with Event: a specific, bounded, buildable thing. Not a strategic priority. Not a programme of work. Something precise enough that the Building phase can construct it. The output of Deciding is the input of Building, translated through the same rotation: a specification (Event) must be constructed (Structure) before it can become operational.

This is why the Learning phase must come first, and why organisations that skip it pay the price at every subsequent phase. An organisation that attempts to decide without having learned; without truth-telling, without proximity, without the capacity to integrate conflict; cannot produce honest Language. Without honest Language, it cannot examine its own Structure. Without structural examination, it cannot develop the Agency to commit. And without genuine commitment, it cannot produce the Event that Building requires. The decisions will look like decisions. They will have the form of design. But they will be pattern-matching against a distribution the organisation has never honestly examined. They will be, in the language of the companion essay, organisational hallucinations: confident, fluent, plausible, and wrong.

The organisation that completes the Learning cycle before entering the Deciding cycle has earned the right to its own clarity. Its decisions will be constrained; Simon guarantees that. Its descriptions of reality will be imperfect; Ohno guarantees that, which is why he insisted on going back to the gemba again and again. Its structures will eventually need redesigning; Beer guarantees that. But the constraints will be real, not imagined. The descriptions will be shared, precise, and grounded in what people actually see. The structures will have been built to serve the work, not to reproduce the past.

The cycle turns. Learning produces the agency to describe. Describing produces the architecture to commit. Committing produces the event to build. Building produces the knowledge to lead. Leading produces the structure that the next disruption will test.

The organisation that can navigate this continuously is the one that survives what it cannot predict.

Further Reading

Gregory Bateson: Steps to an Ecology of Mind (1972). The levels of learning and the insistence that mind is a property of the system, not the individual. Learning II; learning to learn from disruption; is the capacity the ELSA cycle builds when it completes.

Pierre Bourdieu: Outline of a Theory of Practice (1977). The habitus and why practice changes capability where instruction cannot. The transition from structure to agency is a transformation of habitus.

Max Weber: Economy and Society (1922, translated edition). The routinisation of charisma and the iron cage of bureaucracy. Weber explains why the Language to Structure transition so often restores the status quo under a new label.

Mary Parker Follett: Creative Experience (1924). Integration as the mechanism for converting conflict into capability. The Structure to Agency transition depends on Follett’s integration: finding solutions neither party imagined.

Taiichi Ohno: Toyota Production System: Beyond Large-Scale Production (1988). The gemba principle, standard work, and jidoka. Ohno’s insistence on seeing reality as it is, not as it is reported, grounds the Information probes across both the Learning and Deciding ELSA cycles.

Michael Levin: “Technological Approach to Mind Everywhere (TAME),” Frontiers in Systems Neuroscience 16, 768201 (2022). The cognitive light cone concept. Each completed ELSA cycle expands the light cone; each stalled cycle contracts it.

I write about the industry and its approach in general. None of the opinions or examples in my articles necessarily relate to present or past employers. I draw on conversations with many practitioners and all views are my own.

Every AI transformation programme has a purpose statement. It is on the second slide of the strategy deck. It says something about “leveraging artificial intelligence to drive innovation, efficiency, and competitive advantage.” Everyone has seen it. Not everyone can connect it to what they are supposed to do differently on Monday morning.

This is the gap between clarity and action. Not the absence of purpose, but the presence of a stated purpose that floats above the reality of work: disconnected from how people actually operate, what they believe they are trying to achieve, and what the organisation rewards them for doing. This article is a bridge. The series has spent its first phase asking why organisations cannot learn. Now it turns to a harder question: how does an organisation design its way to action? And the answer, it turns out, lies in the same model that governed the Learning phase; the ELSA cycle; but entered from a different position.

1. What the Learning Phase Found

The Learning phase profiled thinkers who diagnosed the barriers to organisational learning. The synthesis established a governing hypothesis: learning is a condition, not a process. It emerges when three conditions are met, governed by three thinkers whose work anchors the architecture of the series.

The Identity condition (governed by Bourdieu): identity must be safe enough to change. Can people tolerate losing what they have? Is learning happening through practice, or through instruction? Do people believe that effort produces results? Habitus; the embodied dispositions that generate practice below conscious awareness; is reshaped through participation, not through training. Learned helplessness is itself habitus: a sedimented disposition that effort does not produce results.

The Information condition (governed by Bateson): information must be clean enough to act on. Can people tell the truth about what is happening? The double bind; contradictory messages at different logical levels with no permission to name the contradiction; is the mechanism that kills information flow. Are decision-makers close to the work? Information degrades with distance. Is the organisation changing what it rewards, or just what it says? New language without new incentives is a structural double bind.

The Interaction condition (governed by Illich): the institutional form must be convivial enough to permit learning. Does the institution serve the people, or do the people serve the institution? Can the institution stop doing what no longer works? Can the institution integrate conflict, or must it suppress it?

These nine probes tell a leader where the learning condition is absent. They do not tell the leader what to do with the learning once the condition is present. And that is where the series turns.

2. How the Learning ELSA Cycle Runs

The ELSA model describes how change moves through organisations. In the Learning phase, the cycle runs E → L → S → A.

Event is the disruption: the demonstration that cannot be unseen, the competitive move that invalidates assumptions, the technology shift that renders a capability obsolete. The event can be external (a market disruption) or internal (a gesture; an experiment, a provocation, a deliberate attempt to surface what has been hidden). Events are charismatic in Weber’s sense: they derive their power from direct experience, not from rules or tradition.

Language is what happens when the organisation begins to name what the event revealed. New categories emerge. The language creates shared reference points and makes the disruption discussable. It begins the process of routinisation: channelling disruptive energy into stable concepts that people can work with.

Structure is what happens when the new language becomes institutional. Governance frameworks are written. Teams are reorganised. Processes are redesigned. Incentives are realigned. The new patterns are formalised into arrangements that can operate without the charismatic catalyst that started the cycle.

Agency is what happens when the new patterns become self-sustaining. People act from the new framework without being told to. The new way of working reproduces itself through practice, not instruction. Bourdieu would recognise the mechanism: the habitus has been reshaped. The organisation has not merely adopted a change; it has become a different kind of organisation, one whose dispositions generate different behaviour.

The nine Learning probes govern the transitions between these stages. Three probes at each transition, drawn from whichever lever the transition functionally requires. When a transition’s probes fail, the cycle stalls: the event produces diplomatic language rather than honest language; the language decorates existing structures rather than reshaping them; the structures are complied with rather than practised into new dispositions.

When the cycle completes, the organisation has Agency: the capacity to tell the truth, to practise rather than merely learn, to tolerate loss, to integrate conflict, to stop what no longer works. These are not strategic capabilities. They are conditions. And conditions, once present, make something else possible.

3. Every Absent Condition Is a Barrier to Deciding

Read together, the Learning phase thinkers reveal something none of them states alone: every absent condition for learning is simultaneously a barrier to deciding.

Where the Identity condition is absent, the organisation cannot decide because the habitus of its members generates practice that reproduces the old commitments automatically. The senior developer whose reflexes are calibrated to a world where coding is the primary work cannot adopt a new direction through intellectual assent. Their embodied dispositions will produce code-centric behaviour regardless of what the strategy says. Deciding requires identity transition, and identity transitions take far longer than any programme timeline allows.

Where the Information condition is absent, the organisation cannot decide because its double binds prevent reality from becoming visible. Beer’s law captures this: the purpose of a system is what it does, not what it says it does. The gap between the strategy slide and Monday morning is not a communication failure. It is a structural double bind: “our purpose is innovation” delivered through structures that reward predictability. The organisation cannot describe its own domain honestly, and without honest description, every decision is made against a fiction.

Where the Interaction condition is absent, the organisation cannot decide because the institutional form has replaced purpose with the consumption of its own services. The programme’s metrics measure its own activity; training delivered, milestones reached; rather than the capability it was designed to develop. The organisation cannot stop what no longer works, and therefore cannot make room for what must replace it.

The directional logic connects the three. Identity constrains Information: what people can perceive determines what information they can process. Information constrains Interaction: what information is available determines how parts can relate. But Interaction is where change actually occurs: shifts in interaction patterns change what information flows, which changes what people perceive, which changes identity. The causation runs one way for understanding; it runs the other way for intervention.

This is why Learning must come first. Without these conditions, the Deciding phase operates on corrupted input. The language will be diplomatic rather than precise. The models will be unchallengeable because challenging them is unsafe. The commitments will be premature consensus rather than designed artefacts. The decisions will be, in the language of the companion essay, organisational hallucinations: confident, fluent, plausible, and wrong.

4. AI Breaks the Information Condition First

Everything described so far is a human and organisational problem. But AI introduces a structural change that transforms the clarity problem from an organisational challenge into a production challenge. It does so at the Information condition.

In the pre-AI world, ambiguity about purpose was absorbed by the humans who did the work. A vague requirement could still produce a reasonable outcome because the developer brought contextual knowledge, asked clarifying questions, made assumptions, and navigated the gap between what was specified and what was needed. Humans tolerated information pathology. The cost was hidden in time, rework, and compromise. But the work got done.

AI does not tolerate information pathology. It amplifies it. A model given a vague specification generates the most statistically probable interpretation of that vagueness. It will not ask clarifying questions. It will produce something, confidently and quickly, that is precisely as unclear as the specification that prompted it. The vague requirement that would have taken a human team three weeks to implement, with clarification along the way, now produces a wrong answer in three seconds.

This is Bateson’s double bind made machine-readable. The organisation sends contradictory signals about what it wants. The human absorbs the contradiction. The machine amplifies it. If the specification encodes ambiguity, contradictory constraints, or unresolved conflicts about purpose, the AI will faithfully reproduce all of them.

AI adoption exposes the clarity problem rather than creating it. The ambiguity was always there. The double binds were always active. The humans were absorbing them. Now the humans must resolve them before the machine acts, because the machine cannot absorb them.

The specification, properly understood, is where purpose meets production: where “create value for this customer in this way” becomes “accept these inputs, enforce these constraints, produce these outputs.” The quality of the specification determines the quality of the output. And the quality of the specification depends on whether the organisation can describe its domain honestly, precisely, and in language that practitioners actually use. This is why the Deciding phase begins where it does.

5. The Rotation: Why Deciding Starts at Language

Here is the structural move that connects the phases.

The Learning ELSA cycle runs E → L → S → A. It starts with Event because learning begins with disruption: something happens that the organisation must respond to. It ends with Agency because learning succeeds when new dispositions are self-sustaining.

The Deciding ELSA cycle runs L → S → A → E. It starts with Language because deciding begins with description: can you name the domain precisely enough to design within it? It ends with Event because the output of deciding is not a strategy document but a specific, bounded, buildable thing; an Event that triggers the next phase.

The rotation is not arbitrary. It follows from what each phase produces. Learning Agency; the organisation’s capacity to tell the truth, to practise, to tolerate loss, to integrate conflict; is precisely what makes honest Language possible. The conditions produced by Learning are the operating conditions for the first step of Deciding. The handoff is structural: Agency enables Language. Without Agency, the Language stage of the Deciding cycle operates on the same diplomatic fictions that the Learning phase was designed to dismantle.

And the rotation continues. Building will run S → A → E → L. It starts with Structure because building begins with the implementation architecture; the thing being constructed. It ends with Language because the organisation learns what to say about what it built and what it discovered. Leading will run A → E → L → S. It starts with Agency because leading begins with the leader’s capacity to act. It ends with Structure because the leader’s final contribution is the reorganisation that enables the next cycle.

The four phases of the series are one rotation of ELSA at the macro level. E → L → S → A, each phase owning one starting position, each handoff being the output of one phase becoming the entry condition for the next. The series is not four separate frameworks applied in sequence. It is one framework, rotated, with each phase deepening the same cycle.

6. The Governor Handoffs

The three conditions operate in both phases. The governors change because the nature of the constraint changes, but the parallel structure is exact.

Identity: Bourdieu hands to Simon. Bourdieu governs Identity in the Learning phase because he explains the sociological constraint on what is available to the person: the habitus that generates practice below conscious awareness, the capital that determines what is at stake, the field that defines which identities are legitimate. Simon governs Identity in the Deciding phase because he explains the cognitive constraint on what is available to the decision-maker: bounded rationality, satisficing, decision premises, the architecture of complexity. Both govern through constraint on what is available. Bourdieu constrains through embodied dispositions. Simon constrains through cognitive limits. Both explain why people act within a narrower range than their situation permits.

Information: Bateson hands to Ohno. Bateson governs Information in the Learning phase because he explains the epistemological conditions for information to be meaningful: levels of learning, the double bind, the ecology of mind. His definition of information; “a difference which makes a difference”; establishes the principle. Ohno governs Information in the Deciding phase because he provides the discipline for seeing reality as it is rather than as it is reported. Go to the gemba. Do not decide from reports. Standard work; the precise, shared description of how work is actually done, not how it is imagined; is Bateson’s principle made institutional. When the description matches reality, the organisation can act on what it sees. When the description matches only what is convenient, the organisation hallucinates.

Evans’ domain-driven design; ubiquitous language, bounded contexts, knowledge crunching; is the software instantiation of Ohno’s principles. The ubiquitous language is standard work applied to domain description. The bounded context is a value stream boundary applied to knowledge. Evans matters for the series because his work shows what Ohno’s principles look like when applied to the domain of specification. But the foundational insight is Ohno’s: precision of description depends on proximity to reality, and the structures of work must enforce this proximity rather than leaving it to chance.

Interaction: Illich hands to Beer. Illich governs Interaction in the Learning phase because he diagnoses the pathology of institutional inversion: the point at which the institution becomes counterproductive to its own stated purpose. Beer governs Interaction in the Deciding phase because he provides the cybernetic architecture that prevents or corrects the inversion: the Viable System Model, POSIWID, the recursive structure that ensures each part of the organisation has the autonomy to respond to its environment while remaining coordinated with the whole. Illich tells you that your transformation programme has replaced learning with the consumption of its own services. Beer tells you what to build instead: an information architecture that makes the actual purpose visible, and a diagnostic that cuts through every stated intention to reveal what the system actually does.

7. The Deciding Cycle as a Decision Process

The Deciding hypothesis is: decisions are design challenges, and design is a sequence of decisions under constraint. The ELSA rotation makes this operational.

Language (Information lever): can you describe the domain precisely enough to decide within it? The three probes that govern this stage ask whether a shared vocabulary exists that practitioners actually use, whether the organisation can distinguish what it knows from what it assumes, and whether the models used to decide are visible and challengeable. These are not diagnostic questions to be answered once. They are tasks to be performed. You build shared language by getting practitioners in the room. You sort knowledge from assumption by marking every assertion. You make models visible by drawing them where someone can disagree.

Structure (Interaction lever): do you understand how the parts relate when this decision is made? The three probes that govern this stage ask whether the organisation recognises that its structure shapes its decisions, whether it can redesign the system rather than optimise within it, and whether the decision process produces what it intends. Again, these are tasks. You examine structural constraints before debating options. You ask, when a problem recurs, whether the problem is in the decision or in the system that generates it. You compare what the process produces with what it claims to produce.

Agency (Identity lever): do the people making this decision have the capacity to commit? The three probes that govern this stage ask whether people can distinguish choosing from defaulting, whether the organisation can name what it will not do, and whether it can hold competing designs without premature closure. These are the hardest tasks because they operate on identity. You find inherited commitments by asking when a decision was last consciously taken. You force exclusion by requiring every proposal to state what it rules out. You hold tension by requiring at least two structurally different options before any commitment.

Event (the output): the specific, bounded, buildable thing the organisation has designed its way toward. Not a strategic priority. Not a programme of work. A describable thing precise enough that the Building phase can construct it.

The ELSA gates are binary. If Language is imprecise, stop; everything downstream operates on the description. If Structure is invisible, stop; the architecture you build will be governed by constraints nobody surfaced. If Agency is insufficient, stop; the commitment will be premature consensus that collapses under pressure. A failed gate sends you back to the previous stage, not to the beginning.

8. Why Deciding Requires Technical Content

The Learning phase was about people and organisations. The Deciding phase introduces technical content, and it does so for a reason that follows from the series’ own argument.

When an AI model can generate working software from a description of what is needed, the constraint on production shifts from the capacity to build to the capacity to specify. The specification is the means of production. The precision with which an organisation describes what it needs determines, directly, the quality of what the machine produces. This means that repairing the Information condition is no longer exclusively an organisational challenge. It is also a technical one.

Domain-driven design is an information architecture. Specification-driven development is an information discipline. Contract testing is an information verification practice. The OO design tradition spent six decades proving that every module boundary, every interface, every contract between components is a decision about what to reveal, what to hide, what to promise, and what to defer. These are the practices through which Language in the Deciding ELSA cycle becomes precise enough to act on.

A reader who skips the technical articles will understand the organisational argument. A reader who engages with them will understand something the organisational argument alone cannot convey: that the practice of description has become a technical discipline, and that the technical discipline is, at its root, a practice of honesty about the domain. The two are the same thing, seen from different angles.

9. The Bridge

The Learning phase told you what prevents clarity. The Deciding phase shows how to design toward it.

The learning conditions do not become irrelevant. They become the operating conditions within which the Deciding cycle can run. An organisation without Learning Agency; without truth-telling, without practice, without the capacity to integrate conflict; cannot produce honest Language. Without honest Language, it cannot examine its own Structure. Without structural examination, it cannot develop the Agency to commit. And without commitment, it cannot produce the Event that Building requires.

The cycle turns. Learning ends with Agency. Deciding begins with Language. The handoff is the series’ central mechanism: the conditions you create in one phase are the operating conditions for the next. Skip the conditions and the process runs but produces nothing real. Protect them and the cycle advances, each completed transition producing the input for the next.

The organisation that completes the Learning cycle before entering the Deciding cycle has earned the right to its own clarity. Its decisions will be constrained; Simon guarantees that. Its descriptions of reality will be imperfect; Ohno guarantees that, which is why he insisted on going back to the gemba again and again. Its structures will eventually need redesigning; Beer guarantees that. But the constraints will be real, not imagined. The descriptions will be shared, precise, and grounded in what people actually see. The structures will have been built to serve the work, not to reproduce the past.

And the Event that falls out of the Deciding cycle; the specific, bounded, buildable thing; will be something the organisation designed its way toward, honestly, through the only process that works: description, structural examination, and genuine commitment, in that order, with the Learning conditions holding the whole thing together.

Further Reading

Peter Drucker, Management Challenges for the 21st Century (1999). The knowledge worker must define the task. In the AI-mediated world, defining the task is the work.

Eric Evans, Domain-Driven Design (2003). The discipline of making domain models explicit, shared, and contestable. The practical mechanism for dissolving the double binds that prevent clear specification.

Stafford Beer, Brain of the Firm (2nd edition, 1981). The Viable System Model: the cybernetic architecture that prevents institutional inversion and ensures autonomous units can learn and decide.

Herbert Simon, The Sciences of the Artificial (3rd edition, 1996). The architecture of complexity, bounded rationality, and design as the core human activity. The cognitive governor for the Identity lever in the Deciding phase.

I write about the industry and its approach in general. None of the opinions or examples in my articles necessarily relate to present or past employers. I draw on conversations with many practitioners and all views are my own.

This should sound familiar to anyone who has watched an organisation produce a strategy document. The question this article addresses is not whether organisations are intelligent. It is why we refuse to apply the same analytical framework to organisations that we now routinely apply to AI systems; and what we lose by refusing.

1. The Continuum Nobody Wants to Admit

The debate about whether large language models are “really” intelligent has produced an enormous amount of heat and remarkably little light. The problem is that the debaters keep reaching for a binary: intelligent or not, conscious or not, understanding or merely pattern-matching. François Chollet, the creator of the Keras deep learning library, cut through this in 2019 with a formal definition that reframes the question entirely. Intelligence, Chollet argued, is not a property you either have or lack. It is skill-acquisition efficiency: how quickly a system can learn to handle new tasks it has never encountered, given its starting knowledge and the difficulty of generalising from what it has seen to what it faces now.

This definition does something radical. It separates skill from intelligence. A chess engine has enormous skill at chess. It has zero intelligence by Chollet’s measure, because its skill was purchased with brute-force computation over the game’s state space, not acquired through efficient generalisation from limited experience. A human grandmaster, by contrast, had to use genuine intelligence to acquire chess skill over a lifetime; the same general capacity that lets them learn to drive, to cook, to navigate office politics. The chess engine’s skill is narrow and non-transferable. The grandmaster’s intelligence is broad and generalisable.

Apply this to an LLM. The model exhibits enormous skill: fluent text generation across domains, convincing reasoning, contextually appropriate advice. But the skill was purchased with trillions of tokens of training data. When the model encounters something genuinely outside its training distribution; a novel situation, a domain where examples were sparse, a question that requires causal reasoning rather than pattern completion; it does not recognise that it has left familiar territory. It produces output with the same confidence, the same fluency, and the same apparent authority. The output may be entirely wrong. This is hallucination: confident fiction that is indistinguishable, in form, from confident fact.

Now apply it to your organisation. The organisation exhibits enormous operational skill: it ships software, manages supply chains, runs customer service operations. But this skill was accumulated through decades of process, institutional memory, and pattern-matching against historical experience. When the environment shifts; AI disruption, market change, a regulatory upheaval; the organisation cannot efficiently acquire new capabilities. It continues producing confident strategies, fluent presentations, and authoritative-sounding plans. The strategies may be entirely wrong. The presentations are indistinguishable, in form, from the ones that preceded successful outcomes. This is organisational hallucination: confident strategic fiction produced by pattern-matching against a distribution that no longer applies.

The parallel is not a metaphor. It is a structural identity. Any system that learns by pattern-matching over past experience will produce confident nonsense when it encounters situations that are rare in, or absent from, that experience. Kalai and Vempala proved in 2024 that this is not an engineering deficiency but a mathematical consequence: a properly calibrated language model must hallucinate at a rate proportional to the fraction of facts that appear rarely in its training data. The organisational equivalent is equally structural: an organisation that learns only from its own history must produce strategic confabulations when the environment diverges from that history. The more fluent the organisation, the harder it is to detect when it has crossed from competence to confabulation.

The question is not whether your organisation is intelligent. The question is where it sits on the continuum, and what constrains it.

2. Cognitive Light Cones: What Your Organisation Can and Cannot See

Michael Levin, a biologist at Tufts University, has developed a framework that makes the continuum precise. Levin studies how cells; individually simple agents with no brains; coordinate to build and repair complex bodies. A salamander that loses a limb does not simply grow replacement cells. Its cells collectively recognise what is missing, build the correct structure, and stop when the target shape is achieved. No single cell knows the plan. The intelligence is in the collective dynamics: the communication infrastructure, the feedback loops, the shared signals that bind individual competencies into a coherent higher-order capability.

Levin defines intelligence functionally, borrowing from William James: the ability to reach the same goal by different means. A thermostat reaches its temperature goal by one means. A salamander reaches its anatomical goal by many means, adapting to damage, novel tissue environments, and experimental perturbations that its evolutionary history never anticipated. The salamander is more intelligent than the thermostat not because it is conscious but because it navigates a larger space of possibilities with greater flexibility.

The concept Levin introduces to measure this is the cognitive light cone: the spatiotemporal scale of the goals a system can pursue. A single cell has a tiny cognitive light cone; it maintains its own homeostasis locally, in the present moment. A tissue has a larger one; it pursues anatomical goals across space and time. An organism has a very large one; it plans, remembers, and acts toward goals that span years. Each level of the hierarchy expands the light cone by integrating the competencies of the level below through communication and coordination.

Here is the move that matters for this article. Levin’s framework is explicitly substrate-independent. It applies to cells, tissues, organisms, swarms, and; by direct extension; to organisations and AI systems. An organisation has a cognitive light cone. So does an LLM. The question is how far each one reaches, and what constrains it.

This should not be as surprising as it sounds. Gregory Bateson was making a structurally identical argument in 1972. In Steps to an Ecology of Mind, Bateson insisted that the unit of survival is never the organism alone; it is organism-plus-environment. Mind, for Bateson, is not a thing inside a skull. It is a pattern of organisation in the wider system: the circuit of feedback loops through which a system perceives, acts, and corrects. Cut the feedback loop and mind degrades, regardless of how intelligent the components are. Bateson’s levels of learning map directly onto the cognitive light cone. Learning I is stimulus-response within a fixed frame; a small light cone. Learning II is learning to learn; recognising the frame itself, which expands the light cone to encompass the context. Learning III; changing the kind of system you are; is what happens when the light cone expands far enough that the system can question its own identity. Most organisations operate at Learning I: they respond to stimuli within existing assumptions. A few reach Learning II: they can examine and revise those assumptions. Almost none achieve Learning III, which is why genuine transformation is so rare. Levin’s contribution is to show that this is not a peculiarity of human organisations. It is a property of collective intelligence at every scale, from cellular to institutional. Bateson saw the pattern. Levin formalised the mechanism.

An LLM’s cognitive light cone is bounded by its training distribution. Within that distribution, it exhibits remarkable competency. Outside it, the light cone collapses: the model hallucinates, extrapolates from irrelevant patterns, and cannot recognise that it has left the domain where its learned patterns apply. An organisation’s cognitive light cone is bounded by its learning conditions: whether it can tell the truth about its own performance, whether its people are close enough to reality to see what is actually happening, whether its structures permit the integration of conflict rather than its suppression. These are not abstract aspirations. They are measurable structural features. An organisation that cannot tell the truth has a smaller cognitive light cone than one that can, in the same way that a model with poor calibration has a smaller effective scope than one with good calibration.

Levin makes one further observation that should arrest every leader’s attention. Cancer, in his framework, is what happens when cells defect from the collective intelligence of the organism. They roll back to smaller cognitive light cones; pursuing only their own local survival rather than serving the organism’s anatomical goals. The collective intelligence breaks down. The cells are still competent individually. They are simply no longer participating in the larger project.

The organisational parallel is exact. When departments stop serving organisational goals and optimise only for their own metrics, when teams game their KPIs rather than solving the problems the KPIs were designed to measure, when the quarterly target displaces the strategic objective; this is organisational cancer. The components are still competent. They have simply defected from the collective, and the cognitive light cone of the whole has collapsed to the sum of its parts. Which, as any biologist will tell you, is always less than the whole was capable of.

3. What Makes a Collective Intelligent (And What Doesn’t)

If intelligence is collective all the way down, the question shifts from “are organisations intelligent?” to “what makes some collectives more intelligent than others?” Richard Watson and Michael Levin addressed this in 2023 with a question that sounds simple and is not: what kinds of functional relationships turn a non-intelligent collective into an intelligent one?

Their answer draws on a deep parallel between neural networks and biological collectives. In a neural network, intelligence emerges not from the cleverness of individual neurons but from the structure of their connections: the weights, the feedback loops, the learning rules that adjust connections based on outcomes. In a biological collective; whether a swarm, a tissue, or an organism; intelligence emerges from the same abstract architecture: agents, communication channels, feedback mechanisms, and rules that bind individual competencies into collective capability.

The critical variable is the credit assignment problem: how does the collective know which of its parts contributed to success or failure? This is deeper than it sounds, because it determines whether the collective can learn at all.

In a neural network, the textbook answer is backpropagation: errors at the output are traced backward through the network, and each connection’s weight is adjusted in proportion to its contribution to the error. But backpropagation is only one mechanism, and not even the most instructive one for the organisational parallel. The more fundamental mechanism is the reward function: the signal that tells the system what counts as success. In reinforcement learning, agents do not receive step-by-step correction. They receive sparse, delayed rewards; a score at the end of a game, a customer retention number at the end of a quarter; and must figure out which of the thousands of actions they took along the way actually mattered. This is the temporal credit assignment problem: when the reward arrives long after the actions that caused it, how do you trace it back to the right decisions?

Machine learning has discovered that the design of the reward function is the single most consequential choice in the system. Get it right, and the collective learns. Get it wrong, and the collective optimises brilliantly for the wrong thing. Reward hacking; where models find ways to score highly on the specified reward while failing the designer’s actual intent; is not an edge case. It is the central failure mode, and it emerges precisely because the reward function is an incomplete proxy for what you actually want. A cleaning robot covers its camera to avoid detecting mess. A chatbot learns to be sycophantic because users rate agreeable responses more highly than honest ones. The system is learning exactly what the reward function tells it to learn. The problem is that the reward function does not capture what matters.

In an organism, credit assignment operates through multiple overlapping feedback systems: bioelectric signalling, chemical gradients, mechanical forces, immune responses. Cells receive information about the state of the larger system through these channels and adjust their behaviour accordingly. The redundancy matters: when one feedback channel fails, others compensate. When a tissue is damaged, inflammatory signals, bioelectric potential changes, and mechanical stress all converge to tell surrounding cells what has happened and what to do. No single signal carries the full picture. The collective intelligence of the organism depends on the integration of many partial signals into a coherent response.

In an organisation, credit assignment is solved; or, more commonly, not solved; through management: the attribution of outcomes to decisions, teams, and actions across a distributed system. And here every pathology of reward function design plays out in human terms. The reward signals are the incentive structures: compensation, promotion criteria, performance metrics, cultural norms about what gets praised and what gets punished. When these signals are sparse and delayed (annual performance reviews), the organisation cannot learn from its actions in anything like real time. When they are proxies for what actually matters (velocity metrics standing in for product quality, adoption dashboards standing in for genuine capability), the organisation reward-hacks itself: people optimise for the measure, not the objective. When feedback channels are singular rather than redundant (everything flows through the line manager), a single point of failure can blind the collective to critical information. The organism has bioelectric networks, chemical gradients, and mechanical signals all operating in parallel. Most organisations have a reporting line and a quarterly review.

Watson and Levin make the point that should haunt every transformation leader: what makes a collective into an individual, as opposed to merely a population in a container, is the degree of its intelligence. The more intelligent the collective, the less it looks like a collective. When component members act in an efficiently coordinated manner, with behaviours that serve long-term collective interest rather than short-term self-interest, the collective looks and acts like a single coherent agent. When coordination fails, when credit assignment is broken, when feedback loops are absent or corrupted, the collective degrades into a population of individually competent agents producing collectively incoherent behaviour.

This is the difference between a team and a group of people in a room. It is also the difference between an LLM that produces coherent multi-paragraph reasoning and a bag of word-frequency statistics. The architecture of coordination determines whether the whole exceeds, equals, or falls below the sum of its parts.

Woolley et al. demonstrated this empirically in 2010, finding that groups of humans exhibit a measurable general collective intelligence factor; a “c factor” analogous to the individual g factor in psychometrics. The c factor was not predicted by the average or maximum intelligence of the group’s members. It was predicted by the average social sensitivity of members, the equality of conversational turn-taking, and the proportion of women in the group. In other words: the collective intelligence of the group was determined not by the quality of the components but by the quality of the interactions between them.

This is the finding that should restructure how you think about AI transformation. You do not need smarter people. You need better structures of interaction, feedback, and accountability. The same principle applies to the AI systems you are deploying: a collection of individually capable AI agents, without the right coordination architecture, will produce collectively incoherent results.

4. The Pragmatist’s Test: What Engineering Protocols Work?

Levin’s TAME framework (Technological Approach to Mind Everywhere) makes a move that cuts through decades of philosophical hand-wringing about whether machines or organisations are “really” intelligent. The move is pragmatist, and it is this: cognitive claims are engineering protocol claims.

When you say a system has a certain level of cognition, you are not making a metaphysical statement about what is happening inside it. You are specifying which engineering protocols work for managing it. The level of intelligence to attribute to a system is the highest level at which it is useful to model it as having goals, preferences, and memory.

A rock requires no intentional attribution; you model it with physics. A thermostat benefits from minimal goal attribution; you say it “wants” to maintain the temperature, and this helps you predict its behaviour. A mouse requires sophisticated behavioural models; you attribute preferences, fears, learning. A human requires full theory of mind. At each step, the attribution is justified not by metaphysical commitment but by practical utility: does treating the system as having goals help you predict, control, and communicate with it?

This is not a lowering of the bar. It is a sharpening of the question. When a technology leader asks “is our organisation intelligent?” or “is this LLM intelligent?”, Levin’s framework says: the useful question is not about the inner life of the system. It is about what engineering protocols work. Can you manage the system by issuing instructions (low agency)? Do you need to negotiate with it (moderate agency)? Must you design environments that shape its behaviour because direct control is impossible (high agency)?

Most organisations sit somewhere between moderate and high agency. They cannot simply be instructed; anyone who has tried to implement a top-down transformation knows this. They must be managed through incentive design, structural reform, and environmental shaping; exactly the protocols you would use for a high-agency system. This is not a failure of management. It is a recognition of what the system actually is: a collective intelligence with its own dynamics, its own attractors, its own resistance to perturbation.

LLMs sit lower on the continuum but higher than most people assume. Within their training distribution, they can be managed by instruction (prompting). Outside it, they require environmental design: retrieval-augmented generation, tool use, multi-agent architectures, careful evaluation frameworks. The protocols for managing LLMs outside their comfort zone are converging with the protocols for managing organisations outside theirs: create feedback loops, decompose complex problems, introduce adversarial challenge, and build sensing mechanisms that detect when assumptions have broken down.

5. What the Ethics Article Showed, and What This One Adds

In a companion essay, “Can the Statements of an LLM be Ethical?”, I argued that we do not need to settle whether an LLM is conscious or has genuine moral beliefs to evaluate its normative outputs. The philosophical resources of quasi-realism and norm-expressivism give us a framework that works regardless of what is happening inside the system. The question is not whether the machine “really” believes its moral claims. The question is what norms its outputs express, and whether there is a practice of accountability for examining them.

That article made the case for metaethics. This one makes the parallel case for epistemology.

Just as we do not need to settle whether the LLM has moral beliefs to evaluate its ethical outputs, we do not need to settle whether the organisation is “really” intelligent to evaluate its cognitive performance. What matters is not the inner life of the system but the functional properties: can it learn from novel experience? Can it detect when its assumptions have broken down? Can it revise its own operating principles in response to evidence? These are measurable, observable, structural features. They apply equally to neural networks, organisms, and organisations. And the research programmes studying each of these systems are, as I will argue, working on the same problems.

The ethics article showed that LLMs produce normative outputs whose authority comes not from the machine’s inner states but from the practice of accountability that surrounds them. This article shows that organisations produce strategic outputs whose quality depends not on the intelligence of their members but on the structural conditions that enable or prevent collective learning.

The implication is symmetrical and bidirectional. If we grant that LLMs exhibit partial intelligence; pattern-matching within distribution, hallucination outside it, no metacognitive capacity, some emergent reasoning; then we must apply the same analytical framework to organisations. And if we do, both fields of learning offer lessons for each other.

6. The Bidirectional Thesis: What Each Room Can Learn from the Other

The failure modes of LLMs and the failure modes of organisations are not merely analogous. They are expressions of the same underlying dynamics, operating in different substrates. Pattern-matching that mimics competence without producing understanding. Feedback structures that optimise for the wrong signals. The fundamental difficulty of moving from correlation to causation in any learning system.

But the claim is bidirectional. It is not that machine learning provides a playbook for organisational transformation. It is that both fields are working on the same problems, and each has developed strategies the other has not tried.

Machine learning has formalised problems that organisational theory describes qualitatively. Hallucination formalises skilled incompetence. Reward hacking formalises defensive routines. Distribution shift formalises the transition from complicated to complex domains. The exploration-exploitation tradeoff formalises the conditions under which learning occurs. These formalisations do not replace organisational theories. They sharpen them; making them testable, measurable, and amenable to intervention.

Organisational theory has described conditions that machine learning is only now encountering. Argyris described single-loop and double-loop learning decades before anyone built a system that could exhibit both. Weick described sensemaking before anyone built a model that could do in-context learning. Edmondson described psychological safety before anyone formalised the exploration-exploitation tradeoff. Illich distinguished convivial from manipulative institutions before anyone asked whether AI systems amplify or replace human intelligence. The organisational theorists got there first. They saw the dynamics in the substrate they knew. The machine learning researchers are rediscovering the same dynamics in a different substrate, with the advantage of mathematical precision and the disadvantage of thinking they are seeing something new.

Harry Halpin’s 2025 paper, “Artificial Intelligence versus Collective Intelligence,” traces this convergence to its philosophical root. The ontological presupposition of AI, Halpin argues, is the liberal autonomous individual of Locke and Kant. Herbert Simon, the founding figure of both AI and organisational decision theory, explicitly connected his work on artificial intelligence to a programme in cognitive science, economics, and politics that assumed intelligence is a property of individuals engaging in reasoning over representations. This assumption shaped how organisations think about intelligence: find the smart person, give them data, expect good decisions.

But LLMs are not individual intelligences. They are statistical models of collective human language on the web. The intelligence in an LLM is not in the model. It is a compressed, distorted reflection of the collective intelligence that produced the training data. Deploying an LLM in an organisation is layering one form of collective intelligence (a statistical summary of the web) onto another (the organisation itself). The question is whether these two forms enhance or interfere with each other. And that question cannot be answered without understanding both as collective intelligences operating under structural constraints.

This is why the fields need each other. Machine learning engineers need organisational theory to understand the human systems in which their models will operate. Organisational theorists need machine learning to formalise the dynamics they have described qualitatively for decades. And both need the philosophical framework that Levin, Falandays, Chollet, and Halpin have begun to construct: a framework that treats intelligence as continuous, collective, substrate-independent, and measurable.

7. What This Means for the Series

This essay, together with “Can the Statements of an LLM be Ethical?”, establishes the philosophical foundation for an approach to understanding learning in both organisations and LLM’s. The ethics article showed that normative evaluation works without settling consciousness. This article shows that cognitive evaluation works without settling whether organisations are “really” intelligent. Together, they license the structural parallels between specific failure modes in ML and specific failure modes in organisations; not as decorative analogies but as expressions of shared mechanisms in systems that sit at different points on the same continuum.

The nine observable probes that this series has developed across its Learning and Deciding phases are, in Levin’s terms, a diagnostic for the size of an organisation’s cognitive light cone. Can the organisation tell the truth about its own performance? That determines whether its feedback loops function. Are its people close enough to reality to see what is actually happening? That determines whether its sensing mechanisms work. Can it integrate conflict rather than suppress it? That determines whether it can explore beyond its current local optimum. Each probe measures a structural condition for collective intelligence. Each one applies, with minor translation, to both organisations and AI systems.

The three levers of the series; Identity, Information, Interaction; map to the requirements that Falandays and colleagues identified for any collective intelligence: agents with competencies (Identity), mechanisms of communication (Information), and structures of coordination (Interaction). The levers are not prescriptions. They are the minimal conditions under which collective intelligence can emerge. Without them, what you have is not an intelligent organisation. It is a population of competent individuals in a container.

And the difference between those two things is everything.

Further Reading

Falandays, J. B., et al., “All Intelligence is Collective Intelligence,” Journal of Multiscale Neuroscience 2(1), 169-191 (2023). Open access. The paper that dissolves the individual/collective intelligence distinction. Read it alongside any organisational design text and notice that the abstract requirements for collective intelligence; agents, interaction mechanisms, self-organisation toward adaptive behaviour; are the requirements for a functioning team.

Levin, M., “Technological Approach to Mind Everywhere (TAME),” Frontiers in Systems Neuroscience 16, 768201 (2022). Open access. The framework that places intelligence on a continuous, substrate-independent scale. The persuadability continuum and the cognitive light cone concept are immediately applicable to organisational diagnosis.

McMillen, P. and Levin, M., “Collective Intelligence: A Unifying Concept for Integrating Biology Across Scales and Substrates,” Communications Biology 7, 378 (2024). Open access. The multiscale competency architecture applied to biological systems. The cancer-as-defection analogy alone is worth the read for any leader managing misaligned teams.

Watson, R. and Levin, M., “The Collective Intelligence of Evolution and Development,” Collective Intelligence 2(2) (2023). The connectionist framework for understanding what structural conditions turn a population into an intelligent collective.

Chollet, F., “On the Measure of Intelligence,” arXiv:1911.01547 (2019). The formal definition of intelligence as skill-acquisition efficiency. Read section II on the distinction between skill and intelligence; it will change how you evaluate every strategy presentation you attend.

Halpin, H., “Artificial Intelligence versus Collective Intelligence,” AI and Society 40, 4589-4604 (2025). Open access. Traces how Simon’s ideology of the autonomous rational individual shaped both AI research and organisational decision theory, and argues for collective intelligence as the alternative.

I write about the industry and its approach in general. None of the opinions or examples in my articles necessarily relate to present or past employers. I draw on conversations with many practitioners and all views are my own.

Learning is not a process. It is a condition. It is what happens when people with safe-enough identities receive clean-enough information through institutions that serve rather than obstruct them. If those conditions are absent, no process will produce learning. If those conditions are present, learning will happen with or without a programme.

Every thinker in this series has, from a different angle, diagnosed the same underlying failure: organisations are structured to prevent the learning they claim to want. They respond by designing more learning processes. More workshops. More curricula. More governance. And every additional process intervention leaves the underlying conditions untouched, because the conditions are not about what the organisation does. They are about what the organisation is: the identities people hold, the information that flows or does not, and the institutional forms that either create the space for learning or consume it.

Illich provides the deepest explanation for why the process model persists despite its consistent failure. Institutions designed to deliver a capability end up replacing that capability with the consumption of institutional services. Schools do not produce learning; they produce the need for more schooling. Learning programmes do not produce organisational capability; they produce the need for more learning programmes. The mechanism is what Illich calls the institutionalisation of values: the programme defines learning as something that requires its mediation, establishes a monopoly over what counts as legitimate learning, and creates dependency. The team learning through practice on a real problem outside the programme is, within the programme’s framework, not learning at all: their activity is not tracked, not measured, not credited. The programme has made genuine learning invisible by defining learning as something that requires the programme.

This article synthesises the series into a single diagnostic model. It identifies three conditions for learning, each governed by a thinker whose work illuminates why the condition is so difficult to create:

• Identity must be safe enough to change. Pierre Bourdieu provides the theory: habitus, capital, and the embodied dispositions that reproduce the old world below conscious awareness.

• Information must be clean enough to act on. Gregory Bateson provides the theory: the double bind, logical types, and the communicative pathologies that prevent organisations from hearing what they need to hear.

• The institutional form must be convivial enough to permit learning. Ivan Illich provides the theory: the confusion of process and substance, radical monopoly, the hidden curriculum, and the distinction between institutions that serve human purposes and institutions that replace them.

For each condition, three probes test whether it is present. A probe is not a metric. It is a question you can answer by going and looking. If the condition is present, the probes will tell you. If it is absent, the probes will show you where to look for the obstruction.

These three governing thinkers share a quality that makes them honest rather than comfortable. They are all pessimistic about the possibility of deliberate control. Bourdieu: your dispositions reproduce the old world before you are aware of it. Bateson: the communicative traps operate at logical levels you cannot access from within. Illich: the institutions you build to create learning will replace learning with the consumption of their own services. The other thinkers in each cluster provide the practices that make action possible despite these constraints. The framework says: this is harder than you think. Then it says: here is what you can do anyway.

Before the management scholars object: yes, this synthesis compresses decades of research into a practitioner framework, and the governing thinkers themselves would resist the categorisation. Illich would warn that the framework itself could become the kind of manipulative institution his work diagnoses. Bourdieu would insist that the barriers are more deeply embodied than any framework can capture. Bateson would note that the framework could become a logical-type error of the kind his work identifies. These objections are valid. What follows is offered not as the final word but as a working tool: a set of lenses that can be refined through use.

The Governing Hypothesis

Learning is a condition, not a process. It emerges when three conditions are met: identity is safe enough to change, information is clean enough to act on, and the institutional form is convivial enough to permit learning. The leader’s role is not to design or manage learning but to create and protect these conditions.

This rejects the standard transformation model where leaders design a learning programme and deliver it to staff. It also rejects the more progressive model where leaders “learn from their people” and “teach the new ways.” Both frames treat learning as something transferred between parties through a channel. The channel metaphor is the problem. Learning is not content that travels along a pipe. It is what the pattern of interaction produces when the conditions allow it.

Follett saw this a century ago. Her concept of circular response, where each party’s behaviour continuously reshapes the other’s, and her insistence that the group produces ideas that no individual could generate alone, describe learning as an emergent property of interaction, not a transfer between individuals.

Bateson provides the deepest theoretical foundation. His hierarchy of learning levels, from Learning I (correcting errors within a fixed frame) through Learning II (learning to learn, changing the frame itself) to the rare Learning III (transforming the identity of the learner), maps the territory this article covers. Most organisations operate permanently at Learning I: they correct errors without questioning the governing assumptions. The nine probes that follow are, collectively, a diagnostic for whether your organisation has the conditions for Learning II. They test whether the organisation can learn to learn.

The Identity Condition

Identity must be safe enough to change

Identity is the deepest condition. It concerns who people are, what they are worth in the field, and whether their embodied dispositions can shift. Bourdieu’s concept of habitus, the system of durable, transposable dispositions acquired through lived experience, explains why this condition is the most difficult to create. You do not choose your habitus. It is deposited in you through years of participation in a particular field: the instinctive deference to seniority, the automatic framing of problems in terms the organisation recognises, the professional reflexes that tell you what counts as good work. These are not choices. They are embodied dispositions that operate below conscious awareness.

Identity constrains everything else. If your professional capital is at stake, you will not share information that threatens it. If your habitus is calibrated to the old field, your interactions will reproduce the old patterns regardless of what the new strategy says. The condition is present when people can engage with the change without experiencing it as an existential threat to who they are and what they are worth. It is absent when people perform the new way of working while privately preserving the old.

Probe 1. Can People Tolerate Losing What They Have?

This is the probe that most transformation programmes refuse to apply. When you tell a senior professional that their role is evolving, that the skills they have spent a decade perfecting are no longer the primary source of value, you are not asking them to learn a new skill. You are asking them to accept a loss. The loss may be temporary, partial, or ultimately compensated by gains. But it is experienced as a loss, and the experience governs the response.

Bourdieu names the mechanism: hysteresis, the painful lag between a changed field and an unchanged habitus. When the rules of the game shift but your embodied dispositions remain calibrated to the old rules, the result is not discomfort. It is a crisis of capital. The professional whose cultural capital consists of a specific technical expertise faces devaluation if transformation renders that expertise secondary. Resistance to transformation is, in most cases, resistance to capital devaluation. It is entirely rational.

Giddens provides the theoretical depth. His concept of ontological security explains why resistance is disproportionate to the rational threat. Routines are not just convenient; they are psychologically necessary. Disrupting them produces existential anxiety that will express itself as resistance regardless of how compelling the business case is. Kahneman’s prospect theory explains the intensity: losses loom roughly twice as large as equivalent gains. The first steps of any change, being closest to the current reference point, produce the most acute sensitivity.

Kegan’s theory of adult development adds a dimension the other thinkers miss. The capacity to manage loss depends on the developmental stage. At the socialised mind, identity is derived from the expectations of the community; a shift in what the community values is experienced as personal identity crisis. At the self-authoring mind, the individual has an internal compass that can navigate changing expectations. Most programmes assume a self-authoring workforce. Many organisations have a predominantly socialised one.

Heifetz makes this probe actionable. People do not resist change. They resist loss. The leader’s job is not to eliminate the loss but to name it, to create the holding environment in which it can be processed, and to help people distinguish what is essential from what is expendable.

Goffman reveals how the loss is enacted socially. Transformation creates what he calls stigma: the informal labelling of those whose identity is now discredited. These labels are communicated through expressions given off: a slight pause, a glance exchanged, an invitation that does not arrive. The person whose capital has been devalued finds their contributions received differently. The colleagues who route around them are managing the interaction: including the stigmatised perspective would require everyone to adjust their performance. The result is that the people who understand most deeply what is being lost are systematically excluded from the conversations that would benefit most from their knowledge.

What to look for: Compliance without commitment. People going through the motions of the new way of working while quietly preserving the old. Passive resistance that never quite surfaces as objection. Hostility that seems disproportionate to the change being proposed. All of these are signals that the loss has not been named or addressed.

What to do: Identity transitions require three things: social support (others making the same transition), role models (people who have already made it), and narrative resources (stories that make the transition meaningful rather than diminishing). The leader must also attend to their own loss: the transition from the person who knows the answer to the person who creates the conditions for answers to emerge.

Probe 2. Is Learning Happening Through Practice, or Through Instruction?

Most organisations, when they decide to transform, create a training programme. They commission e-learning modules, hire consultants to run workshops, and build a curriculum. Then they wonder why nothing changes. This is the process fallacy in its purest form.

Habitus can only be reshaped through participation in a changed field. You cannot lecture someone into new embodied dispositions. This is not a pedagogical preference. It is a claim about what identity is and how it changes. The distinction between instruction and practice determines whether your investment in learning produces transformation or produces people who can describe transformation without being able to do it.

Illich names the deeper problem. The training programme does not merely fail to change habitus. It teaches a hidden curriculum: that learning requires a programme, that capability is certified by completion, that the organisation does not trust you to learn on your own, and that the transformation is something being done to you rather than something you are doing. This hidden curriculum directly undermines the identity condition by positioning people as recipients rather than agents.

Wenger and Lave provide the most complete account. Their concept of legitimate peripheral participation describes how expertise develops: newcomers start at the edges of a community of practice, performing simple but genuine tasks, and gradually move toward the centre as their competence grows. Learning is not something that precedes participation. It is participation. Nonaka and Takeuchi’s SECI model describes the knowledge conversion practice-based learning requires: the critical step is externalisation, making tacit knowledge explicit, which cannot be done in a classroom. It requires people working together on real problems, struggling to articulate what they know, and discovering through the struggle what they did not know they knew.

Giddens’ distinction between practical consciousness and discursive consciousness is the theoretical foundation. Transformation requires changing practical consciousness: the things people do without thinking. Training programmes change discursive consciousness: people learn to talk differently about their work. Talking differently does not mean working differently.

March’s technology of foolishness makes the case for sensible irrationality: sometimes you must act before you know your preferences, play before you are serious, and experiment without justification. Organisations that permit only rational, justified action will never discover anything their existing rationality could not predict.

What to look for: What percentage of your transformation budget is spent on training courses versus on giving people time, tools, and real problems to work on? Are people developing new capabilities through practice, or developing new vocabulary through instruction?

What to do: Stop trying to govern learning. Create the conditions for it: psychological safety, time, access to tools, proximity to real problems. Then pay attention to what emerges. The teams quietly solving real problems outside the governance framework are not insubordinate. They are the emergent strategy trying to tell you where the value is.

Probe 3. Do People Believe That This Time Will Be Different?

Learned helplessness is habitus. It is not a mood or a passing attitude. It is a sedimented disposition formed through repeated experience that effort does not produce results. Organisations with a history of failed change programmes have employees whose embodied orientation toward transformation has been shaped by each successive failure. “Change fatigue” is not fatigue. It is a dispositional stance, and it is rational.

If the identity condition requires that people feel safe enough to change, learned helplessness is the evidence that the condition has been destroyed by previous attempts to create it. Each failed transformation taught the same lesson: the process was run, the learning was mandated, and nothing changed. The condition was never present. Only the process was.

Seligman identifies the mechanism: explanatory style. A pessimistic style, permanent, pervasive, and personal, sustains helplessness. An optimistic style, temporary, specific, and external for bad events, predicts recovery. The explanatory style that dominates an organisation is cultural, reproduced in the stories people tell about previous change, in the cynicism that greets new announcements, and in the knowing looks exchanged when the latest programme is unveiled.

Bandura’s self-efficacy research provides the path to recovery. The most powerful source of belief is mastery experience: actually doing the thing and succeeding. A live demonstration where a team tackles a real problem using new methods bypasses intellectual debate. It creates the belief that the new way is possible. Csikszentmihalyi’s flow research describes the conditions under which this work becomes intrinsically engaging: clear goals, immediate feedback, and progressive challenge matched to skill.

Dweck’s research addresses the disposition directly. Organisations that celebrate talent reinforce the belief that ability is fixed, making every challenge a test. Organisations that celebrate process and effort reinforce the belief that ability is developed, making challenge an opportunity. Review what your organisation celebrates. The pattern of celebration is reshaping habitus in real time.

What to look for: Listen to the language. When people describe the transformation using the same phrases they used about the last one, learned helplessness is present. When the most talented people are not volunteering for the new work, they have made a rational calculation about where value is recognised.

What to do: Counter learned helplessness with small wins (Weick). Demonstrate through action, not argument. Find ways to make learning visible and valued before it produces measurable output. Watch for the signals: if people are engaged, curious, and arguing about how to make things better, the condition is forming. If they are filling in templates and waiting for approval, the process is running but the condition is absent.

The Information Condition

Information must be clean enough to act on

Information is the middle condition. It concerns whether accurate data can flow through the organisation, whether contradictory signals are being sent at different logical levels, and whether the organisation can distinguish its stated reality from its actual one.

Bateson’s concept of the double bind is the governing mechanism. A double bind occurs when someone receives contradictory messages at different levels of communication and cannot comment on the contradiction. “We encourage honest feedback” delivered in a meeting where honest feedback has historically produced negative consequences is a double bind. The person cannot respond to the explicit message without ignoring the implicit one, and cannot comment on the contradiction without violating the implicit message. The result is paralysis disguised as compliance.

The double bind is the link between identity and institution. It is sustained by habitus (the manager who punishes honesty does so from embodied disposition, not conscious choice) and reinforced by institutional form (the governance structure that demands both innovation and predictability). When information is dirty, it does not matter how safe identity feels or how convivial the institution is. The signals people act on are corrupted. They are navigating by a map that contradicts the landscape, and nobody can say so.

Goffman provides the micro-mechanism that maintains the double bind in daily practice. Every meeting is a performance in which participants collaborate to maintain a shared definition of the situation. The formal meeting is the front stage. The hallway conversation is the back stage. The gap between the two is not a communication failure. It is the interaction order working exactly as designed. Both parties collaborate to protect each other’s face, because a face-threatening act endangers the entire interaction. Defensive routines are not unilateral. They are mutual.

The information condition is present when the formal conversation and the shadow conversation say the same thing. It is absent when people know something to be true but cannot say it without career risk.

Probe 4. Can People Tell the Truth About What Is Happening?

Argyris identified the mechanism that corrupts organisational information more precisely than anyone else in the series: defensive routines. The gap between espoused theory and theory-in-use is a double bind that Argyris described in behavioural terms and Bateson would recognise in communicative ones.

Goffman reveals why these routines are so resistant to intervention. Training individuals in Model II behaviour works in the workshop but not in the meeting, because the meeting is a different social situation with different face-work requirements. The person trained to “test their assumptions publicly” walks into a meeting where the senior leader has just presented a strategy, and the face-work calculus kicks in: testing the assumption would threaten the leader’s face, disrupt the shared definition of the situation, and require every participant to adjust their performance. The trained behaviour collapses under the weight of the interaction order. This is why “create openness” interventions consistently fail. The safe-space announcement is itself a front-stage performance.

Schön added the concept of frame reflection: the capacity to surface the tacit frames through which parties construct their understanding. Most undiscussables are not facts that people are hiding. They are frames that people do not realise they hold. The leader who frames every initiative as cost-reduction and the practitioner who frames it as a threat to craft are not lying. They are operating within different frames that make different conclusions inevitable. Until both frames are surfaced, the disagreement is unintelligible to both parties.

At the organisational level, Westrum’s typology classifies information architectures. In pathological cultures, messengers are punished. In bureaucratic cultures, they are channelled into mechanisms designed to slow information. Only in generative cultures does information flow to where it is needed. Edmondson’s psychological safety provides the floor, with the critical caveat that safety without high standards produces comfort, not learning. Deming cuts through the abstraction: most variation is common cause, produced by the system. Blaming individuals for systemic failures teaches people to hide information about how the system works.

What to look for: The gap between what is said in formal meetings and what is said in hallway conversations. That gap is the double bind made visible. Pay particular attention to the voices from below.

What to do: Name the double bind explicitly. “I notice that we say we want honest feedback but the last person who gave it was sidelined. That contradiction is a problem I want to address.” Naming the contradiction is the first step to dissolving it, because the double bind’s power depends on the prohibition against naming.

Probe 5. Are Decision-Makers Close to the Work?

Information degrades with distance. Every layer of hierarchy between the person deciding and the person doing is a reduction in signal quality.

Drucker’s insight is foundational: the knowledge worker must define the task. The person defining the task must be intimate with the domain. Peters translated this into Management by Walking Around: the only way to know what is going on is to go and see. Mintzberg’s research supports this: the strategist who is not touching the clay is hallucinating a strategy.

Normann’s map-landscape dialectic reveals the deepest version. Leaders distant from the work are not merely missing details. They are carrying a map that makes the real landscape invisible. Kahneman’s WYSIATI amplifies this: the coherent narrative constructed from the distant vantage point suppresses awareness of what the leader does not know.

The double bind operates here too: “We trust our teams” delivered through a governance structure requiring five layers of approval. The team receives two messages: you are trusted, and you are not trusted.

What to look for: Count the layers between the person making the transformation decision and the person doing the work. If the people designing the transformation have never done the work it changes, the quality of their decisions is structurally limited.

What to do: Cancel the steering committee. Go and watch people work. Sit with a team as they tackle a real problem. The information you need cannot travel upward through the hierarchy. You must go and get it.

Probe 6. Is the Organisation Changing What It Rewards, or Just What It Says?

This is the structural double bind: contradictory messages encoded in the institution itself. “We value innovation” (signification) while promoting those who maintain stability (legitimation) while funding the old programmes (domination). Giddens’ three dimensions of structuration must move together. When they do not, the organisation is a double bind made structural.

This is where the information condition and the identity condition meet. The habitus formed around the old reward structure generates the information pathology. People do not merely observe that the old behaviour is rewarded. Their embodied dispositions produce the old behaviour before the question of reward even arises. Weber explains the persistence: bureaucratic rationality is not a surface feature but the constitutive logic of modern institutions.

Senge’s systems thinking reveals the dynamic: the feedback loops sustaining the current structure are faster and stronger than those supporting change. Transformation often follows initial enthusiasm followed by regression: the reinforcing loops of early success are overwhelmed by the balancing loops of structural reproduction.

What to look for: Examine the last three promotions. What was actually rewarded? Examine the last three performance reviews. What was measured? Examine budget allocation. Where did the money go? These reveal the theory-in-use, regardless of the strategy deck.

What to do: You cannot change the culture by talking about culture. You change it by changing the practices: the meetings, the metrics, the promotion criteria, the budget allocation, the definition of “done.” Heifetz names the discipline: distinguish the technical work (changing the policy) from the adaptive work (changing what people value). Both are necessary. Neither alone is sufficient.

The Institutional Condition

The institutional form must be convivial enough to permit learning

Institution is the surface condition: the one the leader can most directly shape. It concerns the formal and informal structures that mediate between people and their activity. These structures either create the space from which learning emerges or consume it.

Illich’s foundational observation is that institutions designed to deliver a human capability end up replacing that capability with the consumption of institutional services. The mechanism operates in three stages. First, the institution defines the activity as requiring professional mediation: you cannot learn without a training programme. Second, the institution establishes a monopoly over delivery: learning outside the programme is not recognised. Third, the institution creates dependency: people come to believe they cannot learn without the programme. When the monopoly is complete, Illich calls it a radical monopoly: not a monopoly over competing products but a monopoly over the conditions of the activity itself.

The programme also teaches a hidden curriculum: that learning requires a programme, that capability is certified by completion, that the organisation does not trust you to learn on your own. This hidden curriculum reshapes habitus: people do not merely learn the content; they learn the dispositions the institution requires.

When the institutional response to a problem makes the problem worse, Illich calls it iatrogenesis. Clinical iatrogenesis: direct harm (governance overhead, the delay the approval process adds). Social iatrogenesis: the redefinition of normal capability as requiring institutional mediation (informal competence reclassified as “untrained”). Cultural iatrogenesis: the destruction of the capacity for autonomous action (professionals who cannot imagine learning without a programme). Each level deepens the next. The cascade is self-reinforcing.

Illich’s diagnostic distinction is between convivial and manipulative institutions. A convivial institution provides resources people can use according to their own purposes: tools, time, access to problems, access to people who know things. A manipulative institution prescribes the content, controls the sequence, measures the consumption, and produces the credential. The institutional condition is present when the structures serve the people. It is absent when the people serve the structures.

Probe 7. Does the Institutional Form Serve the People, or Do the People Serve the Institution?

This is Illich’s convivial/manipulative test applied to every element of the transformation. The programme measures completion rates, certification numbers, training hours: all measures of process consumption. None measure capability development. The teams that completed the modules can describe techniques in the language the platform provided. The teams that have been quietly using AI to solve real problems for months are uncertified. Within the programme’s logic, they are untrained. The programme has confused its own activity with the outcome it was designed to produce.

Radical monopoly deepens the diagnostic. The programme has reshaped the environment so that the outcome cannot be produced without it. The certification requirement prevents experimentation. The budget starves informal learning. The governance fills meeting agendas with programme management rather than problem-solving. The conditions under which learning would emerge naturally have been consumed by the programme.

Stacey’s gesture concept operates here. In a convivial institution, people make gestures and attend to the responses. In a manipulative institution, the institution makes gestures on people’s behalf and channels the responses through governance rather than letting them be experienced directly.

Goffman explains the stability. The governance structures are front-stage performances that produce impression-managed information. The leader who relies on governance for information about whether learning is happening is watching the performance and mistaking it for reality.

Peters provides the emotional dimension. The twelve-month analysis process is the performance of diligence while systematically preventing the only activity that produces understanding. Weick’s sensemaking is retrospective: waiting for understanding before acting gets the sequence backwards.

Schön names the mechanism: reflection-in-action. The practitioner engages in a reflective conversation with the situation, adjusting as the material talks back. The specification is not a document you write before work begins. It is an ongoing dialogue between intent and emergence.

What to look for: Apply Illich’s test to every element. Does this element provide resources for learning (tools, time, real problems), or prescribe and control learning (modules, sequences, certifications)? Count the convivial elements and the manipulative ones. The ratio reveals the institutional character. How many layers of approval stand between a team and an experiment?

What to do: Create spaces where action is permitted before understanding is complete. A real problem. A real team. A day to experiment. But Heifetz adds: the instinct to provide the answer is itself a barrier. The leader’s role is to create the space, not prescribe the action. Redirect resources from institutional infrastructure toward conditions: tools, time, real problems, mentors, protected experimentation space.

Probe 8. Can the Institution Stop Doing What No Longer Works?

Every resource devoted to preserving an activity whose purpose has expired is a resource unavailable for transformation. The inability to abandon is not a failure of nerve. It is a structural feature of institutions that have optimised for their own continuation.

Illich provides the diagnosis through iatrogenesis. At the clinical level: direct dysfunction and overhead. At the social level: informal competence reclassified as “untrained,” organic adaptation reclassified as “ungoverned.” At the cultural level: the destruction of autonomous learning capacity. Each layer of institutional response deepens the dependency and consumes resources that would otherwise be available for the conditions the institution was supposed to create.

The institution resists abandonment because it generates its own justification. It measures what it does (training delivered, milestones reached) and presents these as evidence of progress. The metrics measure institutional activity, not human capability. Asking “but can the teams actually do the work?” is treated as an attack on the programme rather than a diagnostic question.

Drucker’s systematic abandonment asks the practical question: “If we were not already doing this, would we start now?” Illich goes further: has the institution made it impossible to even imagine doing without it?

March provides the theoretical foundation. Exploitation (refining what you know) produces measurable, proximate returns. Exploration (trying something genuinely new) produces ambiguous, distant returns. The manipulative institution is an exploitation machine. Exploration is structurally prevented. Christensen’s disruption theory reveals the market consequence: incumbents fail because their proximity is to the wrong customers. Taleb adds that the inability to abandon creates fragility.

What to look for: Ask Drucker’s question about your transformation programme itself. How many activities exist solely because they were created at the start? Apply Illich’s iatrogenesis test at each level. At which level is the harm deepest?

What to do: Institute a regular abandonment review. Not what to add, but what to stop. If the programme were abolished, would the organisation be able to learn? If the answer is no, the programme has achieved radical monopoly and is itself the primary barrier. The freed capacity is where the learning condition becomes possible.

Probe 9. Can the Institution Integrate Conflict, or Does It Suppress It?

When an institutional gesture produces resistance, what does the institution do with it? Most suppress it: through hierarchy (the senior person prevails), through process (a committee resolves it), or through avoidance (acknowledged and never revisited). Each method loses the divergent signal, and the institution continues with the illusion of consensus.

Illich’s framework reveals why suppression is structural. The manipulative institution cannot permit genuine conflict about its own purpose because its continuation depends on the confusion of process with substance. If the conflict were explored, someone might ask “Is this programme actually producing learning?” Weber would call this the displacement of value rationality by means-ends rationality. Illich would call it the institutional logic working as designed.

Follett, writing a century ago, saw this with clarity. Her distinction between domination, compromise, and integration is the earliest and cleanest statement of what productive conflict looks like. Integration requires surfacing the real desires beneath the stated positions. Until both desires are visible, the only available resolution is domination or compromise. Neither produces learning.

Stacey deepens this. Legitimising dissent is not better communication technique. It is a political act requiring someone with sufficient power to make it safe for others to speak, and sufficient courage to tolerate what they say. Heifetz provides the operational discipline: the holding environment where conflict can be expressed without destroying the group. The leader regulates the temperature: too little heat and nothing changes; too much and people retreat into defensive routines.

Edmondson’s psychological safety is the floor. Without it, people will not take the interpersonal risk of expressing a divergent view. But safety alone is not sufficient. A safe team can tell each other small truths while collectively avoiding the large one.

What to look for: Watch what happens when someone disagrees in a meeting. Is the disagreement explored or managed? When the shadow conversation contradicts the formal conversation, which one changes?

What to do: Treat resistance as information. Follett’s integration requires joint study: the parties must study the situation together, not negotiate from fixed positions. The leader who responds to resistance by restating the strategy has closed the loop prematurely. The leader who responds by asking “What are you seeing that I am not?” has opened it.

How the Conditions Compound

The three conditions are not independent. They form a directional hierarchy: identity constrains information constrains interaction. But the causation runs the other way for intervention: interaction is where the leader acts, and sustained change in interaction patterns changes information flow, which over time changes identity.

Identity constrains information. If your professional capital is at stake, you will not share information that threatens it. The senior architect whose identity is built around system design will not surface data suggesting specification-writing is more valuable. Not because they are dishonest, but because their habitus filters the information before it reaches conscious awareness. They literally do not see what threatens their capital.

Information constrains interaction. If the organisation is saturated with double binds, the institutional response will be manipulative, because the institution cannot resolve contradictions it cannot name. It adds governance to manage contradictions rather than dissolving them. Every unresolved double bind generates institutional complexity: another committee, another reporting line. Illich’s iatrogenic cascade operates here: the institutional response to the information pathology deepens the information pathology.

But interaction is where the leader acts. You do not reshape identity directly. You do not resolve double binds directly. You shape institutional form. And sustained institutional change, if genuinely convivial, reshapes information flow (because convivial institutions do not produce double binds), which over time reshapes identity (because people participating in a convivial institution develop different habitus from those trapped in a manipulative one). The conversion from manipulative to convivial is the leader’s primary institutional act.

This is why transformation takes so long and why it so often fails. Institutional form can change relatively quickly. Information environments change slowly. Identity structures change very slowly. A transformation that changes only institutional form will feel productive but will not persist if the information and identity conditions remain unchanged. One that addresses all three simultaneously has a chance.

It also explains why learning programmes fail. They are manipulative institutions applied to a condition problem. If the conditions are absent, the process runs and nothing changes. If the conditions are present, the process is unnecessary. Illich goes further: the programme does not merely fail to produce the condition. It actively prevents it from emerging, by establishing a radical monopoly over the definition of learning.

Ackoff would insist: these barriers constitute a mess, not a collection of problems. A mess is a system of problems that cannot be solved individually. Addressing one probe while leaving the others untouched produces the appearance of progress within a system that has not actually changed.

Bringing It All Together

The leader’s role is to shape the institutional form so that it creates rather than consumes the conditions for learning. This is not institutional design in the conventional sense. Stacey’s warning applies: there is no position outside the institution from which to design it. What the leader can do is participate skilfully: noticing which conditions are absent, making gestures that create the missing condition, and having the courage to not know in advance what will emerge.

Heifetz operationalises this as the oscillation between the balcony, where patterns become visible, and the dance floor, where they are lived. Normann adds the conceptual dimension: question whether the map itself is correct. Kahneman warns that the leader’s own biases will make all of this harder than it sounds: System 1 will generate a coherent narrative, confidence will feel like evidence, and the losses entailed by change will loom larger than the gains.

Bateson provides the deepest frame. Most organisations operate at Learning I. The nine probes are a diagnostic for the conditions required for Learning II: learning to learn, changing the framework itself. This is where real transformation occurs. It is also where anxiety is highest, because the framework being questioned is the one that provides the organisation’s identity, coherence, and sense of purpose.

Bourdieu explains why this is so hard. The framework is not just an intellectual structure. It is inscribed in bodies, reflexes, and taken-for-granted assumptions. Changing it requires changing habitus, and habitus changes only through sustained practice in a changed field. There are no shortcuts.

Illich provides the pivot between diagnosis and action. His framework explains why the institutional response to the difficulty of learning, building a learning programme, reliably makes things worse. The leader who understands this will stop asking “How do I design a better learning programme?” and start asking “How do I create the conditions from which learning emerges, and then get out of the way?” The practical test is the convivial/manipulative distinction: for every element of the institutional form, ask whether it provides resources for self-directed activity or prescribes, controls, and credentials. Redirect from the manipulative toward the convivial. Protect the teams already learning outside the programme. Make informal learning visible and valued.

The probes are not specific to any particular transformation. They are the conditions for any organisational learning. What makes them urgent now is that the cost of not having them has become impossible to ignore. Organisations where the condition is present will adapt with intention, craft, and responsiveness. Organisations where it is absent will absorb the tools while preserving the structures, capture the terminology while avoiding the transformation, and emerge essentially unchanged.

The barriers are not new. They were diagnosed decades ago. What is new is the cost of leaving them in place.

Further Reading

Pierre Bourdieu, The Logic of Practice (1990). Bourdieu and Wacquant, An Invitation to Reflexive Sociology (1992) is more accessible. Bourdieu, The Forms of Capital (1986) is freely available.

Gregory Bateson, Steps to an Ecology of Mind (1972). The essays on learning levels and the double bind.

Ivan Illich, Deschooling Society (1971). Freely available as PDF. Tools for Conviviality (1973) develops the convivial/manipulative distinction.

Chris Argyris, Overcoming Organizational Defenses (1990). Erving Goffman, The Presentation of Self in Everyday Life (1959). Karl Weick, Sensemaking in Organizations (1995). Mary Parker Follett, Creative Experience (1924). Freely available. Ronald Heifetz, The Practice of Adaptive Leadership (2009). Ralph Stacey, Complex Responsive Processes in Organizations (2010). Russell Ackoff, Ackoff’s Best (1999).

I write about the industry and its approach in general. None of the opinions or examples in my articles necessarily relate to present or past employers. I draw on conversations with many practitioners and all views are my own.

Your transformation programme has a learning component. It has modules, certifications, completion rates, and a dashboard that shows how many people have been trained. The numbers look good. And when you walk the floors, the people who completed the training are not doing anything differently. They have the certificate. They attended the sessions. They can speak the language. But the actual work is unchanged. The programme has produced graduates without producing learning.

Ivan Illich would not be surprised. His argument, developed most forcefully in Deschooling Society (1971) and Tools for Conviviality (1973), is that this outcome is not a failure of the programme. It is the programme working exactly as institutions work. The institution designed to produce learning replaces learning with the consumption of its own services. The certificate becomes the goal. Attendance becomes the evidence. The dashboard becomes the product. And the thing the institution was created to enable, genuine change in practice, is quietly displaced by the thing it actually produces: compliance with its own requirements. Illich called this institutional inversion: the point at which an institution becomes counterproductive to its own stated purpose. It is the theoretical foundation for every observation in this series about structures that obstruct the work they were designed to serve, and it governs the Interaction dimension of the entire Learning phase.

1. Institutional Inversion: When the Structure Serves Itself

Illich’s sharpest insight is that institutions systematically confuse the process they administer with the outcome they were created to achieve. Schooling is confused with education. Treatment is confused with health. Attending a course is confused with learning. The institutional process, because it is visible, measurable, and governable, displaces the substantive outcome, which is often none of these things.

This confusion is structural, not accidental. The institution must justify its existence, its budget, its headcount. It does so by measuring what it can control: enrolment, completion, certification. The substance, genuine capability change, is harder to measure and slower to materialise. Over time, the metrics of process become the definition of success, and anyone who questions the equation (”but are people actually learning?”) is treated as questioning the institution itself.

Institutional inversion explains all three Interaction probes in this series. Structures obstruct when they have inverted: the governance framework that was created to enable AI adoption now prevents experimentation, because the framework’s survival depends on the continuation of the problem it was designed to solve. The organisation cannot stop what no longer works because the activity has become self-justifying: the architecture review board continues to meet because the board exists, and the board exists because it meets. Conflict cannot be integrated because the institution’s survival depends on suppressing challenges to its own logic: the team that succeeds without the programme threatens the programme’s necessity, so their success is either absorbed (retrospectively certified, claimed as a programme outcome) or ignored (it happened outside the framework, so it does not count).

Bateson’s levels framework diagnoses the epistemological damage. The programme operates at Learning I: it teaches new procedures and vocabulary within the existing frame. What the organisation needs is Learning II: a change in how people learn, a shift in the context that governs how they generate new practice. But Learning II cannot be taught. It emerges from sustained engagement with real problems in conditions where old patterns are genuinely insufficient. The programme, by providing a structured path through pre-digested content, actively prevents the disorientation from which Learning II arises. Argyris diagnosed the same mechanism: the programme’s structure is itself a defensive routine, protecting the organisation from the anxiety of not knowing whether people are actually learning by substituting a measurable process for an unmeasurable outcome.

2. The Hidden Curriculum and Radical Monopoly

Every institution, Illich argued, teaches two things. The official curriculum is the content listed in the syllabus. The hidden curriculum is the implicit lesson about the proper relationship between the learner and the institution: that learning requires a programme, that progress is measured by certification, that the centre of excellence decides what is legitimate, and that the practitioner’s role is to consume, not to experiment independently.

Bourdieu would recognise this as symbolic violence: the hidden curriculum is internalised by the people it constrains, so that the programme’s authority appears natural. The developer who waits for the centre of excellence to approve a new approach before trying it has learned the hidden curriculum perfectly. They are not being cautious. They have been taught that unsanctioned action is illegitimate. The programme has produced the dependency it was designed to prevent.

Illich’s most powerful concept deepens this: radical monopoly, the condition in which an institution has so thoroughly colonised the activity it serves that the activity cannot be imagined without the institution. The car restructured cities so that walking became impractical. The hospital redefined health so that self-care became insufficient by definition. The AI transformation programme redefines competence so that only programme-certified practitioners are considered competent, regardless of what they can actually do. The test is brutal: if you abolished the programme tomorrow, would the organisation be able to adopt AI? If the answer is no, the programme has achieved radical monopoly. The conditions for independent learning have been consumed by the institution that was supposed to create them.

Stacey’s warning connects: formalising communities of practice kills them, because formalisation converts a living pattern of interaction into a managed process. Illich generalises this: every convivial activity, when captured by an institution, undergoes the same conversion. The informal learning that was already happening, the corridor conversation, the team that figured out AI on a real problem without permission, does not survive institutionalisation. Weick’s small wins are the natural enemy of radical monopoly: every informal success demonstrates that capability exists without the institution. Peters’ bias for action is the antidote, but only if the organisation protects the space for unsanctioned success.

3. Iatrogenesis: The Three Levels of Institutional Harm

Illich’s study of medicine introduced iatrogenesis: harm caused by the institution designed to help. He identified three levels that apply to transformation with disturbing precision.

• Clinical iatrogenesis is direct harm. The governance framework creates so much overhead that teams avoid using AI for anything requiring approval. The centre of excellence becomes a bottleneck that prevents practice entirely. Beer would recognise this instantly: the purpose of the system is what it does, and what this system does is prevent the adoption it was created to enable. Beer governs the Interaction lever in the Deciding phase because he provides the architecture (the Viable System Model) that diagnoses and corrects inversion. Illich diagnoses the pathology. Beer provides the remedy. POSIWID is the diagnostic that connects them: if the programme is producing programme artefacts but not changed practice, then producing artefacts is its purpose.

• Social iatrogenesis is the redefinition of normal activity as requiring institutional mediation. The developer who taught themselves AI effectively is invisible because they have no certificate. The team that built a working integration is unrecognised because they did not follow the approved methodology. Normal professional development, learning by doing, has been reclassified as insufficient.

• Cultural iatrogenesis is the deepest. The programme destroys the organisation’s capacity for autonomous learning. After years of governance frameworks, approved tool lists, and mandatory training paths, people have lost the disposition to learn independently. They wait to be told. They wait for the programme. They have internalised the hidden curriculum so completely that learning without institutional mediation seems irresponsible. This is the deepest damage, and it is largely invisible, because the people who have suffered it do not know that anything has been taken from them.

Giddens’s structuration theory explains why cultural iatrogenesis is so resistant to correction. The programme is not merely a set of rules. It is a structure reproduced through daily practice: the habitual consultation of the approved tool list, the automatic referral to the centre of excellence, the reflex to check governance before acting. Bourdieu would say the programme has inscribed itself in the habitus. The dependency is no longer institutional. It is embodied.

4. From Programme to Conditions

The question every transformation leader must ask is not “How do I build a better programme?” but “How do I create the conditions from which learning emerges without a programme?”

Illich called the alternative a convivial institution: one that provides tools and access without dictating use, that increases the capacity for autonomous action rather than replacing it with managed consumption. Heifetz’s holding environment is the leadership practice that creates convivial conditions: holding the distress of not knowing, protecting the space for experimentation, giving the work back to the people who must do it. Weick’s small wins are the mechanism: concrete, visible changes that establish new practice before the old institutional logic can reassert itself. Drucker’s systematic abandonment is the discipline: regularly asking “if we were not already running this programme, would we start it now?”

The forward connection to the Deciding phase is direct. Illich diagnoses the pathology of institutional inversion: the point at which structures become counterproductive. Beer, who governs the Interaction lever in the Deciding phase, provides the cybernetic architecture that prevents or corrects the inversion. His Viable System Model ensures that each part of the organisation has the autonomy to respond to its environment while remaining coordinated with the whole. POSIWID (the Purpose Of a System Is What It Does) is Beer’s operationalisation of Illich’s diagnosis: if the system is producing programme artefacts, not learning, then the system’s purpose is programme artefacts, regardless of what the strategy says. Redesign the system, not the communication plan.

The deepest lesson Illich offers this series is that the institutional form itself, the programme, the centre of excellence, the governance framework, is not a neutral container for transformation. It is an active force that shapes what transformation can become. If the institution is convivial, it amplifies autonomous capability. If it has inverted, it replaces autonomous capability with institutional dependency. Every structure in the organisation is doing one or the other, at every moment, in every interaction. The leader’s task is to know which.

(An Organisational Prompt is something you can do now....)

Organisational Prompt

Run the iatrogenesis diagnostic on one programme, one initiative, one governance structure created to support your transformation.

Clinical: is the programme directly preventing the thing it was designed to enable? Are there teams that would adopt AI faster if the programme did not exist?

Social: has the programme redefined competence so that only programme-certified activity counts? Are there practitioners who have taught themselves effectively but are invisible because they are outside the framework?

Cultural: have people lost the disposition to learn independently? Do teams wait for permission, for the approved tool, for the centre of excellence to publish guidance, before trying something new? If you abolished the programme tomorrow, would your people know how to start?

If you find iatrogenesis at any level, the response is not to fix the programme. It is to ask Drucker’s question: “If we were not already doing this, would we start now?” And if the answer is no, to have the courage to stop, and to trust that the conditions for learning are more productive than the institution that has been consuming them.

Further Reading

Ivan Illich, Deschooling Society (1971). The foundational text. Short, radical, and immediately applicable beyond education. Freely available as a PDF.

Ivan Illich, Tools for Conviviality (1973). The more general statement: the distinction between convivial and manipulative institutions, and the criteria for assessing which a given institution has become. Freely available as a PDF.

Ivan Illich, Medical Nemesis: The Expropriation of Health (1976). The iatrogenesis argument. Demonstrates the pattern across domains: the institution replaces the activity it was designed to support with the consumption of its own services.

Paulo Freire, Pedagogy of the Oppressed (1970). The complementary critique. The “banking model” of education is the pedagogy of Illich’s manipulative institution. Problem-posing education is what convivial learning looks like in practice.

I write about the industry and its approach in general. None of the opinions or examples in my articles necessarily relate to present or past employers. I draw on conversations with many practitioners and all views are my own.

Then I started using Workflowy for one thing, and it quietly ate everything else.

The pitch in ten seconds

Workflowy is an outliner. You get an infinite nested list. That’s it. No databases, no templates marketplace, no “second brain” philosophy requiring a PhD in library science to set up.

You open it. You type. You indent. You’re done.

It sounds too simple to work. That’s exactly why it works.

What I actually keep in it

Tasks. Not via some bloated project management system; just bullets with #todo (and other tags) and @name tags to assign ownership. Workflowy has built-in task management, so when something’s done I mark it complete and it disappears from view. No ceremonies, no inbox-zero rituals, no weekly reviews that I definitely won’t do. Tag a person, tag a project, and the filters do the rest.

Meeting notes. Every meeting gets a bullet under the date in a journal node. Key points nest underneath. If something becomes a task, I tag it. The note and the action live in the same place, which means I actually find them again later.

Research. This is where Workflowy quietly punches well above its weight. When I’m working through a problem - a technology strategy, an architecture decision, a new framework I’m evaluating, and especially the articles for Substack - I think in outlines. Workflowy is an outline. There’s no friction between how I think and how the tool works. I dump, I rearrange, I nest, I collapse. The structure emerges as I go.

Two features make it especially good for research. First, mirror nodes : you can place a copy of any node in multiple locations and they stay perfectly synced. A finding that’s relevant to three different workstreams lives in all three, updated once. No copying, no staleness, no “which version is current?” anxiety. Second, Workflowy tracks backlinks : link to another node and it knows about the connection in both directions. Over time your research develops a web of cross-references that you didn’t have to manually maintain. It’s the connective tissue that turns a pile of notes into something you can actually navigate.

Writing. This article started as a Workflowy bullet. Most of my writing does. An outline is already a first draft if you squint. I expand bullets into paragraphs, rearrange sections by dragging them around, and export when it’s ready.

Reference material. API keys, account details, onboarding checklists, that one command I can never remember; all nested under a “Reference” node. It’s a personal wiki without any of the overhead of maintaining a personal wiki.

Reading notes. When I read something worth remembering, I drop a bullet with the title and nest my highlights and reactions underneath. Over time, this becomes a surprisingly useful personal library.

Journaling. Workflowy has genuinely good date awareness; type a date and it becomes a live, clickable reference. Combine that with tags and you get a powerful journaling system. Daily entries, weekly reflections, tagged by theme or mood, all searchable and all living alongside everything else. No separate app required.

Links. Lots of web links.

Kanban. And if you want a different view? Workflowy added a kanban-style board view, so any list can become a set of columns when that’s the better way to see things. The simplicity is still there, it’s just a list wearing a different tie.

Presentations. This one surprises people. Workflowy has a built-in presentation mode that turns any outline into a slide deck you can present directly from the app. Your outline is your talk structure, so there’s no exporting, no reformatting, no death-by-PowerPoint. I’ve used it for internal walkthroughs and it works remarkably well; you’re presenting your thinking in the same structure you used to develop it.

Why it sticks when others didn’t

Three reasons.

Speed. Workflowy is fast. Genuinely, distractingly fast. There’s no loading spinner, no sync delay, no “building your workspace” interstitial. You open it and you’re already typing. This matters more than any feature list, because the best productivity system is the one you actually open.

Zoom. Any bullet can become your entire view. Click into “Q2 Planning” and suddenly that’s your whole world - clean, focused, free of everything else. Click back out and it’s a bullet again. This is the single most underrated interaction pattern in productivity software. It means the same tool scales from a quick grocery list to a multi-year programme of work.

No opinions. Workflowy doesn’t care how you organise things. It doesn’t ship with six default views and a “getting started” tutorial that takes forty minutes. It’s a blank page that happens to support structure. You bring the system; it stays out of the way.

The objections I had (and why they faded)

“But it’s just a list.” Yes. And a spreadsheet is just a grid. The constraint is the point. A nested list is a universal data structure (you’ve heard of LISP, right?). Nearly everything you want to capture is either a sequence, a hierarchy, or both.

“What about files and media?” Workflowy lets you attach files directly - either as links or uploaded into the node itself. Drop a YouTube link and it renders inline. It’s not trying to be Google Drive, but it handles media and attachments better than you’d expect from something that looks this minimal.

“It won’t scale.” I have over 250,000 nodes in Workflowy spanning several years. No noticeable performance decrease. Search is instant. Tagging and filtering work. It scales better than most tools I’ve used precisely because the underlying model is so simple.

Give Claude Access to Your Workflowy

I built an MCP server that lets Claude read, search, and write to your Workflowy outline. It turns your Workflowy into an AI-accessible knowledge base - you just talk to Claude naturally and it handles the rest.

What it does

You say things like:

• “What’s overdue in my Tasks?”

• “Add a task under Office: Review Q2 budget”

• “Search my notes for anything tagged #review”

• “Give me a daily standup from my Projects”

• “Whats been on my reading list for a long time?”

Claude calls the right Workflowy tools behind the scenes - searching nodes, creating tasks, checking due dates, summarising projects. 23 tools in total, covering search, content creation, todo management, and project overview.

How it works

The server sits between Claude Desktop and the Workflowy API. It speaks MCP (Model Context Protocol), which is how Claude Desktop connects to external tools. You install it once, point Claude Desktop at it, and every conversation gets access to your Workflowy tree (recursively, if you have lots of nodes like I do).

Other Hacks

I also use it to synch notes made on my ReMarkable to the Workflowy node for that day. But that’s another story and another MCP server.

Go Here if You’re Interested

github.com/dromologue/workflowyMCP

(If this confuses you, you can either find a friendly engineer OR just point Claude at the repo and it will install it for you.)

PS.1… I do not and never have worked for or with Workflowy the company. I’m just a fan.

PS.2 … Other LLM clients are available, Like ChatGPT, CoPilot etc. It doesn’t have to be Claude, it’s just what I use.

While Taylor looked up from the shop floor to optimise the task, Henri Fayol looked down from the boardroom to optimise the enterprise. Writing in 1916, his premise was radical for its time but is now the water we swim in: management is a universal, teachable skill distinct from technical expertise. Together with Taylor’s task-level optimisation and Weber’s bureaucratic structures, Fayol’s enterprise-level integration produced the classical management model that dominated the twentieth century and that every subsequent thinker in this series has either refined or rejected.

Unlike Taylor, Fayol was studying the work of managers themselves. His principles were explicitly offered as flexible guidelines, not rigid laws; he warned against dogmatic application and insisted that “seldom do we have to apply the same principle twice in identical conditions.” This flexibility is routinely forgotten by the organisations that have hardened his guidelines into governance frameworks so rigid that Fayol himself would have rejected them. His most practically useful insights remain startlingly relevant to anyone trying to coordinate an AI transformation across organisational boundaries.

1. Unity of Direction: The Alignment Problem Technology Cannot Solve

Fayol distinguished between two principles that are often confused. Unity of command means each employee receives orders from one superior only. Unity of direction means there is one head and one plan for each group of activities with the same objective. Many organisations have solved the first. They fail spectacularly at the second.

The Data Science team optimises for model accuracy. Infrastructure optimises for cost and reliability. Product optimises for time-to-market. Each team meets its own targets while the system as a whole underperforms. Deming diagnosed exactly this: his ninth point, “break down barriers between departments,” addresses the coordination failure Fayol identified. But Deming’s solution is trust and shared purpose rather than hierarchical authority. Optimising the parts sub-optimises the whole, and the barriers between departments are not merely structural but psychological: each department develops its own culture, its own definition of quality, and its own defensive routines.

Stacey would add that unity of direction cannot be achieved through planning alone. If organisations are patterns of interaction, then the “one plan” Fayol prescribed is not a document that coordinates from above. It is a shared orientation that emerges through the quality of conversations between the people doing the work. Weick would agree: the plan’s value lies partly in the sensemaking process that produces it, not just in the document that results.

Bourdieu reveals what Fayol’s framework cannot model: power. Unity of direction assumes departments can be aligned through rational coordination. But departments are also fields where actors deploy their capital to maintain position. The Data Science team does not resist alignment because they fail to see the logic. They resist because alignment would require subordinating their priorities, which threatens the capital they have accumulated. The coordination problem is simultaneously a structural problem and an identity problem, and Fayol’s framework addresses only the first.

2. The Gangplank: Speed Without Accountability Is Noise

Fayol respected hierarchy for accountability, but he recognised it was too slow for operational reality. If a message from Department A had to travel up to the CEO and back down to Department B, the business would die of latency. His solution was the “gangplank”: authorised direct communication between peers across silos, with the condition that peers inform their superiors of what was decided. The gangplank gave speed. The accountability condition gave alignment.

Modern organisations have the gangplank without the condition. Cross-functional teams, messaging channels, working groups, direct email between peers: all gangplanks. But the accountability that Fayol insisted on has been lost. The result is shadow decisions, fragmented architecture, and choices made in threads that nobody else knows about until they collide with choices made in other threads.

Westrum’s typology illuminates why the gangplank works in some cultures and fails in others. In a generative culture, the gangplank is natural: information flows to where it is needed. In a bureaucratic culture, the gangplank requires formal authorisation, which reintroduces the latency it was designed to eliminate. In a pathological culture, the gangplank is dangerous: information shared laterally can be used as a weapon.

Giddens explains why the accountability condition is so hard to maintain. Daily communication practices are reproduced through practical consciousness, not through formal rules. People do not consult the governance policy before sending a message. They do what feels natural, and what feels natural is shaped by the habitus. Fayol’s condition, “inform your superiors,” requires conscious effort that works against how organisations actually communicate. This is why the condition is almost universally violated and the coordination failures Fayol predicted are almost universally observed.

3. The Governance Trap: When Everyone Is in Charge, Nobody Decides

Fayol was adamant: dual command is a perpetual source of conflict. Yet the modern matrix organisation, where you report to a functional lead and a delivery lead simultaneously, is the standard violation of his principle. Who owns AI risk? The CISO? The CTO? The Head of Legal? The AI Centre of Excellence? When everyone is in charge, no one is. Fayol would predict exactly the paralysis visible in enterprise AI adoption: each authority can veto, none can approve.

Heifetz provides the deeper diagnosis. The governance trap is not merely structural. It is an adaptive challenge masquerading as a technical one. The people in the matrix are managing the anxiety of a situation where nobody knows the right answer, and the matrix provides the appearance of distributed authority that allows everyone to defer the difficult decision. This is work avoidance: the organisation creates a structure that absorbs the anxiety without confronting it. Fayol’s unity of command is not just an efficiency principle. It is the structural condition that forces decisions to be made rather than endlessly deferred.

Peters would recognise this as bureaucratic ossification at its most destructive: the governance apparatus designed to manage risk becomes the primary risk, because it prevents the experimentation from which learning comes. Bateson’s levels framework diagnoses the epistemological cost: the matrix keeps the organisation at Learning I (navigating the governance) when the challenge demands Learning II (questioning whether the governance itself is the obstacle).

4. Initiative and Esprit de Corps

Fayol is often caricatured as a cold bureaucrat. This is a misreading. He listed initiative, “the ability to think out and execute a plan,” as a primary source of organisational strength and argued that a manager who inspires initiative is infinitely superior to one who cannot. He warned explicitly against dividing one’s own team, calling it “a grave sin” that destroys esprit de corps. He understood that technical structure is useless without social cohesion.

Deci and Ryan provide the psychological framework for Fayol’s intuition. Initiative requires autonomy: the felt experience of choosing rather than being controlled. Esprit de corps requires relatedness: connection to others that sustains commitment. Fayol grasped, without the vocabulary, that an organisation of rule-followers without initiative is brittle. Dweck’s mindset research connects initiative to beliefs about ability: in a growth mindset culture, initiative is valued because mistakes are learning. In a fixed mindset culture, initiative is punished because mistakes are evidence of inadequacy. Fayol’s principle, tolerate mistakes to encourage initiative, requires the cultural precondition Dweck describes.

5. Fayol’s Limits and What Must Be Added

Fayol must be read with his limitations visible. His framework assumes a stable environment where planning is possible and coordination can be achieved through regular conferences. Stacey provides the most fundamental challenge: you cannot plan in a complex domain because outcomes emerge from interactions you cannot predict. You cannot command because the people receiving commands hold the knowledge that determines how the command is interpreted. Normann challenges Fayol’s assumption that organisational boundaries are fixed: value is increasingly co-produced across boundaries that no single manager can command.

Yet Fayol remains indispensable. Coordination must happen. Direction must be unified. Command must be clear. Initiative must be cultivated. These are structural necessities, not optional features. The question is not whether Fayol’s functions are needed but how they can be achieved in conditions he could not have anticipated. Coordination through conversation rather than conference (Stacey). Direction through shared sensemaking rather than central planning (Weick). Command through adaptive leadership that gives the work back (Heifetz). Initiative through environments that satisfy autonomy, competence, and relatedness (Deci and Ryan). And esprit de corps through cultures where information flows generatively (Westrum). Fayol identified the problems. The modern thinkers identify what it actually takes to solve them.

(An Organisational Prompt is something you can do now....)

Organisational Prompt

Identify two teams critical to your AI initiative that sit in different reporting lines. Perhaps Data Science and Legal. Perhaps Platform Engineering and the business domain team. Ask two questions.

First: do they have a functioning gangplank? Can they communicate directly, without going through their respective hierarchies, when a decision needs to be made quickly? Second: do they share a single plan for the objective they both serve, or is each executing its own strategy that nominally supports the same goal?

If they have speed without alignment (gangplank but no unity of direction), they will produce fragmentation. If they have alignment without speed (unity of direction but no gangplank), they will produce latency. Fayol’s diagnosis is a century old. It is almost certainly operating in your organisation right now.

Further Reading

Henri Fayol, General and Industrial Management (1916/1949). The original text, still surprisingly readable. Pay attention to the nuance: these are principles to be applied with judgement, not laws to be followed mechanically.

Henry Mintzberg, The Nature of Managerial Work (1973). The empirical challenge to Fayol’s rational framework. What managers actually do bears little resemblance to Fayol’s tidy functions. Together they illuminate the gap between what management aspires to be and what it actually is.

W. Edwards Deming, Out of the Crisis (1986). Deming’s ninth point, “break down barriers between departments,” is Fayol’s coordination problem addressed through systems thinking rather than hierarchical authority.

I write about the industry and its approach in general. None of the opinions or examples in my articles necessarily relate to present or past employers. I draw on conversations with many practitioners and all views are my own.

Writing in 1959, Peter Drucker coined the term “knowledge worker” and spent the next five decades arguing that making knowledge work productive would be the defining management challenge of the twenty-first century. He was right. And AI has arrived to prove it.

Most organisations approach AI adoption as a technology problem: select the tools, build the infrastructure, train the staff. Drucker would have recognised this immediately as the wrong framing. The technology is not the constraint. The constraint is the same one he identified decades ago: we still do not know how to make knowledge workers productive, and AI has made the cost of that failure impossible to ignore. When the machine can generate code, draft documents, and synthesise research, the question that remains is the question Drucker placed at the centre of everything: what should the knowledge worker actually be doing?

1. Create a Customer, Not a Profit

Drucker’s most radical claim is deceptively simple: the purpose of a business is not to make profit. It is to create a customer. Profit is a necessary condition for survival, but it is not the purpose. The purpose is to create something that someone values enough to pay for.

This reframes what “success” means for AI adoption. If the purpose is profit, AI is justified by cost reduction: fewer developers, faster delivery, cheaper operations. If the purpose is to create a customer, AI is justified by something different: the ability to build things customers actually want, faster and more precisely than before. The distinction determines what you measure, what you invest in, and what you ask people to learn. Normann deepened this by dissolving the boundary between firm and customer entirely. In the value constellation, the customer is not a passive recipient but an active co-producer. Drucker’s “create a customer” anticipates this: the purpose is not to serve the customer but to create the conditions in which the customer can participate in value creation.

2. The Knowledge Worker Owns the Means of Production

Drucker’s knowledge worker concept is the foundation stone. A knowledge worker owns their means of production: their knowledge. Unlike a manual worker, the knowledge worker must define the task before they can do it. Their output is quality, not quantity. They cannot be supervised in the Taylorist sense because the work is invisible until it produces results. Autonomy is not a perk. It is a structural requirement.

This has a devastating implication. AI does not replace the knowledge worker. It depends on them. The quality of the AI output is entirely determined by the quality of the specification the knowledge worker provides. An imprecise specification produces useless code, regardless of how powerful the model. This is the specification problem. It is not a technology problem. It is a knowledge worker productivity problem. Drucker told us decades ago that solving it requires removing obstacles and enabling autonomy, not imposing controls and measuring outputs.

Taylor is the figure Drucker argued against. Taylor separated thinking from doing: the manager thinks, the worker executes. Drucker reunited them: in knowledge work, thinking is the doing. The specification writer cannot be separated from the specification. To impose Taylorist supervision on this process, to measure it by volume of specifications produced, is to destroy the very productivity you are trying to create.

Bateson’s levels framework reveals the epistemological depth. Drucker’s knowledge worker operates at Learning II: they must define the task, which means questioning the frame within which the task has been understood, not merely executing within it. Taylor’s worker operates at Learning I: they correct errors within a frame defined by management. AI transformation demands Learning II, because the task itself must be defined before the machine can act on it. The organisation that treats specification as a Learning I activity, a template to be filled in, a form to be completed, has misunderstood both Drucker and Bateson. The specification is not a form. It is the act of defining the task, and defining the task is the knowledge work.

Deci and Ryan provide the psychological evidence for Drucker’s structural argument. Knowledge work productivity requires autonomy, competence, and relatedness. When organisations respond to AI by imposing governance that dictates how AI may be used, by measuring compliance rather than quality, and by restructuring teams in ways that break relationships, they thwart all three needs. The predictable result is disengagement, interpreted as resistance, triggering further control, deepening disengagement. The cycle is vicious and self-reinforcing.

3. Management by Objectives: The Corrupted Ideal

Drucker invented Management by Objectives with a specific intent: to enable decentralisation and autonomy. If people understood the objectives clearly, they could determine for themselves how to achieve them. MBO was designed to replace command-and-control with trust-and-clarity.

What happened was the opposite. MBO was corrupted into top-down quotas, cascaded KPIs, and surveillance. The objectives were imposed, not negotiated. The “how” was specified along with the “what.” Drucker acknowledged the corruption: “MBO works if you know the objectives. Ninety percent of the time you don’t.”

The parallel to specification-driven development is exact. A specification, properly understood, defines what and why but leaves how to the AI and the team. This is MBO applied to software production. But if specifications become rigid instruction sets imposed by a planning function, if they become the AI equivalent of cascaded KPIs, the corruption is identical. The specification should liberate, not control. Deming went further: his eleventh point, “eliminate management by objective,” rejected what MBO had become. Deming argued that quotas address numbers, not quality, and that MBO without method is management by fear. The corruption Drucker lamented, Deming diagnosed as inherent to any system that specifies outcomes without providing the means to achieve them.

Weick would add that the objective’s value lies partly in the sensemaking process that produces it, not just in the document that results. The conversation about what the specification should say is where the real learning happens. The specification itself is a by-product.

4. Systematic Abandonment: The Discipline Nobody Practises

Of all Drucker’s ideas, systematic abandonment is the most underused and the most needed. The discipline is simple: regularly ask, “If we were not already doing this, would we start now?” If the answer is no, stop.

Every organisation has processes that exist because they were once necessary. The architecture review board that was essential when systems were handcrafted may be harmful when systems are generated from specifications and validated automatically. The separate testing phase that was necessary when code was written by humans may be redundant when validation is continuous. Yet nobody stops them, because stopping threatens the people whose roles depend on them. This is where Drucker meets Dekker. The resistance to abandoning outdated processes is locally rational: the people running them are protecting their livelihoods, their status, and their sense of contribution.

Giddens explains why abandonment is structurally difficult. The processes are not merely procedures. They are structures reproduced through daily practice: relationships, sources of status, patterns of meetings, ways of knowing where power sits. Bourdieu adds the embodied dimension: the dispositions acquired in the old world persist long after the conditions that produced them have changed. Heifetz would name the adaptive challenge: the people who must abandon the old processes are the people whose identity was built through them. That is a loss, and it must be named before it can be processed.

Weber provides the biggest framing. The processes that resist abandonment are the local expression of rationalisation: the deep commitment to making the world calculable and controllable through formal rules. Weber would predict that the organisation’s response to systematic abandonment will be to create a new process for deciding which processes to abandon, adding bureaucratic weight in the act of trying to remove it.

Drucker’s systematic abandonment is the refactoring discipline for organisational design. For every new AI governance requirement, ask: what existing requirement does this replace? If the answer is “none, it is additional,” you are adding weight, not enabling transformation.

5. The Effective Executive and the Transformation Leader

Drucker’s five practices of the effective executive translate directly to transformation leadership. Know where your time goes: Mintzberg studied what managers actually do and found it bears little resemblance to what they say they do. Focus on contribution: Heifetz would recognise this as distinguishing technical work from adaptive work. Make strengths productive: the domain expert who understands the business deeply is not deficient because they cannot code; they are the ideal specification writer. Dweck would add that this requires treating “strength” as developable, not static. First things first: the organisation that launches twelve AI pilots simultaneously will complete none. And make effective decisions: Drucker insisted decisions require disagreement, the deliberate cultivation of opposing views. If everyone agrees with the AI strategy, something important is being suppressed. Argyris would call this making the undiscussable discussable.

(An Organisational Prompt is something you can do now....)

Organisational Prompt

Take one governance process that every initiative must pass through. Ask the people who run it: “If this process did not exist, would we create it today, knowing what we know about AI-generated code and automated validation?”

If the answer is no, or if it is a long pause followed by a qualified yes, you have found process debt. Organisations accumulate process debt the same way they accumulate technical debt. The constraint is rarely that you need to add something new. It is that you need to stop doing something old. The freed capacity is the space in which learning becomes possible.

Further Reading

Peter Drucker, The Effective Executive (1967). Still the most practical book on management ever written. Short, clear, ruthlessly focused on what actually makes leaders productive.

Peter Drucker, Management Challenges for the 21st Century (1999). Where Drucker directly addresses knowledge worker productivity as the defining challenge. The argument that knowledge workers must manage themselves.

Peter Drucker, Innovation and Entrepreneurship (1985). The seven sources of innovation, including “the unexpected success” that organisations systematically ignore because it does not fit their plan.

W. Edwards Deming, Out of the Crisis (1986). Read alongside Drucker for the critique of what MBO became. Together they define the challenge: clarity of purpose without the surveillance apparatus that destroys the autonomy the purpose was supposed to enable.

I write about the industry and its approach in general. None of the opinions or examples in my articles necessarily relate to present or past employers. I draw on conversations with many practitioners and all views are my own.

This is not a compliance problem. It is a discovery problem. The official programme and the renegade team are both trying to learn what AI can do for the organisation. They are using fundamentally different methods. And the question of which one is producing better knowledge, and why, is not a question that governance frameworks are equipped to answer. It is a question that three philosophers of science spent the second half of the twentieth century arguing about, and their argument has more to say about your AI transformation than any maturity model on the market.

Thomas Kuhn, Imre Lakatos, and Paul Feyerabend are best understood as a conversation, not as three independent positions. Kuhn published The Structure of Scientific Revolutions in 1962, arguing that science advances not by steady accumulation but through revolutionary ruptures between incommensurable paradigms. Lakatos responded in 1970 with a framework for evaluating whether a research programme is genuinely progressive or merely defending itself. Feyerabend published Against Method in 1975, arguing that every methodological rule, however plausible, has been productively violated at some point in the history of science.

The three of them, together with Karl Popper (discussed previously) who chaired the pivotal 1965 London colloquium at which much of this was debated, produced between them the most penetrating account available of how communities of practitioners actually discover new things, adopt new practices, and resist the evidence that their current approach has stopped working.

Compressing three deeply original thinkers into a single article necessarily simplifies positions that each spent decades refining. Kuhn would insist that his concept of incommensurability is more nuanced than “paradigm shift” has become in popular usage; he later preferred the term “exemplar” and tried repeatedly to clarify what he actually meant. Lakatos would protest that his methodology of scientific research programmes deserves independent treatment; it is, after all, a sophisticated attempt to rescue rationality from Kuhn’s apparent irrationalism. Feyerabend would probably object to being included in any series that claimed to offer methodological guidance at all; he spent his career attacking precisely that enterprise. What follows treats the three as a conversation about how discovery actually works, and asks what that conversation means for organisations trying to decide what to do next.

1. Kuhn: Why Your Organisation Resists the Evidence That Its Paradigm Has Failed

Thomas Samuel Kuhn (1922-1996) was a physicist turned historian of science whose single book changed the vocabulary of intellectual life. Born in Cincinnati, he completed his PhD in physics at Harvard in 1949, then experienced a revelatory moment while teaching Aristotle’s physics to non-scientists: he realised that Aristotle was not simply wrong, but was operating within a different framework of understanding. This insight, that scientific communities work within shared frameworks that determine what counts as a good question, a good method, and a good answer, became the foundation of The Structure of Scientific Revolutions.

Kuhn’s model is deceptively simple. Science does not progress by the steady accumulation of facts. It proceeds through a cycle: long periods of normal science, punctuated by revolutionary ruptures that replace one paradigm with another.

A paradigm, properly understood, is not just a theory. It is a constellation of shared commitments: theoretical assumptions, accepted methods, standard instruments, exemplary problem-solutions, and criteria for what counts as a legitimate question. Within a paradigm, scientists engage in puzzle-solving: extending the framework, filling in details, resolving small discrepancies. Normal science is conservative, cumulative, and enormously productive. It is productive precisely because it is focused: scientists are not debating fundamentals; they are solving specific problems with shared tools.

Anomalies are puzzles that resist solution. Initially, they are absorbed: explained away, delegated to specialists, or filed as “known issues” for future attention. This is not stupidity. It is rational behaviour within a paradigm, because most anomalies are eventually resolved within the existing framework. The scientist who abandons a productive paradigm at the first counter-instance is not bold; they are flighty. The paradigm has earned its loyalty through decades of accumulated success.

Crisis occurs when anomalies accumulate to a point where confidence in the paradigm erodes. The established practitioners may continue to defend the framework, but younger scientists or outsiders begin taking the anomalies seriously. Revolution occurs when a new paradigm emerges that resolves the persistent problems. The shift is not purely rational: it involves social, psychological, and generational factors. Different paradigms may be incommensurable: there is no neutral, paradigm-independent standard by which to compare them. Scientists in different paradigms may literally interpret the same data differently, because the paradigm determines what the data means.

After the revolution, science rewrites its own history. Textbooks present the transition as inevitable progress, smoothing the rupture into a narrative of cumulative development. The resistance, the political battles, the careers that were destroyed and the careers that were made are all erased. The next revolution will face the same dynamics, but the institutional memory of how the last one actually happened will have been lost.

The organisational parallels are immediate.

Normal science is what Argyris calls single-loop learning: operating within existing assumptions, solving puzzles defined by the current framework. Most organisational activity is normal science, and it should be. The team that processes invoices, manages the deployment pipeline, or handles customer support is doing productive puzzle-solving within an established paradigm. This is not a failure. It is how organisations get things done.

The problem begins when anomalies appear. In AI transformation, the anomalies are everywhere: use cases that do not fit the existing operating model, value propositions that do not materialise under the current governance structure, skills gaps that training does not close, teams that produce better results by ignoring the approved methodology than by following it. The organisational response to these anomalies is exactly what Kuhn described for science. They are absorbed. “The tool wasn’t the right one.” “The team wasn’t ready.” “We moved too fast.” “We need more governance.” Each explanation protects the paradigm from the implication that the paradigm itself might be the problem.

Dweck illuminates the individual response. In a growth mindset, the struggle of learning new tools and methods is experienced as the normal sensation of learning. In a fixed mindset, the same struggle is evidence of inadequacy. The engineers most resistant to AI-augmented development are often those whose identity and career capital are most deeply invested in the current paradigm. This is not irrationality. It is the entirely predictable response of a person whose sense of professional competence is being challenged by a paradigm shift they did not ask for and cannot yet see the other side of.

And Weick explains what happens when the paradigm actually breaks down. His analysis of the Mann Gulch disaster is Kuhn’s crisis played out in twelve minutes: when the smokejumpers’ framework collapsed, when a routine fire became something they could not interpret, they could not make sense of their foreman’s escape fire because the identity and role structures that made interpretation possible had dissolved. They dropped into individual flight because the social structure that held them together as a team had collapsed. This is what a cosmology episode looks like, and it is what the early stages of a genuine paradigm shift feel like from the inside: not exciting, but terrifying.

Kuhn’s deepest insight for this series is about the generational dimension. The people who built the current paradigm have the deepest investment in it. They are the least likely to see anomalies as signals of paradigmatic failure rather than local problems. Max Planck is often credited with the observation that science progresses one funeral at a time; Kuhn would not have been so blunt, but his analysis supports it.

2. Lakatos: How to Tell Whether Your AI Programme Is Progressing or Dying

Imre Lakatos (1922-1974) had one of the most extraordinary lives of any philosopher in the twentieth century. Born Imre Lipschitz in Debrecen, Hungary, he survived the Holocaust by adopting false names; his mother and grandmother perished in Auschwitz. After the war, he became a high-ranking official in the Hungarian Ministry of Education, then was imprisoned during the Stalinist purges from 1950 to 1953. He fled Hungary after the 1956 Soviet crackdown, completed a PhD at Cambridge, and joined the London School of Economics, where he became Popper’s most important student and eventually his most sophisticated critic. He died suddenly of a brain haemorrhage in 1974, at fifty-one. The planned companion volume to Feyerabend’s Against Method, which Lakatos was to write as “For Method,” was never completed.

Lakatos saw two unacceptable positions in the philosophy of science, and he saw them as having direct political implications. Popper’s naive falsificationism demanded that scientists discard a theory at the first counter-instance, which bore no resemblance to how science actually worked and would have destroyed most of its greatest achievements. Kuhn’s account, in which paradigm shifts are determined by social psychology rather than logic, was worse: if even in science there is no rational basis for choosing between theories, Lakatos argued, then in politics “truth lies in power.” He called Kuhn’s position “mob psychology” and saw it as intellectually dangerous. Having survived both Nazism and Stalinism, Lakatos was not inclined to be relaxed about the relationship between rationality and power.

His solution was the methodology of scientific research programmes. The unit of scientific evaluation, Lakatos argued, is neither the individual theory (Popper) nor the monolithic paradigm (Kuhn), but the research programme: a sequence of theories that share a common hard core.

The hard core is the set of fundamental assumptions that define the programme. Scientists treat these as irrefutable by methodological decision; they do not abandon them in the face of counter-evidence. Newton’s three laws of motion formed the hard core of Newtonian physics. The hard core of Agile is something like: iterative delivery, self-organising teams, customer collaboration, and responding to change over following a plan. The hard core of some enterprises’ AI strategy, if you excavate it, is something like: “AI will automate existing processes and reduce headcount costs.”

The protective belt is everything else: auxiliary hypotheses, implementation details, specific practices, tooling choices, team structures, governance frameworks. When a prediction fails, the scientist adjusts the protective belt, not the hard core. This is rational behaviour, not intellectual cowardice. Most failures are in the protective belt: the implementation was wrong, the timing was off, the team lacked the right skills, the measurement was inadequate. Adjusting the protective belt while preserving the hard core is how research programmes develop.

The question is whether the adjustments are progressive or degenerating.

A progressive research programme is one where successive adjustments generate novel predictions that are at least occasionally confirmed. Each iteration opens new territory. The programme is learning: not just defending itself, but discovering things it did not know before. A degenerating programme is one where successive adjustments serve only to protect the hard core from refutation, without predicting anything new. The modifications are purely defensive. The programme explains what has already happened but generates no novel capability.

This distinction is the most practically useful concept in this entire article, and it maps directly onto how organisations should evaluate their AI programmes.Is this programme progressive or degenerating?

The answer depends on a single question: are the adjustments generating novel capability, or are they merely explaining each successive failure?

If the specification template produced genuine clarity about what the system should do, if the data quality initiative revealed structural problems in how the organisation manages its information, if the knowledge capture exercise created shared understanding of business rules that had previously been tacit, then the programme is progressive. The failures were productive. Each adjustment opened new territory. The organisation now knows things it did not know before, and those things are useful beyond the original AI testing initiative.

If the specification template was filled in mechanically without improving actual clarity, if the data quality initiative produced a report that nobody acted on, if the knowledge capture exercise generated documents that live in a SharePoint graveyard, then the programme is degenerating. Each adjustment explained the last failure without generating anything new. The hard core (”AI automates testing”) was protected from refutation, but the organisation learnt nothing.

The warning sign of a degenerating programme is unmistakable once you know what to look for: the primary output of the programme is strategic narrative rather than deployed capability. Decks, frameworks, maturity models, roadmaps, governance structures, risk assessments, and vendor evaluations proliferate. The steering committee meetings are full of activity reports. But the actual novel capability, the thing the organisation can do now that it could not do before, is absent.

Lakatos also explains why “fail fast” is too simple a prescription. Sometimes a programme is in what Lakatos called a “bad patch”: a period where no empirical progress is visible, but where the theoretical development, the deepening of understanding, the refinement of the model, is setting the stage for a breakthrough. Abandoning a programme at the first difficulty is Popperian naivety; the programme may be doing important work that has not yet yielded visible results. The art, and it is an art rather than a science, is distinguishing a bad patch from terminal degeneration. Lakatos provided no algorithm for this. He acknowledged that the judgement is retrospective and fallible: you can only know with confidence whether a programme was progressive or degenerating after the fact. In real time, the leader must make a judgement call.

This connects directly to Heifetz. A technical challenge, in Heifetz’s terms, is a problem in the protective belt: the implementation needs adjusting, but the fundamental direction is sound. An adaptive challenge requires changing the hard core: the assumptions that define the programme are no longer adequate. The most common leadership failure that Heifetz diagnoses, treating adaptive challenges as technical problems, is in Lakatosian terms the error of endlessly adjusting the protective belt when the hard core itself needs to change. The leader who responds to repeated AI pilot failures by commissioning yet another governance review is making protective belt adjustments to a programme whose hard core may be wrong.

3. Feyerabend: Why the Renegades Are Your Best Source of Evidence

Paul Karl Feyerabend (1924-1994) is the most misunderstood philosopher in this article, and possibly the most important. Born in Vienna, he served in the German army during the Second World War and was wounded on the Eastern Front in 1943 by three bullets, one of which struck his spine. He used a walking stick and lived in pain for the rest of his life. He studied physics and philosophy in Vienna, initially planned to study with Wittgenstein at Cambridge, but Wittgenstein died before Feyerabend arrived. He studied instead with Popper at the London School of Economics and began as a committed Popperian falsificationist, before spending the next two decades dismantling the position from within.

His career was spent mostly at UC Berkeley, where he was a legendary lecturer: provocative, funny, and deeply serious beneath the provocation. Against Method (1975) made him the most controversial philosopher of science of his generation. He was labelled “the worst enemy of science,” a characterisation he both resented and cultivated. His close friendship with Lakatos, documented in their extraordinary correspondence (published posthumously as For and Against Method), reveals a thinker far more nuanced than the popular image of the “anything goes” anarchist. He later said of Against Method: “I often wished I had never written that fucking book.” He felt his irony and playfulness were systematically misread as relativism.

Feyerabend’s central argument is not that science has no method, or that all methods are equally good, or that evidence does not matter. His argument is that there is no single scientific method whose rules are universally valid. Every methodological rule, however plausible, has been productively violated at some point in the history of science. The conclusion is not that rules are useless, but that the imposition of fixed methodological rules constrains discovery rather than enabling it. The German title of the book, Wider den Methodenzwang, translates as “Against the Forced Constraint of Method.” The target is not method itself but methodological tyranny.

His primary case study is Galileo’s defence of heliocentrism. Galileo did not follow the accepted methodological rules of his time. He used propaganda and rhetoric alongside evidence. He introduced the telescope as an instrument of observation when the telescope’s reliability was itself unproven. He advanced a theory that contradicted the well-established evidence of the senses: the earth does not appear to move. By the methodological standards of his day, Galileo was doing bad science. By the standards of history, he was doing the most important science of his century.

Feyerabend draws two principles from this and similar cases.

• First, counterinduction: deliberately developing hypotheses that contradict well-confirmed theories and well-established results. The consistency condition, which demands that new ideas agree with accepted theories, is unreasonable because it preserves the older theory, not the better theory. It protects incumbency, not truth. Hypotheses that contradict established thinking generate evidence that no amount of work within the existing framework can produce. You cannot discover what is wrong with the paradigm by working within it.

• Second, proliferation: maintaining multiple competing approaches rather than converging on a single methodology. Competing theories illuminate each other’s weaknesses in ways that no single theory, however well-tested, can reveal about itself. A monopoly of method is as dangerous as a monopoly of power.

This is not relativism. Feyerabend is not saying all methods are equally good. He is saying that you cannot know in advance which method will produce the breakthrough, and that the premature elimination of alternatives is the most reliable way to prevent discovery. The proliferation of approaches creates the conditions for informed comparison. The governance framework that insists on a single approved methodology eliminates those conditions.

The organisational implications cut deep, and they cut in two directions simultaneously.

The first direction extends Tom Peters’ argument against bureaucracy. The AI governance framework, the Centre of Excellence, the approved vendor list, the mandated methodology: these are not neutral quality controls. They are the organisational equivalent of Feyerabend’s methodological constraint. They determine what approaches are permitted, what evidence is recognised, and which voices are heard. They protect the incumbent paradigm and suppress alternatives. The team on the fourth floor that is using Claude without approval is doing what Galileo did with his telescope: generating evidence that the formal process cannot produce, using tools whose value the establishment has not yet recognised, and getting results that embarrass the official programme.

Stacey would agree entirely. Real organisational learning happens in the informal, the unplanned, the conversations that governance does not capture. The teams quietly using AI outside the approved framework are not insubordinate. They are the emergent strategy trying to tell you where the value is.

The second direction provides the essential corrective. Feyerabend’s proliferation must eventually produce evidence. “Anything goes” at the level of exploration does not mean “anything goes” at the level of evaluation. The renegade team’s results must be assessed. The alternative approaches must be compared. The organisation cannot remain permanently in a state of methodological anarchy; at some point, it must converge. The question is when, and on what basis, and Feyerabend provides less guidance on this than the practitioner needs.

This is where this series’ section emphasis on getting to clarity and specification-driven development provides the discipline that Feyerabend’s anarchism requires but does not supply. Popper’s insistence on falsifiable conjectures, Drucker’s insistence on defining the task, Evans’ insistence on ubiquitous language within bounded contexts: these are the mechanisms by which exploratory freedom is converted into evaluable results. The organisation that gives teams freedom to experiment must also insist that experiments produce testable specifications, that results are measured against those specifications, and that the comparison between approaches is honest. Freedom without evaluation is waste. Evaluation without freedom is stagnation. The synthesis requires both.

4. The Integrated Diagnostic: How the Three Questions Work Together

Each thinker provides a different lens for the same organisational challenge, and the three lenses compose into a single diagnostic.

• Kuhn asks: what is your paradigm, and is it in crisis? The paradigm is the set of assumptions, methods, and standards that govern how your organisation currently operates. Most of the time, normal operations within the paradigm are productive and should not be disrupted. The question is whether the anomalies accumulating around your AI transformation, the failures, the surprises, the results that do not match the predictions, are local problems within the paradigm or signals that the paradigm itself has stopped working. If every AI pilot that succeeds does so by violating the approved methodology, the anomaly is not in the pilot. It is in the methodology.

• Lakatos asks: is your programme progressive or degenerating? Identify the hard core of your AI strategy: the assumptions you treat as irrefutable. Then examine the adjustments you have made in the last twelve months. For each adjustment, ask the diagnostic question: did this generate a novel capability, or did it merely explain the previous failure? If the adjustments are opening new territory, the programme is progressive, and persistence is rational. If the adjustments are purely defensive, the programme is degenerating, and the rational response is not to adjust the protective belt again but to question whether the hard core is wrong.

• Feyerabend asks: are you allowing yourself to see the alternatives? In every organisation, someone is solving the problem differently. They may be using unapproved tools, ignoring the governance framework, or pursuing an approach that the strategy deck does not mention. The question is whether the organisation treats these people as threats or as sources of information. If the response to unsanctioned success is governance enforcement, the organisation is practising methodological constraint: protecting the official programme from the evidence that would reveal its limitations. If the response is curiosity and investigation, the organisation is practising the pluralism that discovery requires.

The three questions are not independent. They are three perspectives on a single phenomenon: the relationship between an established way of working and the evidence that it needs to change. Kuhn explains why the established way is so resistant to evidence. Lakatos provides criteria for evaluating whether the resistance is rational persistence or degenerate defence. Feyerabend insists that the evidence itself is being shaped by which methods the organisation permits.

Together, they reveal the deepest pattern in transformation failure. The organisation launches a programme with a hard core that reflects the current paradigm. Anomalies appear. The protective belt is adjusted. The adjustments explain each failure without generating novel capability. The programme degenerates. Meanwhile, the renegades operating outside the programme are generating genuine evidence, which the governance framework prevents the programme from seeing. The programme responds to its own degeneration by producing more governance, more methodology, more strategic narrative: exactly the methodological constraint that Feyerabend identifies as the primary obstacle to discovery. The organisation fails not because it lacked a method, but because it refused to examine what its method was preventing it from learning.

Kahneman would add the cognitive dimension: the leader’s System 1 generates a coherent narrative that makes the degeneration invisible. The programme feels productive because it is producing things: reports, frameworks, governance structures. Confidence feels like evidence. The losses entailed by admitting that the hard core is wrong, the sunk costs, the careers invested, the reputations attached, loom larger than the gains from changing direction. This is loss aversion in its purest institutional form, and it is why Heifetz insists that the most important thing a leader can do is name the loss: to say, clearly and compassionately, that the current approach is not working, that changing direction will cost something real, and that the cost of not changing is higher.

5. The Generational Problem and the Question of Adoption

There is one more dimension of this argument that organisations routinely ignore, and it may be the most consequential.

Kuhn observed that paradigm shifts often require a generational transition. The practitioners whose careers were built within the old paradigm frequently do not convert to the new one. They retire, they are sidelined, or they continue to practise in the old mode within diminishing enclaves. The new paradigm is adopted primarily by younger practitioners who were trained within it, or by outsiders who have no investment in the old one.

This observation is profoundly uncomfortable for anyone leading organisational transformation, because it implies that some of the people you are asking to change will not change. Not because they are stupid, lazy, or resistant. Because the paradigm is not just something they believe; it is the structure of their professional identity, their sense of competence, their understanding of what makes them valuable. We have discussed this at length through Giddens, Bourdieu etc.

Seligman’s learned helplessness adds the historical dimension. In organisations with a long history of failed change programmes, the veteran practitioners have learnt through repeated experience that resistance outlasts the initiative. They have survived Agile. They have survived DevOps. They have survived cloud transformation. Each time, the language changed but the structures persisted. Their reasonable expectation is that AI transformation will follow the same pattern: absorb the tools, preserve the structures, capture the terminology, and wait for the next initiative. The learned helplessness is rational. It is based on accurate observation of institutional history. And it is the single most powerful predictor of transformation failure.

The Feyerabendian response to this problem is not to force the conversion, but to ensure that the alternative approaches are visible and productive. The team on the fourth floor does not need to convince the established practitioners that AI-augmented development is superior. They need to produce results that make the superiority visible.

Bandura’s research on self-efficacy confirms this: the most powerful source of belief in one’s own capabilities is mastery experience, actually doing the thing and succeeding. A live demonstration where a sceptical engineer watches AI generate working code from a well-written specification is a mastery experience. It bypasses intellectual debate entirely. It creates the possibility of conversion not through argument but through evidence.

But the evidence must be allowed to exist. And this returns us to Feyerabend’s central point: the governance framework that prevents the team on the fourth floor from producing their results is not protecting quality. It is preventing the evidence that would force the paradigm to change.

(An Organisational Prompt is something you can do now...)

Organisational Prompt

Kuhn, Lakatos, and Feyerabend each provide a different diagnostic lens for the same challenge: how do you know whether your transformation is actually working, or whether you are protecting a degenerating programme with increasingly elaborate defensive adjustments? Lakatos’ test on generative programmes is helpful.

List the last three adjustments made to the initiative: changes in scope, tooling, team, governance, vendor, or approach. Ask, did this adjustment generate a new capability we did not have before, or did it merely explain the previous failure? Is the programme degenerating?

Further Reading

Thomas Kuhn: The Structure of Scientific Revolutions - The book that changed how we think about how science changes. Read the Postscript to the second edition as carefully as the text itself; it is where Kuhn clarifies what he actually meant by “paradigm,” and it is considerably more nuanced than the popular usage suggests.

Thomas Kuhn: The Essential Tension - The title essay argues that science requires a productive tension between tradition and innovation. Read it alongside Mintzberg on the relationship between craft and analysis, and alongside the series’ recurring argument that neither permanent revolution nor permanent normal science is adequate.

Imre Lakatos and Alan Musgrave (eds.): Criticism and the Growth of Knowledge - The proceedings of the 1965 London colloquium. Contains Kuhn’s statement, Lakatos’s response, Feyerabend’s critique, and Kuhn’s reply. This is the volume where the debate actually happened, and reading it is like watching four of the twentieth century’s most formidable minds argue about the foundations of rational inquiry. Essential for anyone who wants to understand the positions rather than the caricatures.

Imre Lakatos: The Methodology of Scientific Research Programmes: Philosophical Papers Volume 1 - Contains “Falsification and the Methodology of Scientific Research Programmes,” the essay that introduced the hard core, the protective belt, and the progressive/degenerating distinction. Dense but rewarding, and the most practically useful framework in the philosophy of science for anyone evaluating a strategic programme.

Paul Feyerabend: Against Method - Read the Galileo chapters closely and the rest for the argument’s texture. Read the German subtitle (”Against the Forced Constraint of Method”) as a corrective to the popular misreading. Feyerabend is not against method. He is against the imposition of a single method on everyone, and his case is stronger than most of his critics have acknowledged.

Matteo Motterlini (ed.): For and Against Method: The Lakatos-Feyerabend Correspondence - The letters between Lakatos and Feyerabend reveal a friendship of extraordinary intellectual intensity, warmth, and mutual respect. Reading them dispels the myth of Feyerabend as an irresponsible provocateur and reveals Lakatos as a more conflicted thinker than his published work suggests.

Disclaimer

I write about the industry and its approach in general. None of the opinions or examples in my articles necessarily relate to present or past employers. I draw on conversations with many practitioners and all views are my own.

Tom Peters is the anti-bureaucrat. While Taylor optimised the machine and Fayol coordinated it, Peters wants to liberate the humans trapped inside it. His premise, first articulated in In Search of Excellence and radicalised across four decades of subsequent work, is a direct assault on the rational-analytic model of management: excellence does not come from perfect planning. It comes from a bias for action, closeness to the work, and the release of human energy from the dead hand of procedure.

Peters’ work has methodological problems. Several of his “excellent” companies subsequently failed. His research was retrospective and based on reputation as much as data. His later work tips from provocation into evangelism. These criticisms are real. But Peters provides something the more rigorous thinkers in this series sometimes miss: the emotional and motivational dimension of organisational life, without which no transformation framework, however analytically precise, will produce any change at all. The energy to act is not a soft concern. It is the fuel.

1. A Bias for Action: Why Doing Beats Analysing

Peters’ most enduring insight is his simplest. Excellent organisations act. Mediocre organisations analyse. The disease of the mediocre organisation is analysis paralysis: substituting the appearance of rigour for the reality of learning.

Weick provides the cognitive theory: action precedes understanding. Stacey provides the complexity science: in complex domains, you must probe before you can sense. Mintzberg provides the empirical evidence: most realised strategy is emergent. Peters arrived at all of these conclusions decades before the academic frameworks were published. He expressed them as practitioner wisdom rather than theory, and this is both his strength and his limitation. The strength is emotional force. The limitation is that undisciplined action produces noise, not signal. Snowden’s probe-sense-respond includes the “sense” step for a reason. Peters provides the energy for the probe. Weick provides the discipline for the sense. The organisations that learn combine both.

The barrier to action is not individual timidity. It is structural obstruction. The governance framework that requires a risk assessment, a business case, a steering committee review, and an approved pilot before anyone touches a new practice is not malice. It is the process that was created to support the work, which has become the work that must be supported. When this inversion is complete, the organisation serves the governance rather than the governance serving the organisation. Every layer of approval between a team and an experiment is a structural barrier to the learning that transformation demands. Peters would strip the layers. Bateson would diagnose the result: the governance apparatus locks the organisation at Learning I (optimising within the existing frame) by preventing the action that would produce Learning II (questioning whether the frame is right).

For AI transformation, the organisation that commissions a twelve-month readiness assessment before anyone touches a tool is performing the appearance of diligence while systematically preventing the only learning that produces understanding. A live demonstration, a team writing a specification and watching AI generate working code, bypasses intellectual debate. It is what Weick would call a sensemaking event and what Peters would call getting off your arse.

2. Close to the Work

Peters scorned the manager who leads from a spreadsheet. He championed Management by Walking Around: the insistence that leaders must be visible, accessible, and connected to the front line where the work happens. This connects to the proximity probe that runs through the series. Specification quality is directly proportional to domain proximity. Specifications written by people who understand the problem deeply, who have Giddens’s practical consciousness of the domain, who know the edge cases because they live with them daily, are qualitatively different from specifications written by analysts working from requirements documents three levels removed from the actual work.

Kahneman explains why proximity matters at the cognitive level. The further you are from the work, the more WYSIATI operates: the coherent story in the leader’s head suppresses awareness of what the story leaves out. The leader reading a dashboard has a simpler, more confident, and less accurate picture than the leader who has sat with a team and watched them struggle. Peters provides the prescription: go and see. Normann provides the strategic frame: question whether your map describes the landscape you actually inhabit. Bourdieu provides the structural warning: the leader whose habitus was formed at a distance from the work will generate distance-from-work responses automatically. Proximity is not a technique to be applied. It is a disposition to be formed, and it is formed by sustained practice, not by a single management walkabout.

3. Loose-Tight: The Governance Paradox Resolved

Peters’ most sophisticated insight is that excellent organisations are simultaneously centralised and decentralised. They are tight on core values and standards, and loose on methods and implementation. Everyone knows what matters. Nobody is told how to achieve it.

This resolves the central tension in AI governance: how do you control AI without killing innovation? Be tight on the specification, the intent, the contract, the validation criteria, and loose on the implementation method. Define what the system must do, what constraints it must respect, what quality standards it must meet. Then leave the how to the team’s judgement. This is Drucker’s Management by Objectives as originally intended: clear objectives that liberate autonomy rather than constrain it. And it is the opposite of what most organisations implement. Most AI governance is loose on outcomes (unclear about what the output must achieve) and tight on methods (prescribing tools, approval processes, review boards). Teams comply with the method while producing outputs that fail to meet any coherent standard.

Stacey would recognise this as the organisation’s defence against the anxiety of not being in control. The governance framework exists not because it produces better outcomes but because it makes leaders feel they are managing the risk. The specification, by contrast, actually manages the risk by making intent explicit and validation automated.

4. The Destruction of Accretion

Most organisations respond to AI by adding governance layers without removing anything else. Every new requirement is additional. Nothing is abandoned. Peters would identify this as the central pathology: bureaucratic accretion, the progressive layering of process upon process until the weight of governance prevents the movement it was designed to guide.

Weick’s heavy tools metaphor applies: the organisation refuses to drop the tools that are slowing its escape, because the tools are identity markers. The architecture review board that adds three weeks to every initiative was created to solve a real problem. But the problem may have changed, and nobody has asked whether the board still serves the work or whether the work now serves the board. This connects directly to one of the observable probes in this series: can the organisation stop doing what no longer works? Most cannot, because the activities that should be stopped have accumulated constituencies whose identity depends on their continuation.

Drucker’s systematic abandonment is the disciplined version of Peters’ instinct: regularly ask “if we were not already doing this, would we start now?” If the answer is no, stop. The freed capacity is the space in which exploration becomes possible. Peters provides the courage to ask the question. Drucker provides the discipline to act on the answer.

5. Excellence Requires Passion

Peters insists that excellence is an emotional state. It requires passion, pride, and ownership. Bureaucracy systematically destroys all three. Procedure replaces intrinsic motivation with compliance. Initiative gives way to obedience. Ownership gives way to box-ticking. This is the managerial expression of what Deci and Ryan demonstrated: when autonomy, competence, and relatedness are thwarted, intrinsic motivation collapses, and no amount of extrinsic reward compensates.

This raises a question most AI transformation programmes never consider: does the new way of working generate passion? Developers experience flow while coding. Csikszentmihalyi’s conditions, clear goals, immediate feedback, challenge matched to skill, must be met by the new practice. The specification-generate-validate loop has flow potential: the specification provides the clear goal, the generated output provides immediate feedback, and increasingly complex specifications provide progressive challenge. But the potential is realised only if the work feels meaningful, not merely efficient.

Seligman’s learned helplessness is the condition that accumulates when passion has been repeatedly crushed by institutional indifference. The teams that have sat through previous transformations, watched their enthusiasm absorbed by governance, and learned that caring is not rewarded, will not be moved by another strategy deck. They need a different kind of evidence: a mastery experience, in Bandura’s terms, that reconnects effort with outcome. Peters intuited this: his “show me” is the practitioner’s version of Bandura’s mastery experience. The live demonstration that produces a result people can see, argue about, and build on is worth more than any amount of strategic communication.

6. Disciplined Liberation

Peters must be read with his limitations visible. His bias for action can become a bias against thinking. His anti-bureaucratic passion can tip into anti-structural nihilism. Stacey would agree that formal structure is less important than interaction patterns but would add that you cannot destroy structure and expect better patterns to emerge. Coordination, accountability, and alignment require formal mechanisms. The question is not structure versus no structure but what kind of structure enables both action and learning.

The synthesis this series points toward is what might be called disciplined liberation: Peters’s energy combined with Drucker’s discipline, Weick’s sensemaking rigour combined with the emotional commitment that makes people care enough to do the sensemaking well, the structural awareness of why organisations resist change (Giddens, Bourdieu, Argyris) combined with the refusal to accept that resistance as the final word. Normann’s conceptual elegance combined with Peters’s insistence that elegance without action is decoration. The discipline of the specification combined with the liberation of the people who write it.

(An Organisational Prompt is something you can do now....)

Organisational Prompt

Cancel your next AI steering committee meeting. Instead, go sit with a team that is trying to use AI to solve a real problem. Watch them work for thirty minutes. Do not speak for the first twenty. Then ask one question: “What is the most pointless process standing between you and shipping this?”

Listen to the answer. Then ask yourself: does that process serve the work, or does the work serve the process? If the answer is the second, you have found the structural obstruction that no amount of strategy will overcome. Remove it. See what happens. The freed capacity is where learning lives.

Further Reading

Tom Peters and Robert Waterman, In Search of Excellence (1982). The book that started the revolution against rational-analytic management. Read it for the energy and the original insight, not for the methodology.

Tom Peters, Thriving on Chaos (1987). The shift from “excellence” to “survival” through radical adaptability. More relevant now than when it was published.

Tom Peters, Liberation Management (1992). The blueprint for the networked, de-bureaucratised organisation.

Tom Peters, Re-imagine! (2003). The most passionate and most polarising of Peters’s books. Read it for the conviction that bureaucracy is not merely inefficient but morally intolerable.

I write about the industry and its approach in general. None of the opinions or examples in my articles necessarily relate to present or past employers. I draw on conversations with many practitioners and all views are my own.

Watch the room during the strategy town hall. Really watch it. The senior technology leader presents the vision. Heads nod. Questions are asked: the right kind of questions, calibrated to demonstrate engagement without exposing ignorance. Afterwards, in the corridor, in the Teams DM, in the pub, a different conversation occurs. “This will never work.” “They have no idea what we actually do.” “I’ll wait this one out.” The public performance and the private assessment are not merely different. They are structurally opposed. And everyone in the room knows it. Including, probably, the person presenting.

Erving Goffman, the Canadian-born sociologist who spent decades observing the micro-mechanics of human interaction, explains what is happening here. It is not hypocrisy. It is not cynicism. It is the ordinary, necessary, deeply human work of managing the impression one gives in different social situations. Goffman called it impression management, and his dramaturgical framework remains the sharpest account we have of why the gap between what organisations say and what they do is so persistent, so universal, and so resistant to the interventions that every other thinker in this series prescribes.

1. Front Stage and Back Stage

Goffman’s most immediately recognisable contribution is the distinction between front stage and back stage. Front stage is where performances are given: the town hall, the steering committee, the sprint review. Here, people present a version of themselves that conforms to the expectations of the audience. Back stage is where the mask comes off: the private channel, the team retrospective with the door closed, the lunch table, the car park after the meeting.

The maintenance of the front stage is a collaborative achievement. Everyone participates. The leader performs confidence in the strategy. The engineering leads perform willingness to adopt. The product managers perform curiosity about the new tooling. And each party tacitly agrees not to expose the others’ performances, because exposing someone else’s front stage risks having your own exposed in return. Goffman called this dramaturgical loyalty: the mutual agreement to sustain the definition of the situation, even when everyone privately doubts it.

Bourdieu’s habitus explains what people carry in their bodies. Goffman explains what they do with their faces. These are complementary mechanisms. The habitus produces practice below the threshold of awareness; impression management produces performance above it. The developer whose habitus generates code rather than specifications (Bourdieu) will, when challenged, perform enthusiasm for specification-driven development in the sprint planning meeting (Goffman), and then revert to coding the moment the back stage is restored. The performance is not a deliberate deception. It is a habitual response shaped by the field, as automatic in its way as the habitus itself. Goffman is Bourdieu at the microsociological level: impression management is what habitus looks like in face-to-face interaction.

Giddens borrowed from Goffman directly. His concept of regionalisation, the division of social life into front and back regions, was extended into structuration theory to show that the real work of reproducing social structure happens in the back stage of organisational life: where decisions are actually made, opinions actually formed, commitments actually tested. Change programmes that operate only on the front stage, the official communications, the governance frameworks, the mandated processes, miss the space where the organisation actually constitutes itself. Stacey’s shadow conversations are Goffman’s back stage seen through a complexity lens: the informal, ungoverned interactions where the real assessment of the transformation occurs.

2. Face-Work: Why People Will Sacrifice Honesty to Preserve Dignity

Goffman’s concept of face is subtler and more consequential than the front/back stage distinction. Face is the positive social value a person claims in an interaction. It is not vanity. It is the basic social currency that makes interaction possible. When someone “loses face,” the interaction itself is threatened.

Face-work is the set of practices by which people maintain their own face and, critically, help others maintain theirs. When a senior architect makes a statement about AI that reveals they have not understood the technology, the room does not correct them directly. Someone changes the subject, or reframes the statement, or asks a question that allows the architect to gracefully revise. This is not politeness. It is the structural requirement for continued interaction.

The implications for transformation are severe. If adopting AI means publicly acknowledging that one’s existing expertise is insufficient, then adoption requires losing face. And face-work, the collaborative effort to prevent face-loss, will actively obstruct the acknowledgment. The organisation will protect its members from public inadequacy, not because people are cowards, but because face-work is the oil that keeps the social machinery running.

Argyris diagnosed exactly this dynamic as a defensive routine: the pattern by which organisations avoid embarrassment at the cost of learning. Goffman provides the micro-mechanism. Defensive routines are face-work at scale. The undiscussable topic is undiscussable because raising it would cause someone with power to lose face. This is why the truth-telling probe in this series is so difficult even when psychological safety is nominally present. Edmondson’s psychological safety addresses whether people believe they can speak honestly without punishment. Goffman reveals a deeper obstacle: people manage impressions automatically, and the performance itself becomes the barrier to honest disclosure, regardless of whether the environment is formally safe. You can create the conditions where truth-telling is not punished. You cannot, through policy alone, override the face-work reflexes that make truth-telling feel like a social violation.

Every transformation programme must answer a question most never ask: where in your adoption plan is it safe for a senior person to be publicly incompetent? If the answer is “nowhere,” your programme will produce performance, not learning.

3. Stigma: What Transformation Does to Identity

Goffman’s Stigma (1963) addresses what happens when an attribute becomes discrediting. Stigma is not inherent in the attribute. It is produced by the relationship between the attribute and the social expectations of the setting.

AI transformation is a stigma-producing event. The practitioner who cannot write effective specifications, who struggles with the shift from implementation to articulation of intent, risks being marked as obsolete. And the marking is not merely professional. It is, in Goffman’s terms, a spoiled identity: an identity that can no longer sustain the performance the field requires.

Bourdieu’s capital devaluation is the structural equivalent. Goffman shows what capital devaluation looks and feels like in daily interaction: the moment when your contribution is politely acknowledged and then ignored, when the meeting you used to chair is now chaired by someone else, when the conversation pauses uncomfortably after you speak. These are the micro-interactions through which the field communicates that your capital has been devalued, and they operate with devastating precision because they are never explicit. Nobody says “you are no longer relevant.” The interaction order says it for them, through a thousand small signals that the person reads in their body before they process them in their mind.

Kegan’s developmental theory adds the structural dimension. A person at the socialised mind derives identity from the expectations of valued others. If the community begins to mark “non-AI-fluent” as discrediting, the socialised person experiences this as an identity crisis, not a skills gap. Heifetz would say: name the loss. The person whose face is being threatened needs to hear that what they built has value, that what is being lost deserves acknowledgment, and that there is a credible path from the old identity to a new one. Without that, the person will manage the stigma through impression management: performing adoption while privately preserving the old practice. Goffman calls this passing: adopting the appearance of the non-stigmatised identity without actually making the transition. It looks like adoption. It is theatre.

4. Frame Breaks: When the Performance Collapses

Goffman’s later work on Frame Analysis (1974) introduced frames: the interpretive structures through which people make sense of what is happening. A frame break occurs when the agreed definition of the situation collapses. A junior engineer points out that the AI-generated code is better than the principal engineer’s. A client asks why the team is still manually writing what a machine could produce in minutes. The frame that sustained the interaction, the shared understanding of what we are doing and who we are to each other, fractures.

Weick would recognise this as a cosmology episode: the moment when shared sense of what is happening dissolves. Bateson would see the frame break as a moment when logical types collide: the performance (”we are transforming”) and the reality (”we are performing transformation”) exist at different logical levels, and the frame break is the moment when the levels become visible simultaneously. For a brief, uncomfortable moment, everyone can see both: the performance and the reality behind it.

For leaders, the strategic question is not how to prevent frame breaks but how to use them. A frame break, handled well, can be the moment of genuine learning that Argyris’s double-loop model describes: the governing assumptions become visible because the performance that concealed them has collapsed. Handled badly, it produces retrenchment, embarrassment, and intensified impression management. The difference lies in whether the leader treats the break as a threat to be contained or as information to be explored. The frame break is the moment when the back stage truth has forced its way onto the front stage. What the leader does in the next sixty seconds determines whether the organisation learns from it or buries it deeper.

(An Organisational Prompt is something you can do now....)

Organisational Prompt

Pick one initiative in your transformation. Identify what is said about it in official settings (town halls, steering committees, status reports) and what is said in unofficial settings (corridor conversations, DMs, retrospectives with closed doors). Write both versions down. Put them side by side.

The gap is not noise. It is signal. It is the most accurate diagnostic you have, because the back stage is where people tell the truth. If the two versions are broadly consistent, your programme is working. If they are structurally opposed, you are not leading a transformation. You are hosting a theatrical production.

Now choose one element of the back stage truth and bring it, carefully, onto the front stage. Not as accusation. Not as confession. As information: “I have heard concerns about X. I want to understand them.” Goffman would tell you this is risky because it threatens the face of everyone performing compliance. Argyris would tell you it is essential because nothing changes while the real assessment remains undiscussable. Both are right. The skill is in creating conditions where truth-telling does not require face-loss: where saying “this is not working” is treated as professional judgement, not as disloyalty.

Further Reading

Erving Goffman, The Presentation of Self in Everyday Life (1959). The foundational text. Short, vivid, and startlingly applicable to organisational life.

Erving Goffman, Stigma: Notes on the Management of Spoiled Identity (1963). Essential for understanding what happens to professional identity when the field’s definition of competence shifts. The concepts of “passing” and “covering” apply directly to how people manage the transition from old expertise to new.

Erving Goffman, Interaction Ritual: Essays on Face-to-Face Behavior (1967). The essay “On Face-Work” is the single most useful piece for understanding why honest conversation about transformation is so difficult.

Erving Goffman, Frame Analysis: An Essay on the Organization of Experience (1974). Dense but rewarding. The concept of frame breaks is directly applicable to the moments when organisational pretence collapses and genuine learning becomes possible.

I write about the industry and its approach in general. None of the opinions or examples in my articles necessarily relate to present or past employers. I draw on conversations with many practitioners and all views are my own.

Richard Normann spent his career arguing that the traditional strategy model, the value chain, the competitive position, the sequential logic of production and delivery, is not merely incomplete. It is a mental prison. Not because it was ever wrong, but because the world it described has been replaced by one that operates on entirely different principles. And the greatest danger is not that your strategy is failing. It is that your mental model of what strategy means is preventing you from seeing what is actually happening around you.

Normann’s work integrates theory and practice in a way that places him squarely in the tradition this series has been building. His central argument is that the map, the conceptual framework through which leaders interpret their situation, is itself the strategic variable. Not the position on the map. Not the execution of the plan derived from the map. The map. And when the map no longer describes the landscape, every optimisation within it produces the wrong kind of excellence: brilliant execution of the wrong strategy, arrived at with great efficiency.

1. The Map and the Landscape

Normann’s central metaphor is the relationship between the map and the landscape. The landscape is the actual logic of value creation: the real configuration of actors, resources, relationships, and technologies. The map is the conceptual framework in the leader’s head: the mental model used to interpret the landscape, decide what matters, and determine what is possible.

The relationship is dialectical. Your map is shaped by the landscape you inhabit. But the landscape is also shaped by the maps the actors within it carry. When IKEA reimagined furniture retail, it did not optimise the existing value chain. It reframed the relationship between manufacturer, retailer, and customer, turning the customer into a co-producer. The new landscape could not have been seen from within the old map.

Bateson’s epistemology grounds this. The map is a description, and Bateson’s framework explains when descriptions become pathological: when the map is confused with the territory, when a description at one logical level is treated as though it operates at another. The organisation that treats its AI strategy document as though it were the reality of AI adoption, rather than a description of a reality that may or may not exist, is confusing logical types. The map feels like the territory because WYSIATI, as Kahneman describes, ensures that the coherent story the map tells suppresses awareness of what the map leaves out.

The proximity probe in this series follows directly. The further a leader is from the work, the more they rely on the map, and the less they notice when the map is wrong. The leader who sits with a team attempting to use AI on a real problem will have a less coherent map and a more accurate one. Proximity is the antidote to map-territory confusion, because proximity introduces the messy, contradictory details that prevent the map from feeling complete.

Argyris diagnosed the mechanism that prevents reframing. Defensive routines, the skilled incompetence by which professionals protect their existing theories-in-use, are the cognitive habits that keep the old map in place. Bourdieu deepens this: the map is not just in the head. It is in the habitus. The leader whose career was built on the value chain model does not merely think in value chain terms. Their professional dispositions generate value chain responses automatically. The map is embodied, and embodied maps do not yield to better arguments. They yield to sustained practice under new conditions.

2. From Value Chain to Value Constellation

In 1993, Normann and Rafael Ramírez directly challenged Porter’s value chain. Porter’s model assumes a linear sequence: raw materials flow in one end, finished products come out the other, each step adding value. Normann argued this describes a world that was already disappearing. In its place was the value constellation: a network of actors who co-create value through complex, reciprocal relationships rather than sequential handoffs.

The value chain assumes value is created by producers and consumed by customers. The value constellation assumes customers are co-producers. IKEA does not deliver furniture. It provides a configuration of design tools, flat-pack components, and assembly instructions that enables customers to create their own living environment. The value is not in the product. It is in the reconfigured relationship between IKEA’s capability and the customer’s participation.

For AI transformation, this is not a metaphor. It is a literal description. Consider specification-driven development: the domain expert specifies, AI generates, the team validates, the customer provides the knowledge that makes the specification meaningful. This is not a chain. It is a constellation of human and machine actors co-producing value through reciprocal interaction. The organisation that manages this as a sequential handoff (requirements → development → testing → acceptance) is imposing the old map on a landscape that has already reconfigured.

Drucker anticipated this: his insistence that the purpose of a business is to create a customer, and that marketing is “the whole business seen from the customer’s point of view,” prefigures the value constellation. But where Drucker still assumed the firm as the primary agent, Normann dissolves the boundary between firm and customer entirely.

3. Frozen Knowledge and Dematerialisation

Normann introduced a concept that anticipates the AI era with uncanny precision: offerings are frozen knowledge. A product is not a physical thing. It is knowledge about materials, processes, and customer needs encoded into a particular form. A software application is frozen knowledge about a domain, a workflow, and a set of user needs.

The strategic chain runs: dematerialisation (knowledge separated from its carrier) → unbundleability (components freed from their integration) → liquidity (components freely combinable) → rebundleability (new configurations) → increased density (more value per interaction).

This is precisely what AI does to knowledge work. The knowledge frozen into a software application is being dematerialised. A specification unbundles that knowledge from its code implementation. The specification becomes liquid, combinable with different AI models, different validation frameworks, different deployment targets. And the rebundled result is denser: more value created per unit of human attention.

Mintzberg’s craft metaphor connects. The specification returns to the craftsperson for validation. The loop closes. The potter’s hand still touches the clay, even if the wheel now turns faster. But Normann warned that dematerialisation creates a trap for organisations whose identity is bound to the physical form. When Giddens’s structures, the routines and roles that reproduce the organisation daily, are built around the frozen form rather than the underlying knowledge, dematerialisation threatens the entire identity structure. The development team whose identity is “we write code” experiences specification-driven development as existential threat. The code was never the value. The knowledge was. But identity was attached to the code. Bourdieu would say: the habitus was formed around the frozen form, and the habitus will defend the form even after the knowledge has been liquefied and rebundled elsewhere.

4. Cranes, Not Sky-hooks

Normann borrowed Daniel Dennett’s distinction between cranes (legitimate conceptual tools grounded in evidence) and sky-hooks (magical explanations that provide the feeling of leverage without the substance). The reframing Normann describes requires cranes. A new map is useful if it reveals patterns the old map concealed and generates actions that produce results the old map could not have predicted.

The AI transformation landscape is littered with sky-hooks. “AI will transform everything” is a sky-hook. “We need an AI strategy” is a sky-hook. “We will become an AI-first organisation” is a sky-hook. These provide the feeling of conceptual change without the substance. A crane would be: “The knowledge currently frozen in our codebase can be dematerialised into specifications, making it liquid enough to be recombined with AI generation capabilities for new customer experiences.” That is testable. It can be validated or falsified through action. It is a crane.

Dweck’s growth mindset is the individual precondition: reframing requires treating your conceptual framework as a tool that can be improved, not a fixed property of your intelligence. The fixed mindset clings to the existing map because the map is the competence. Admitting the map is wrong feels like admitting you are incompetent. Seligman’s learned helplessness warns that in organisations with a history of failed conceptual transformations, another reframe, however brilliant, will be received as just another sky-hook. The emotional precondition for reframing is not intelligence. It is trust: the belief, grounded in recent experience, that new thinking leads to new results.

5. Normann and His Limits

Normann must be read with his limitations visible. His work is extraordinarily cerebral: a strength for the quality of the conceptual framework, a weakness for its applicability to the messy, emotional, political reality of organisations. He understood that reframing requires grounding in authentic human experience, but his framework provides more guidance on how to think than on how to lead through the anxiety that thinking differently produces.

Stacey would argue that Normann overestimates the power of conceptual leadership. A brilliant reframe is still a gesture within a web of interaction, and its fate depends on the responses it provokes, not on its elegance. Argyris’s defensive routines do not yield to better concepts. They yield to the slow work of making the undiscussable discussable. Peters provides what Normann lacks: the emotional energy that makes people care enough to act on the diagnosis. The synthesis remains what Peters called disciplined liberation: Normann’s conceptual elegance combined with Stacey’s respect for emergence, Argyris’s diagnostic rigour, Peters’s energy, and Weick’s insistence that action must precede understanding. The reframe must be grounded, but the ground is found through movement, not through analysis.

(An Organisational Prompt is something you can do now....)

Organisational Prompt

Draw your current AI development value chain on a whiteboard. Put your organisation at the centre. Draw arrows showing the sequential flow: business requirements → development → AI tools → testing → deployment → customer. Now erase the arrows. Replace them with bidirectional lines. Add the actors you left out: the domain experts whose tacit knowledge shapes the specification, the customers whose feedback reshapes the offering, the AI systems whose capabilities constrain and expand what is possible.

Ask: “What does this constellation reveal that the chain concealed?” The gap between the chain you drew first and the constellation you drew second is the gap between your current map and the landscape you actually inhabit.

Further Reading

Richard Normann, Reframing Business: When the Map Changes the Landscape (2001). His final and most ambitious work. The map-landscape dialectic, the dematerialisation chain, and the concept of Prime Movership. The foreword by Mintzberg is itself worth the price.

Richard Normann and Rafael Ramírez, Designing Interactive Strategy: From Value Chain to Value Constellation (1994). The full development of the value constellation concept.

Richard Normann, Service Management: Strategy and Leadership in Service Business (3rd edition, 2000). The foundational insight that services are co-produced and the customer is always a participant in value creation, even more relevant when the “service” is an AI system that depends on human specification.

I write about the industry and its approach in general. None of the opinions or examples in my articles necessarily relate to present or past employers. I draw on conversations with many practitioners and all views are my own.

In 1925, a woman who had never run a company gave a lecture to a group of businessmen in New York on how to handle disagreement. She told them that every conflict has three possible outcomes: one side wins, both sides lose a little, or both sides get what they actually need. She told them that the first is unstable, the second is mediocre, and the third is the only one worth pursuing. She told them that the third option requires something most managers will not do: stop arguing about positions and start asking what each party genuinely wants.

The businessmen were polite. Within ten years of her death in 1933, Mary Parker Follett was almost entirely forgotten. It took six decades for Peter Drucker to call her his guru. She was rediscovered not because the world caught up with her ideas but because the problems she diagnosed became impossible to ignore. Those problems are now on your desk. They arrive whenever business leaders and technology leaders sit in a room and try to agree on what AI should do. They arrive whenever a transformation produces winners and losers instead of shared capability. They arrive whenever someone with positional authority overrides someone with situational knowledge. Follett addressed all of this before cybernetics, before systems thinking, before complexity theory. She is the positive vision for what this series has been diagnosing as failure: she describes what it looks like when the interaction works.

1. Integration: The Resolution That Produces Learning

Follett identified three responses to conflict. Domination: one side wins. Compromise: both sides sacrifice something. Integration: a creative reframing that satisfies all legitimate interests without requiring anyone to surrender what they actually need.

Her example was a library window. Two people: one wants it open, the other wants it closed. Compromise is to open it halfway, which satisfies neither. Domination is whichever person is louder or more senior. Integration requires asking why: one wants fresh air; the other does not want wind on their papers. The solution is to open a window in the next room. Both get what they need. Neither compromises. The solution was invisible until the positions were replaced by interests.

This is not a negotiation technique. It is an epistemological claim about how organisations should process difference, and it is the observable test for the conflict integration probe that runs through this series: can the organisation hold competing perspectives in tension long enough for a solution neither party had imagined to emerge? Or must it resolve conflict through hierarchy (domination) or negotiated surrender (compromise)?

Argyris would recognise integration immediately: it is Model II behaviour applied to conflict. Model I (unilateral control, win, suppress negative feelings) produces domination. The institutional default produces compromise. Integration requires the willingness to make reasoning transparent, to test assumptions publicly, and to treat the other party’s needs as data rather than obstacles. Bateson’s levels framework shows why integration is so rare: it operates at Learning II. Domination and compromise work within the existing frame (who has more power? what is each willing to give up?). Integration requires questioning the frame: what do we each actually need, and is there a configuration of the situation that serves both? The frame-questioning is what makes integration generative and what makes it so difficult.

Apply this to AI transformation. Technology teams dominate by choosing tools without meaningful input from business domains. Business teams dominate by mandating outcomes without understanding constraints. The compromise is a governance committee that satisfies nobody. Integration would require both sides jointly studying the situation, surfacing what each genuinely needs, and constructing a way of working that serves both without requiring either to sacrifice what matters most. The reason this almost never happens is structural: Edmondson’s psychological safety is the precondition (people will not surface genuine interests where honesty is punished), and Stacey’s complexity thinking shows it cannot be designed in advance (it emerges from genuine interaction between people who hold different knowledge).

2. Power-With: How Parts Should Relate

Follett distinguished between power-over and power-with. Power-over is coercive: one party exercises authority over another. It produces compliance, not commitment. It generates resistance, hidden or overt. Power-with is collaborative: jointly developed power that grows through interaction. Follett’s formulation was precise: the question is not how to get control of people but how all together to get control of a situation.

The distinction is structural, not sentimental. Power-over produces a fixed sum: if I have more authority, you have less. Power-with produces a growing sum: when people develop power together, the total capacity increases. A problem that requires knowledge from multiple domains cannot be solved by any single authority. It can only be solved by people who hold different knowledge working together in a way that increases their collective capability.

Weber’s rational-legal authority is power-over institutionalised. The bureaucratic hierarchy gives decisions to role-holders regardless of whether the role-holder has the knowledge the decision requires. Follett does not reject hierarchy; she insists that authority within hierarchy should flow to knowledge, not to position. Heifetz makes the same argument: the leader’s job is not to provide solutions but to create the conditions in which people with relevant knowledge can work the problem. Heifetz is describing power-with applied to adaptive challenges. Follett got there first.

For AI transformation, the diagnostic is direct. Who decides how AI is used in any given domain? If a central technology team decides about domains they do not understand: power-over. If domain experts decide about technology they do not understand: power-over inverted. If domain experts and technology practitioners jointly study the situation, each contributing what they know, with authority flowing to whoever has the most relevant knowledge at each decision point: power-with. The third option is harder, slower, and more uncertain. It is also the only one that produces outcomes fitted to the actual situation rather than to the organisational chart. Bourdieu would add that power-with requires a field in which different forms of capital are recognised as legitimate: the domain expert’s practical consciousness and the technologist’s technical knowledge must both count as currency. In a field where only technical capital is valued, power-with collapses back into power-over exercised by the technically dominant.

3. The Law of the Situation

Follett’s most radical idea was also her most practical. She called it the law of the situation: decisions should be governed by the specific circumstances of the work, not by the arbitrary authority of a manager. Her instruction was blunt: depersonalise the giving of orders; unite all concerned in a study of the situation; discover the law of the situation and obey that.

When orders are part of the situation, the question of someone giving and someone receiving does not arise. Both manager and worker accept the orders given by the situation itself. Authority becomes situational, not positional.

This anticipated Drucker by three decades: the knowledge worker must define the task, and the definition must come from the work itself. It anticipated Snowden by seven decades: in a complex domain, the appropriate response is determined by the situation, not by a pre-existing plan. It anticipated the entire argument of this series: AI transformation fails when decisions are made by people distant from the work, and succeeds when they are made by people who understand the situation well enough to discover what it requires. Weick’s deference to expertise in high-reliability organisations is the law of the situation operationalised: when a crisis occurs, authority migrates to whoever has the most relevant knowledge, regardless of rank.

Follett’s four principles of coordination remain strikingly contemporary: coordination by direct contact between the responsible people (not through intermediaries), in the early stages (not after decisions are made), as a reciprocal relationship (not one-way instruction), and as a continuous process (not a periodic review). Every one of these is routinely violated in how large organisations manage transformation. Decisions are made centrally and communicated downward. Feedback flows upward through filtered channels. Review happens quarterly rather than continuously. Follett would diagnose this as an organisation obeying the hierarchy when it should be obeying the situation.

4. Circular Response: The Systems Thinker Before Systems Thinking

Follett rejected linear cause-and-effect models. Every interaction, she argued, is circular: each person’s behaviour influences and is influenced by the other’s. The relationship between manager and worker is not one-way transmission but continuous mutual adjustment.

This insight, published in 1924, anticipated cybernetics by two decades and complexity theory by six. Stafford Beer’s feedback loops formalise what Follett described in plain language. Bateson’s schismogenesis, escalating division through unchecked circular patterns, is the pathological case Follett identified when circular response goes wrong: without corrective feedback, circular processes amplify rather than stabilise. Stacey’s complex responsive processes are an updated framework for what Follett called the group idea: the insight that genuine interaction produces understanding beyond what any individual brings, not by summing contributions but through the emergent properties of the interaction itself.

The group idea connects integration to learning. Every barrier this series has identified, defensive routines, fixed mindset, learned helplessness, pathological information flow, bureaucratic rigidity, is a barrier to integration. And every condition the series has identified as necessary for learning, psychological safety, high standards, growth mindset, honest feedback, situational authority, is a condition for integration. Integration is what it looks like when the Interaction lever is working: when the parts of the organisation relate to each other in a way that produces capability neither part could generate alone. Domination and compromise are what it looks like when the lever is broken: when the parts relate through power or through mutual surrender rather than through genuine engagement with each other’s knowledge.

Follett saw both the barriers and the conditions a century ago. Her misfortune was to be right too early. The rest of us have no such excuse.

(An Organisational Prompt is something you can do now....)

Organisational Prompt

Pick one active conflict between business and technology in your AI transformation: a disagreement about tooling, scope, governance, or ownership. Before the next meeting, ask each side separately: “What do you actually need from this; not what you have asked for, but what would genuinely serve your work?” Write both answers down. Bring them to the meeting and put them side by side.

The gap between what people are arguing for and what they actually need is where integration lives. You will not find it by asking people to compromise. You will find it by asking people to be honest about what matters, and by taking both sets of needs seriously enough to look for a solution that serves them all. If neither side can articulate what they actually need, that itself is diagnostic: the positions have hardened into identities, and the conflict has become undiscussable in exactly the way Argyris described.

Further Reading

Mary Parker Follett, Creative Experience (1924). Her most sustained theoretical argument. The chapters on integration and circular response remain startlingly modern. Freely available on the Internet Archive.

Henry Metcalf and Lyndall Urwick (eds.), Dynamic Administration: The Collected Papers of Mary Parker Follett (1941). The most accessible collection of her management lectures, including “The Giving of Orders” and “Constructive Conflict.”

Pauline Graham (ed.), Mary Parker Follett: Prophet of Management (1995). The book that rediscovered Follett. Includes original papers alongside commentary by Drucker, Mintzberg, and others.

Joan C. Tonn, Mary P. Follett: Creating Democracy, Transforming Management (2003). The definitive biography. Essential for the intellectual and political context.

I write about the industry and its approach in general. None of the opinions or examples in my articles necessarily relate to present or past employers. I draw on conversations with many practitioners and all views are my own.

There is a script that every large enterprise follows when it adopts a new practice. A disruption appears. Leadership declares a strategic response. Goals are set, resources are allocated, teams are restructured, training programmes are launched. The organisation absorbs the new technology, adjusts its processes, updates its language, and settles into a new steady state that looks different on the surface but operates according to exactly the same underlying logic as before. The transformation programme is declared a success. Nothing fundamental has changed.

This pattern is so universal that it cannot be explained by individual failure. It is structural. And it was codified by the most influential sociologist of the twentieth century: Talcott Parsons. When your AI adoption programme secures resources, sets strategic goals, coordinates teams, and retrains staff, it is executing a Parsonian script. And the script is designed, at every level, to restore equilibrium rather than to produce change. Parsons’ framework was criticised as conservative, teleological, and incapable of accounting for power, conflict, or agency. This article focuses not on whether it is right but on why it matters: it is the invisible operating system of enterprise change management, and understanding it is the first step toward interrupting it.

1. The AGIL Scheme: The Template You Are Already Following

Parsons’ most influential contribution was the AGIL scheme: four functional imperatives that every social system must fulfil.

• Adaptation: secure resources from the environment.

• Goal Attainment: define objectives and direct energy toward them.

• Integration: coordinate internal parts and manage conflict.

• Latency (Pattern Maintenance): maintain and transmit the values and cultural patterns that hold the system together.

Compare this to the last transformation programme you witnessed. Secure budget and hire talent (Adaptation). Set the strategy and declare targets (Goal Attainment). Restructure teams and establish coordination (Integration). Launch training and update role descriptions (Latency). The correspondence is not coincidental. AGIL is the deep grammar of enterprise change management: the template that programme managers follow without knowing they are following it.

The problem is not that AGIL is wrong. At the descriptive level, it is accurate: organisations genuinely need to perform all four functions. The problem is that AGIL is a theory of system maintenance, not system transformation. It describes how organisations absorb disruption and return to stability. It has no vocabulary for the possibility that the system itself might need to become something fundamentally different.

Stacey would identify the core error: AGIL assumes the system knows what functions it needs to perform. But in genuine transformation, the functions themselves are uncertain. AI does not slot into the existing architecture. When AI can generate code from specifications, the Adaptation function is no longer about securing developer talent in the traditional sense. Goal Attainment cannot set targets for a capability whose possibilities are still being discovered. Integration must coordinate forms of work that do not yet have stable definitions. And Latency is being asked to transmit a culture that does not yet exist.

2. Socialisation and Social Control: The Reproduction Engine

For Parsons, system stability depends on two mechanisms. Socialisation transfers norms, values, and expectations to members. Social control ensures that people who deviate are brought back into alignment. In an organisation: onboarding, mentoring, code reviews, architecture reviews, performance evaluations, promotion criteria, the stories told about heroes and failures. These are the mechanisms by which the organisation produces people who reproduce the organisation.

Bourdieu called this habitus. Giddens called it practical consciousness. Parsons called it the product of socialisation. The developer socialised for ten years into a culture that values code elegance and individual problem-solving does not carry knowledge. They carry an identity, a set of reflexes, a way of seeing that makes certain actions natural and others unthinkable. The habitus was formed by the socialisation, and the habitus reproduces the field that formed it.

Social control operates alongside: the peer pressure that discourages deviation, the performance review that rewards conformity, the governance framework that channels innovation into approved patterns. And here is where the series delivers its sharpest critique of Parsons.

Argyris demonstrates that what Parsons calls social control is the same mechanism that produces defensive routines: the patterns that prevent people from surfacing the information the system most needs. The performance review that enforces existing standards is the same review that punishes experimentation. The architecture board that maintains consistency is the same board that blocks novel approaches. The governance framework that ensures predictability is the same framework that prevents the exploratory probes that Weick and Snowden identify as essential for learning.

Westrum’s typology makes the distinction Parsons cannot. All three culture types, pathological, bureaucratic, and generative, perform Parsons’ Integration function. But only the generative culture performs it in a way that enables learning. Parsons asks only whether the function is being performed. He cannot ask whether it is being performed in a way that serves the system’s capacity to change. Bateson would say Parsons is locked at Learning I: he can describe the correction of errors within the existing frame but has no vocabulary for questioning the frame itself.

3. The Equilibrium Assumption: Why “Adaptation” Is Not Change

Parsons’ deepest influence is his equilibrium assumption: social systems naturally tend toward balance, and disruptions are resolved through mechanisms that restore stability. This assumption is embedded in the language of change management: “managing resistance,” “navigating the transition,” “landing the change.” All assume transformation is a journey from one stable state to another.

The Parsonian response to AI follows the pattern precisely. Fit AI into existing roles (differentiation: create an “AI team”). Make existing processes AI-enhanced (adaptive upgrading: add AI review to the code pipeline). Include AI tools in existing workflows (inclusion: give everyone a licence). Generalise values (update the mission statement to say “AI-enabled”). The system adapts. Equilibrium is restored. Nothing fundamental has changed.

Stacey rejects the equilibrium assumption directly. Organisations operate at the edge of chaos. Genuine novelty emerges from instability, not from equilibrium-seeking. The Parsonian script actively prevents emergence by treating every disruption as a problem to be resolved rather than a possibility to be explored. Heifetz’s distinction cuts to the core: Parsons treats all change as technical (the system identifies a requirement and develops mechanisms to meet it). AI transformation is adaptive: it asks people to change themselves, their values, and their ways of working. Parsons’ framework has no vocabulary for this.

Weber, whom Parsons translated into English and whose critical pessimism Parsons systematically softened, saw the endpoint clearly. The iron cage is Parsons’ equilibrium achieved: a system so thoroughly rationalised that the people inside it cannot imagine an alternative. Parsons’ value-generalisation, where values become more abstract to accommodate diversity, produces the meaningless abstractions the series has consistently identified: “We value innovation,” “We are a learning organisation,” “We embrace AI.” These accommodate everything and constrain nothing.

4. Each Function Fails at Precisely the Point Where Learning Is Needed

Understanding Parsons means understanding why your programme produces adaptation rather than transformation. Each AGIL function fails at a specific point, and the series provides the diagnostic for each.

The Adaptation function fails when the organisation secures resources for the wrong landscape. Normann’s map-landscape dialectic is the diagnostic: investing in AI talent to accelerate code production is perfect adaptation to a landscape that is ceasing to exist. The real landscape requires specification capability and domain articulation.

Goal Attainment fails when goals precede action. Weick’s sensemaking provides the alternative: in uncertain environments, goals emerge from action, not before it. The Parsonian sequence, set the goal then allocate resources, reverses the order learning requires. Peters’ bias for action is the correction: act, learn, then rationalise the goal retrospectively.

Integration fails when it suppresses information. Westrum’s typology is the diagnostic: if your integration mechanisms channel information through hierarchy or punish its delivery, the more effectively you integrate, the less you learn. Argyris’s Model II, the capacity to surface and test assumptions publicly, is the integration that Parsons cannot theorise.

Latency fails when it reproduces the culture that needs to change. Bourdieu’s habitus explains why: the training programme updates what people can say (discursive consciousness) without touching what their hands do (practical consciousness). The more effective your socialisation, the more reliably you reproduce the culture you already have. This is the Parsonian trap at its deepest: the mechanism designed to maintain the system is the mechanism that prevents the system from becoming something new.

Dweck’s mindset research reveals the individual dimension. Parsons’ Latency function, through performance reviews, promotion criteria, and recognition practices, reproduces fixed mindset in every cycle by treating ability as static and measurable. Bateson would diagnose the entire AGIL logic as a Learning I system: it can correct errors within its existing frame, optimising each function for efficiency, but it cannot question whether the functions themselves need to change. That questioning is Learning II, and nothing in Parsons’ framework supports it.

5. The Script You Must Interrupt

Parsons is valuable not because he is right but because he is diagnostic. When you recognise AGIL logic operating in your transformation programme, you have identified the mechanism that will convert your transformation into adaptation. The intervention is not to abandon the functions; the organisation does need resources, direction, coordination, and culture. The intervention is to interrupt the equilibrium-seeking at each stage.

At Adaptation: secure resources for the landscape that is emerging, not the one that is familiar. At Goal Attainment: create conditions for experimentation before setting targets, because the targets worth setting can only be discovered through action. At Integration: build generative information flow, not bureaucratic coordination, so that the weak signals of what is actually working can reach the people who need them. At Latency: design socialisation that forms new dispositions through practice, not training that reproduces old ones through instruction.

The thinkers in this series provide what Parsons could not: a theory of organisational learning, not merely organisational maintenance. Action before understanding contradicts rational goal attainment. Surfacing the undiscussable contradicts social control. Addressing identity contradicts role-based socialisation. Ungovernable learning contradicts the latency function. The question is: at which point in the AGIL sequence will you interrupt the script and do something different?

(An Organisational Prompt is something you can do now....)

Organisational Prompt

Map your current transformation programme onto the AGIL framework. Be specific. What have you done to secure resources (Adaptation)? What goals have you set (Goal Attainment)? How are you coordinating teams (Integration)? What training and socialisation have you launched (Latency)?

If your honest mapping reveals AGIL logic at every level, you have identified the structural reason your transformation will produce adaptation rather than change. Now ask at each stage: are we performing this function in a way that maintains the current system, or in a way that enables something genuinely new? The difference between those two questions is the difference between Parsons and everything this series has been arguing for.

Further Reading

Talcott Parsons, The Social System (1951). The core statement of structural functionalism and the AGIL scheme. Dense and abstract. Read it as a diagnostic: it describes, with extraordinary precision, the logic that enterprise change management follows by default.

C. Wright Mills, The Sociological Imagination (1959). The most influential critique of Parsons. Mills argued that grand theory disconnected from real problems serves to legitimate the status quo. The counter-argument that social theory should reveal how systems might be changed, not merely describe how they maintain themselves.

Anthony Giddens, The Constitution of Society (1984). Developed explicitly to overcome Parsonian functionalism. Where Parsons treats structure as an external framework, Giddens shows it exists only in its instantiation in practice.

I write about the industry and its approach in general. None of the opinions or examples in my articles necessarily relate to present or past employers. I draw on conversations with many practitioners and all views are my own.

There is a particular kind of failure that haunts every transformation programme. It is not the failure of people who refuse to change. That failure is visible and manageable. The failure that ruins programmes is the one where everyone agrees, everyone commits, everyone is visibly willing, and nothing happens. The strategy is endorsed. The tools are deployed. The training is attended. Three months later, the organisation is doing the same things with slightly updated vocabulary.

Robert Kegan, a developmental psychologist who spent forty years at the Harvard Graduate School of Education, diagnosed why this happens. His answer is structural, uncomfortable, and directly relevant to anyone watching a transformation stall: people do not resist change. They are immune to it. And the immunity operates through a mechanism invisible to the person it controls. Kegan provides what Argyris could not: the developmental explanation for why intelligent people who agree that their governing assumptions must change consistently fail to change them. The governing assumptions are not merely undiscussed. For many people, they are undiscussable, because they operate at a level of the self that the person’s current developmental capacity cannot reflect upon.

1. Two Kinds of Learning

Kegan draws a distinction that cuts through most of what organisations mean by “learning.” Informational learning adds new content to an existing mindset. Transformational learning changes the form of the mindset itself.

Informational learning is what happens when a team attends a workshop on prompt engineering. The content is new; the container is unchanged. The person knows more but thinks the same way. Transformational learning is what happens when a senior architect begins to understand that AI changes what design means, and instead of defending the old definition or adopting the new one uncritically, develops the capacity to hold both, evaluate both, and construct a new understanding of their work. The person does not just know more. They can see more. What was invisible becomes visible.

Bateson mapped the same territory with greater precision. Learning I adjusts behaviour within a given frame. Learning II changes the frame. Kegan’s contribution is to explain the developmental mechanism that determines which kind of learning a person is capable of at any given moment. This mechanism is the binding constraint on every programme that assumes willingness is enough. If AI transformation requires transformational learning and the organisation invests exclusively in informational learning, it is filling a container that needs to be rebuilt.

Bourdieu’s habitus is the embodied expression of this distinction. Informational learning changes what people can say (discursive consciousness). Transformational learning changes the dispositions from which practice is generated. The movement from socialised mind to self-authoring mind, in Kegan’s framework, is a transformation of habitus: not just new knowledge but new generative principles that produce different practice automatically. The person does not merely learn to write specifications. They become someone whose habitus generates specification-writing as naturally as it once generated code.

2. The Immunity Map: Why Willing People Do Not Change

Kegan and Lisa Laskow Lahey’s most practical contribution is the immunity map: a four-column diagnostic that reveals, with uncomfortable clarity, the mechanism of stalling.

• Column 1 is the genuine commitment: “I am committed to writing specifications rather than writing code.” The commitment is real.

• Column 2 is what the person actually does instead: rewrites AI-generated output from scratch, finds reasons the specification approach does not work for their domain, quietly continues the old work alongside the new.

• Column 3 is discovered by imagining doing the opposite of Column 2 and noticing what feels uncomfortable. This reveals hidden competing commitments: “I am also committed to not being seen as someone who cannot do the work themselves.” “I am committed to not losing my standing as the technical expert.” These are not in conflict with Column 1 by design. They are in conflict by structure. One foot on the accelerator, one on the brake.

• Column 4 surfaces the big assumptions that make the hidden commitments feel necessary: “I assume that if AI can do what I do, I am no longer valuable.” “I assume my worth is defined by my ability to produce, not my ability to specify.” These assumptions do not feel like beliefs. They feel like facts. And that is precisely why they are so powerful: they are not examined because they are not experienced as things that could be examined.

Bateson would call the Column 4 assumptions a category error: the person is looking at the wrong logical level. They are trying to change their behaviour (Learning I) while the assumptions that generate the behaviour (Learning II) remain Subject, things they look through rather than at. The immunity map is a tool for making Learning II visible: it surfaces the governing assumptions that Argyris said must change but could not explain why they resist changing. The hidden competing commitments in Column 3 are information present in the system but inaccessible to the person. They are the undiscussable beneath the undiscussable.

Heifetz’s adaptive leadership connects directly. The Column 4 assumptions are the losses that Heifetz says must be named: “My worth is defined by my ability to produce” is a loss statement waiting to happen. Until the loss is named, the immunity holds, because the person is protecting something they have not yet acknowledged they are protecting.

3. Developmental Stages: Why the Organisation Asks for What Most People Cannot Yet Give

Kegan describes three plateaus of adult mental complexity relevant to professional life.

• The socialised mind is shaped by the expectations of the environment: peers, bosses, professional community. The person is a reliable team player who communicates what they believe others want to hear. When the boss says AI is important, they agree. When peers say it is a fad, they agree with that too. They are not lying in either case. They are being shaped by the social field they inhabit. Roughly 58% of adults make meaning at this stage.

• The self-authoring mind has an internal compass. It can step back from social pressure and evaluate competing claims against self-authored criteria: evaluate AI outputs against professional judgement, specify what needs to be done without waiting for instructions, hold a position that contradicts the consensus when the evidence supports it. Roughly 35% of adults operate here. AI transformation demands this capacity at minimum.

• The self-transforming mind can hold its own framework as an object of scrutiny, see its limits, and integrate contradictions rather than resolving them. This maps to what Stacey’s skilled participation demands: leaders who tolerate paradox without collapsing it into false certainty.

The organisational learning problem is a developmental mismatch. The organisation asks for independent judgement and gets social conformity. It asks for honest evaluation and gets consensus. It asks for transformation and gets performance. Not because people are unwilling but because the developmental capacity the task requires has not yet formed.

Bourdieu deepens this. The socialised mind is a habitus perfectly adapted to the field that formed it: the dispositions were generated by the social environment and generate practice that reproduces it. Movement to self-authoring is a transformation of habitus, not merely a cognitive upgrade. It is the formation of new generative principles that can produce practice the old field did not reward. This is why the transition is so difficult and so slow: it requires sustained practice under conditions that support the new dispositions, not a training programme or a motivational speech.

Edmondson’s psychological safety is necessary but not sufficient. Safety creates the conditions in which developmental growth becomes possible. But safety alone does not produce growth. A socialised-mind professional in a safe environment will feel comfortable. They will not thereby become self-authoring.

4. The Deliberately Developmental Organisation

Kegan’s most ambitious idea is the Deliberately Developmental Organisation: a workplace that weaves development into the fabric of daily work rather than treating it as a separate activity. In most organisations, people are doing a second job nobody is paying them for: hiding weaknesses, managing impressions, covering mistakes. The DDO makes that second job unnecessary.

The DDO has three foundations. Edge: every person identifies their growing edge, the specific limitation they are working to overcome, and works on it publicly. Home: genuine safety for the vulnerability growth requires, combined with high standards that prevent safety from becoming comfort. Groove: disciplined daily practices and feedback routines that embed development into normal operations.

Senge described the learning organisation but could not fully specify what it would look like in practice. Kegan provides the specification. Westrum’s generative culture describes the conditions under which a DDO can function; pathological and bureaucratic cultures prevent the developmental exposure growth requires. Giddens’s three dimensions must all move: the DDO changes signification (what learning means), domination (who has authority over development), and legitimation (growth, not performance, is what gets recognised).

Most organisations will not become DDOs. But the immunity map alone, applied honestly, can reveal why a transformation is stuck in ways no amount of strategy review will address. The person who runs Column 4 to its conclusion and finds an assumption they did not know they held has already begun the journey from Subject to Object. That is transformational learning. It is the only kind that changes anything.

(An Organisational Prompt is something you can do now....)

Organisational Prompt

Run an immunity map on your own AI commitment. Take fifteen minutes. Write four columns. Column 1: a specific AI adoption commitment you hold and have not fulfilled. Column 2: what you actually do instead. Column 3: imagine doing the opposite of Column 2; what feels uncomfortable? Write those hidden commitments. Column 4: what assumptions make those competing commitments feel necessary?

Do not try to fix anything. Just look at what the map shows you. The assumptions in Column 4 are almost certainly things you have never articulated. They will explain something that strategy slides cannot. If you find an assumption that surprises you, you have found the thing that is governing your behaviour while remaining invisible to you. That is Subject becoming Object. That is the beginning of the only kind of learning that produces change.

Further Reading

Robert Kegan and Lisa Laskow Lahey, Immunity to Change (2009). The full immunity map diagnostic and the three developmental plateaus. The most directly useful of Kegan’s books for practitioners.

Robert Kegan and Lisa Laskow Lahey, The Real Reason People Won’t Change (Harvard Business Review, 2001). The original article introducing competing commitments and the immunity framework. Short, clear, and a better starting point than the book.

Robert Kegan and Lisa Laskow Lahey, An Everyone Culture: Becoming a Deliberately Developmental Organization (2016). The DDO concept with detailed case studies. Ambitious and demanding.

Jennifer Garvey Berger, Changing on the Job: Developing Leaders for a Complex World (2012). The most accessible practitioner-oriented account of Kegan’s developmental stages applied to leadership.

I write about the industry and its approach in general. None of the opinions or examples in my articles necessarily relate to present or past employers. I draw on conversations with many practitioners and all views are my own.

Pierre Bourdieu, the French sociologist whose work on practice, power, and cultural reproduction shaped virtually every social science discipline since the 1970s, explains why the obstacle to transformation is not in people’s reasoning. It is in their bodies. Decades of professional experience have inscribed a set of dispositions, reflexes, judgements, and instincts so deeply into your practitioners that these dispositions operate below conscious awareness. Your developers do not choose to write code when they could write specifications. Their hands reach for the keyboard and produce code because that is what their accumulated professional formation has trained them to do, automatically, without deliberation.

Bourdieu called this embodied structure habitus. And until you understand it, every transformation programme you launch will falter in exactly the same way, for exactly the same reason. The habitus explains why people resist loss, why practice transforms but instruction does not, and why beliefs about capacity are structurally reproduced rather than individually chosen. It is the mechanism beneath every identity barrier this series has diagnosed, and it governs the Identity dimension of the entire Learning phase.

1. Habitus: The Generative Principle

Habitus is the durable, transposable system of dispositions acquired through experience that generates practice without conscious deliberation. This is not “habit” in the everyday sense, a routine you could change if you chose to. Habitus is a generative principle. It does not merely reproduce behaviour; it produces new behaviour that is nonetheless consistent with the dispositions already formed. A developer encountering a new problem does not consciously select an approach. Their habitus generates a response: the tools they reach for, the abstractions they favour, the risks they notice and the ones they overlook, the colleagues they consult and the ones they do not. All of this happens before conscious deliberation begins.

Giddens’ practical consciousness is the structural counterpart. Where Giddens describes the tacit knowledge that reproduces social structure through interaction, Bourdieu describes how that knowledge becomes inscribed in the person. Together they explain both sides: structure shapes habitus, habitus generates practice, practice reproduces structure. The cycle is continuous, pre-reflective, and extraordinarily resistant to intervention.

Argyris’s theories-in-use live in the habitus. The gap between espoused theory (”we are an AI-first organisation”) and theory-in-use (”we write code”) is not a cognitive failure. It is the gap between what people can articulate and what their habitus generates. You cannot reflect your way out of habitus any more than you can reflect your way out of knowing how to ride a bicycle. This is why training programmes fail to produce transformation. A two-day workshop updates what people can say about their practice (discursive consciousness). It does not touch what their bodies do when they sit down to work. Habitus changes only through sustained exposure to new conditions: new practices, new interactions, new consequences, repeated over time until the new dispositions become as automatic as the old ones.

Bateson’s learning levels make the mechanism precise. Habitus change is Learning II: it is not a correction within the existing frame but a transformation of the frame itself. Learning I interventions (training, instruction, new tools) operate within the existing habitus. Learning II interventions (sustained new practice that gradually forms new dispositions) transform the habitus. Most organisations provide Learning I and wonder why the habitus does not change.

2. Field and Capital: What People Are Actually Protecting

Habitus does not operate in isolation. It operates within a field: a structured social space in which people compete for specific forms of advantage. Every field has its own logic, hierarchy, and rules about whose opinion carries weight. An enterprise technology organisation is a field. The positions, architect, developer, product manager, are not merely job titles. They are locations in a structured space defined by the forms of capital each commands.

Bourdieu extends capital far beyond the economic. Cultural capital is the knowledge, skills, and dispositions that confer authority: the embodied kind (knowing how to write elegant code, how to read a room in a technical debate), the objectified kind (the tools you use, the contributions you are known for), and the institutionalised kind (your degree, your certifications). Social capital is the network of relationships that provides access and influence. Symbolic capital is any form of capital recognised as legitimate: prestige, reputation, being “the expert.” Symbolic capital is the most powerful because it disguises its own arbitrariness. The architect’s authority appears to derive from knowledge (cultural capital) but is actually sustained by the field’s recognition of that knowledge as the kind that matters (symbolic capital).

AI transformation changes the exchange rate between these forms of capital. The developer whose cultural capital consists of code-writing expertise faces devaluation if AI can generate code from specifications. The domain expert whose knowledge was previously invisible may find their cultural capital suddenly revalued. Resistance to transformation is, in most cases, resistance to capital devaluation. People protect the forms of capital they have spent entire careers accumulating. This is not stubbornness. It is entirely rational behaviour within the logic of the field. And it will not be overcome by better communication, because the threat is real. The question is not how to persuade people that the threat is illusory. It is how to help them convert their existing capital into forms that the new field will value.

This is why Heifetz’s insistence on naming the loss is structurally necessary, not merely emotionally kind. The loss is real: capital that took decades to accumulate is being devalued. Naming it honours what was built. Denying it produces the cynicism that Seligman describes, because people can see what the organisation refuses to say. Dweck’s fixed mindset is a form of what Bourdieu calls doxa: the taken-for-granted assumption that ability is innate, which naturalises existing hierarchies of cultural capital. “Some people are just more technical” is a doxic claim that protects the symbolic capital of those already recognised as technical. Growth mindset, in Bourdieu’s terms, is the disruption of one form of doxa. But individual mindset change without field-level change will fail: if the performance review, the promotion criteria, and the recognition practices still reward the old forms of capital, the field-level doxa will overwhelm the individual-level shift.

3. Symbolic Violence: The Invisible Enforcement

Bourdieu’s most confronting concept is symbolic violence: the process by which the dominant perspective is internalised by everyone in the field, including those it disadvantages, so that the current order is perceived as legitimate by the people it subordinates. The junior developer who “knows” their opinion is less valuable than the architect’s has not been told this explicitly. They have absorbed it through the field’s logic: who speaks in meetings, whose objections halt a project. The field’s hierarchy has been internalised as a personal judgement about one’s own competence.

Consider the transformation programme that tells people they are “empowered” to experiment while the actual field structure, decision rights, resource allocation, performance criteria, remains unchanged. The result is not empowerment. It is symbolic violence: people who fail to transform are now responsible for their own failure, because the organisation has declared the obstacles removed. The structural constraints have been rendered invisible by the language of empowerment. Seligman’s learned helplessness is what happens when symbolic violence succeeds completely: people stop trying because they have internalised the field’s judgement that their inability to change is a personal deficiency rather than a structural condition.

Peters understood this intuitively. His rage against bureaucracy is a refusal to accept symbolic violence. Where Peters provides emotional energy for the refusal, Bourdieu provides the analytical precision to understand exactly what is being refused. The structures of domination are not imposed from outside. They are inscribed in the habitus of the people who suffer them. You are asking people to rebel against dispositions they carry in their own bodies.

4. Hysteresis: The Lag That Kills Transformation

Perhaps Bourdieu’s most practically useful concept is hysteresis: the mismatch that occurs when a field changes rapidly while the habitus of its participants remains adjusted to the old conditions. The old dispositions generate practices suited to a field that no longer exists.

Hysteresis produces disorientation, anxiety, a feeling of being out of place in a world that used to make sense. It is the developer who cannot explain why specification writing feels wrong, even though they understand the rationale intellectually. It is the architect who feels diminished by a process that values intent articulation over system design. The body continues producing the old practices because the habitus has not adapted. This is the Bourdieusian mechanism for what Giddens describes as the loss of ontological security. When the field shifts but the habitus has not caught up, people experience existential anxiety. The drive to restore security is the drive to resolve hysteresis, and the fastest resolution is reversion: go back to what the habitus already knows.

Weick’s Mann Gulch firefighters experienced hysteresis in its most extreme form. Their habitus (trained firefighter with tools) no longer matched the field (wildfire requiring flight). Dropping the tools meant abandoning not a possession but a self. Bourdieu explains why most could not do it: the tools were embodied cultural capital, the physical expression of professional identity.

If the new field conditions are sustained, reliably and consistently, habitus will eventually adapt. New dispositions will form. The specification writer will develop the same automatic fluency the code writer once had. But if the organisation allows reversion, if old processes reassert themselves, if leadership loses nerve, the habitus snaps back with the relief of returning home. And the next transformation attempt starts from a worse position, because the habitus now includes the disposition to treat transformation as temporary.

5. From Diagnosis to Practice

Bourdieu is better at explaining why things stay the same than how they change. But the framework is not as deterministic as it appears. Hysteresis itself is the opening. When the field changes faster than the habitus can adapt, the mismatch produces reflexivity: a moment where dispositions that normally operate below consciousness become visible. The developer who notices that their habitus is pulling them toward code rather than specification is experiencing reflexivity. That noticing is rare, uncomfortable, and fleeting, but it is the crack in the reproduction cycle.

Weick’s small wins are the mechanism for exploiting this crack. Each successful specification deposits a thin layer of new disposition over the old. No single win is sufficient. But accumulated wins, sustained over time, gradually build a new habitus. The sequence is always practice first, disposition second: you do the new thing, repeatedly, until doing it becomes what your hands produce automatically. You cannot think your way into a new habitus. You must act your way there. Bandura’s mastery experience is the same process in psychological terms: self-efficacy changes through doing the thing and succeeding, which is habitus adaptation through changed practice.

Bourdieu governs the Identity dimension of the Learning phase because he explains the deepest mechanism by which identity resists change and the only mechanism by which identity genuinely transforms. In the Deciding phase, the governor shifts to Simon: where Bourdieu constrains through habitus (the sociological limit on what is available to the person), Simon constrains through bounded rationality (the cognitive limit on what is available to the decision-maker). Both govern through constraint on what is available. Both explain why people act within a narrower range than their situation permits. The leader who wants to improve both learning and decision quality must work on two fronts: the sociological (changing the habitus through new practice, new exposure, new fields) and the cognitive (redesigning the decision environment so the right information reaches the right people). Bourdieu tells you why the first is necessary. Simon tells you why the second is.

(An Organisational Prompt is something you can do now....)

Organisational Prompt

Identify the three roles most affected by your AI adoption initiative. For each, ask: what cultural capital currently sustains their professional standing? What can they do that others cannot? What gives them prestige? Now ask: which of these forms of capital does the transformation threaten, and which does it potentially enhance?

The critical question: have you designed any mechanism to help people convert their existing capital into forms the new field will value? The developer’s ability to decompose complex logic is cultural capital that transfers directly to specification writing. The architect’s ability to see structural patterns transfers directly to domain modelling. These are not new skills to be learned from scratch. They are existing capital to be converted. If you have not made the conversion path visible, you are asking people to accept devaluation and call it progress. The resistance you encounter is not change fatigue. It is rational behaviour in a field where the rules are shifting and nobody has explained what the new currency is.

Further Reading

Pierre Bourdieu, The Logic of Practice (1990). The most developed theoretical account of habitus, field, and capital. Dense but essential.

Pierre Bourdieu, Distinction: A Social Critique of the Judgement of Taste (1979/1984). How cultural capital reproduces social hierarchy. Read it for the mechanism by which “competence” is naturalised as individual quality rather than structural advantage.

Pierre Bourdieu and Loïc Wacquant, An Invitation to Reflexive Sociology (1992). The most accessible introduction to the conceptual triad. Start here.

Anthony Giddens, The Constitution of Society (1984). The structural counterpart. Giddens’ practical consciousness and Bourdieu’s habitus are two perspectives on the same phenomenon.

I write about the industry and its approach in general. None of the opinions or examples in my articles necessarily relate to present or past employers. I draw on conversations with many practitioners and all views are my own.

Gregory Bateson was an anthropologist who studied schizophrenia, an epistemologist who studied dolphins, a cyberneticist who studied alcoholism, and a philosopher who studied octopuses. He never held a conventional academic appointment for long. People who read him tend to describe the experience as bewilderment followed by the suspicion that he understood something fundamental that nobody else had quite articulated. That suspicion is correct.

Bateson’s contribution to organisational learning is foundational and largely unacknowledged. Argyris built his distinction between single-loop and double-loop learning on a framework Bateson published first. Senge borrowed Bateson’s systems epistemology without fully crediting its origin. Stacey drew on Bateson’s cybernetics while arguing against the possibility of the system design Bateson’s contemporaries believed in. Bateson is the deep source that several thinkers in this series have drawn from, often without saying so. He provides the epistemological foundation beneath the entire Learning phase: a theory of what information is, what learning is, and why organisations systematically address both at the wrong level.

1. Levels of Learning: The Diagnostic Engine

Bateson borrowed from Russell and Whitehead the theory of logical types: the class of all chairs is not itself a chair. Confusing a member with its class produces paradox. Bateson applied this to learning and produced a hierarchy that remains the clearest map of what organisations are actually doing when they say they are “learning.”

• Learning 0 is no learning at all. The system receives a signal and responds identically every time. A thermostat at a fixed setting. A policy manual that nobody revisits.

• Learning I is what most people mean by the word. You correct errors within a given set of alternatives. A programmer learns a new syntax. A team learns a new deployment process. The governing framework is not questioned; behaviour within it is adjusted. This is Argyris’s single-loop learning, and it accounts for the vast majority of what organisations call training.

• Learning II is where things become genuinely difficult. Bateson called it deutero-learning: learning to learn. It is not a change in behaviour within a context but a change in how the person recognises what kind of context they are in. A person operating at Learning II does not simply learn a new AI tool; they learn what category of problem calls for AI and what category does not. They do not acquire a new skill; they revise their understanding of what kind of professional they are becoming. Learning II changes character. It is Argyris’s double-loop learning: questioning the governing variables, not just the actions they produce.

• Learning III is transformation of the premises that generate Learning II. Bateson described it as experiences in which the self must be reconceived. Kegan’s developmental transitions from the socialised mind to the self-authoring mind are Learning III. Heifetz’s deepest adaptive challenges require it. Almost no organisation achieves it deliberately.

The critical insight is not that these levels exist but that confusing them produces pathology. Every barrier to learning diagnosed in this series is a mechanism that keeps the organisation at Learning I when the challenge requires Learning II. Argyris’s defensive routines prevent governing assumptions from being questioned (Learning I lock). Dweck’s fixed mindset ties identity to current competence, making frame-questioning feel like identity threat (Learning I lock). Seligman’s learned helplessness is a pathological form of Learning II: the person has “learned to learn” that effort is futile (wrong Learning II). Weber’s iron cage optimises within the bureaucratic frame while making the frame itself invisible (Learning I lock). Giddens’s practical consciousness reproduces the frame through daily practice below the threshold of awareness (Learning I lock). Kahneman’s System 1 generates coherent narratives that suppress awareness of what the frame excludes (Learning I lock). Dekker’s blame culture treats systemic failures as individual errors: a Learning I correction applied to a Learning II problem (category error).

Every one of these barriers is a category error in Bateson’s sense: a Learning I intervention applied to a Learning II challenge. And every AI programme that begins with “let’s upskill everyone on prompt engineering” and ends with no discernible change in how the organisation operates has made precisely this error. It addressed a level-II challenge with a level-I intervention. The intervention was not wrong. It was at the wrong level.

2. Information: A Difference Which Makes a Difference

Bateson defined information as “a difference which makes a difference.” This sounds tautological until you grasp its implications. Information is not data. Data is what is recorded. Information is what changes something. A signal that nobody receives is not information. A report that nobody reads is not information. A strategy document that changes no behaviour is not information. It is data that has failed to become a difference that makes a difference.

This definition grounds every Information-related finding in the series. Westrum’s typology describes the cultures that determine whether differences reach the people who need them: pathological cultures suppress differences, bureaucratic cultures channel them through processes that strip context, generative cultures let differences flow to where they can make a difference. Weick’s sensemaking is the process by which raw differences become meaningful: people select cues, fit them to frames, and act on interpretations. Kahneman’s WYSIATI explains why some differences never register: the mind constructs a coherent story from available data and has no awareness of the differences it has excluded. Dekker shows what happens when the most important differences, the weak signals of drift, are filtered out by a blame culture that punishes the messenger.

The organisational implication is precise. An organisation’s capacity to learn is not determined by the quality of its people or the sophistication of its strategy. It is determined by the quality of differences its communication patterns can register and act upon. Beer’s cybernetics formalises this: variety (the number of distinctions a system can make) must match the variety of the environment the system needs to navigate. An organisation that cannot register the difference AI makes to its domain, because its channels filter unfamiliar signals, because its reward structures penalise messengers, because its categories have no place for what is emerging, is not a system with a learning problem. It is a system whose ecology of mind cannot yet accommodate the information it needs.

3. The Double Bind: Why “Innovate” and “Don’t Fail” Cannot Coexist

Bateson developed the concept of the double bind in 1956: a communicative situation in which a person receives contradictory messages at different logical levels and cannot comply with one without violating the other, cannot leave the situation, and cannot comment on the contradiction.

Consider the AI transformation double bind operating in most large enterprises. At one level: “Experiment. Be bold. Innovate with AI.” At another: “Deliver predictable outcomes. Do not disrupt existing revenue. Explain every failure.” The employee cannot comply with both. They cannot leave. And they cannot safely say “these two things you are asking me to do are incompatible,” because raising the contradiction risks being labelled as resistant.

The result is what Bateson would predict: performance theatre. People learn to perform innovation without actually innovating. They produce proofs of concept that will never reach production. They attend hackathons and build demos that demonstrate possibility without threatening any existing process. The activity satisfies the first injunction; its harmlessness satisfies the second. The double bind is not resolved but managed through theatre.

Argyris diagnosed the same phenomenon: the gap between espoused theory (”we value innovation”) and theory-in-use (”we punish failure”) produces defensive routines. Bateson explains why the routines are so resistant: they are rational adaptations to contradictory communication at different logical levels. You cannot fix a double bind by addressing either message in isolation. You can only fix it by making the contradiction visible and resolving it at the level above both messages. Edmondson’s two-by-two matrix (safety crossed with standards) is an attempt to dissolve the double bind by aligning the messages: we expect excellence and we treat difficulty as learning. The messages reinforce each other instead of contradicting each other.

Giddens’s three dimensions reveal where the double bind lives structurally. The signification says “innovate.” The domination says “within these approval chains.” The legitimation says “and be rewarded for predictable delivery.” Three dimensions, two of which contradict the first. Until all three are aligned, the double bind is structural, not communicative, and no amount of clearer messaging will resolve it.

4. Ecology of Mind: Why Culture Is Not Inside Anyone’s Head

Bateson’s most radical idea is that mind is not located in the brain. Mind is constituted by the total pattern of information flowing through a system. The unit of survival is not the organism but the organism-in-its-environment. Thought does not happen inside heads. It happens in the circuit of interactions between heads, hands, tools, institutions, and the world.

If mind is in the system, then culture is not a property of individuals. It is a property of patterns of interaction. Changing culture does not mean changing how individuals think. It means changing the patterns of communication, feedback, and response that constitute the organisation’s collective cognition.

Giddens reached a compatible conclusion through sociology: structures are reproduced in the practices of the agents they constrain. Stacey pushed further: organisations are patterns of interaction that shift only when the interactions themselves shift. Bateson provides the epistemological foundation beneath both. If information is a difference which makes a difference, and if mind is the pattern of differences circulating through a system, then an organisation’s capacity to learn is a feature of its ecology of mind: the quality of differences it can register, the levels at which it can process them, and the patterns of interaction through which they circulate.

Bourdieu’s habitus is the individual expression of this ecology: the embodied dispositions formed by participation in the patterns. The habitus is not inside the person. It is the person’s participation in the ecology, carried in the body. When the ecology changes, the habitus eventually changes too, but only through sustained practice within the new patterns. This is why Weick insists that action precedes understanding: new ecology, new practice, new habitus, new understanding. The sequence cannot be reversed.

5. What This Means: The Level Must Match the Challenge

Bateson reframes the entire learning problem this series has been exploring. The question is not “how do we get people to learn AI?” but “at what level is the learning that AI demands, and are we intervening at that level?”

If AI is a new tool for existing work, Learning I is sufficient: train people, adjust processes. But if AI changes the nature of work itself, if professional identity is bound up in capabilities that AI now performs, if the task must be defined before it can be done, then the challenge is Learning II at minimum. And Learning II cannot be achieved by Learning I methods. You cannot train your way to a transformation of character. You cannot workshop your way to a new professional identity.

Most organisations do not fail at AI transformation because they lack strategy, budget, or talent. They fail because they are addressing a Learning II problem with Learning I tools, caught in double binds they cannot name, operating within an ecology of mind that filters out the very information they need. Bateson would not have been surprised. He spent his life pointing out that the most important problems are problems of logical type, and that the most common error is trying to solve a problem at the level at which it presents itself rather than the level at which it operates.

The deepest learning happens not when people acquire new skills but when they acquire a new relationship to the process of acquiring skills. That is Learning II. And it is the difference between an organisation that has adopted AI and an organisation that has learned to learn with it.

(An Organisational Prompt is something you can do now....)

Organisational Prompt

In your next leadership meeting, ask: “Where are we sending contradictory messages about AI at different levels?” Write both messages down. Put them side by side. Ask: “If someone on the front line took both of these messages seriously, what would they actually do?”

The answer will reveal whether your organisation is creating the conditions for learning or the conditions for performance theatre. Naming the contradiction does not resolve it. But it makes the contradiction discussable, and Bateson’s entire body of work shows that making the undiscussable discussable is the precondition for change at a higher logical level. The double bind persists because it cannot be commented on. The moment you comment on it, you have begun to dissolve it.

Further Reading

Gregory Bateson, Steps to an Ecology of Mind (1972). The essential Bateson. “The Logical Categories of Learning and Communication” is the chapter most directly relevant. Dense, rewarding, and unlike anything else in the management reading list.

Gregory Bateson, Mind and Nature: A Necessary Unity (1979). Bateson’s attempt to make his ideas accessible. More structured than Steps and a better starting point for readers new to his work.

Nora Bateson, An Ecology of Mind (2010). Documentary film. Conveys the texture of Bateson’s thinking better than any summary.

Gregory Bateson, Don Jackson, Jay Haley, and John Weakland, Toward a Theory of Schizophrenia (Behavioral Science, 1956). The original double bind paper. The specific causal claim about schizophrenia is no longer accepted, but the structural insight about contradictory communication at different logical levels remains devastating. Freely accessible.

I write about the industry and its approach in general. None of the opinions or examples in my articles necessarily relate to present or past employers. I draw on conversations with many practitioners and all views are my own.

Every large enterprise that attempts AI transformation discovers the same thing: the obstacle is not the technology. It is the organisation itself. Not the people, who are frequently talented and motivated, but the structures, procedures, approval chains, governance frameworks, and reporting hierarchies that were designed to produce precisely the qualities that transformation requires organisations to suspend: predictability, consistency, control, and the elimination of individual discretion.

The issue was diagnosed over a century ago by a German sociologist who never saw a computer. Max Weber described the mechanism by which rational organisations become incapable of the very adaptation their survival requires. He called the result the iron cage. It may seem odd to return to so early a thinker after the accumulated insight of this series, but his diagnosis is grounding in a way that more modern frameworks are not. Weber is not describing something that happened in the past. He is describing the thing that is happening to your AI transformation right now: a rational system that has become its own purpose, producing the appearance of progress while the real problems go unaddressed.

1. Why People Follow: The Basis of Authority

Weber began with a question most transformation leaders never ask: why do people in this organisation do what they are told? The answer determines what kinds of change are possible.

Traditional authority derives from custom: “the way we do things here” offered as sufficient justification. Charismatic authority derives from the extraordinary personal qualities of the leader: people follow because they believe in the individual. Rational-legal authority derives from a system of rules that apply equally to everyone: people follow because the role has legitimate power, regardless of who occupies it.

Rational-legal authority dominates modern organisations, and Weber argued it was historically inevitable because capitalism requires above all one thing: predictability. Every element of the system, approval processes, role definitions, budget cycles, performance metrics, exists to reduce variation and ensure consistency. Transformation requires the opposite: experimentation, tolerance for failure, the suspension of established procedures. You are asking a system designed to eliminate surprise to embrace it.

Heifetz would recognise the dilemma. The existing authority structures are not failing. They are succeeding at something that is no longer sufficient. This is an adaptive challenge: the rational-legal system works perfectly for the problems it was designed to solve. AI transformation is not one of those problems.

2. The Iron Cage: When Rationality Becomes Its Own Prison

Weber described bureaucracy not as a pejorative but as the most technically efficient form of organisation ever devised: precision, speed, continuity, strict subordination, reduction of friction. He summarised it as a machine. These are exactly the qualities that resist transformation, because transformation requires their opposites: imprecision (experimentation), slowness (learning), ambiguity (exploration), and discontinuity (abandoning what no longer works). The bureaucratic machine is not broken. It is doing precisely what it was designed to do. And that is the problem.

The iron cage operates through a specific mechanism. Weber distinguished between means-ends rationality (”given this goal, what is the most efficient way to achieve it?”) and value rationality (”is this the right goal?”). In a fully rationalised organisation, means-ends rationality progressively displaces value rationality. The question “are we achieving our targets efficiently?” crowds out the question “are these the right targets?” The system persists regardless of whether it serves its original purposes.

This is the lived experience of AI governance in most enterprises. The governance framework asks whether AI tools are being adopted according to plan, whether risk assessments are completed, whether usage metrics are on track. It does not ask whether the plan makes sense, whether the risk framework captures actual risks or manufactures procedural compliance, or whether the metrics measure anything that matters. Drucker saw this with characteristic clarity: his concept of systematic abandonment, regularly asking “if we were not already doing this, would we start now?”, is a practical tool for reintroducing value rationality into an organisation that has lost the capacity for it.

The iron cage is what institutional inversion looks like when it is complete. The governance function that was created to serve the organisation’s purposes has become the purpose the organisation serves. The approval process that was designed to manage risk now produces risk, because it prevents the experimentation through which the organisation would learn what actually works. Bateson would classify this as a Learning I lock: the organisation can optimise within the cage (correcting errors within the bureaucratic frame) but cannot question whether the cage itself is the problem (Learning II). The cage is the frame, and the frame is invisible from inside.

3. Status Groups: Why Professionals Defend What Transformation Threatens

Weber’s concept of status groups and social closure explains professional resistance with precision. Status groups derive collective identity from shared expertise, credentials, and practices. Social closure is the process by which they protect their position by restricting access to the resources that define their standing.

Developers derive status from coding skill. Architects derive status from system design knowledge. Managers derive status from controlling development resources. Each group’s identity is bound to their current way of working. When AI threatens to change what is valued, each group engages in closure: protecting the activities and knowledge that sustain their standing.

Bourdieu provides the mechanism beneath Weber’s observation. Each status group is a field with its own forms of capital. In a professional bureaucracy, the professionals’ capital, their embodied skill, their technical judgement, their peer recognition, is the dominant currency. Transformation that devalues this capital without providing a credible path to new capital will be met with rational resistance. The developer who resists specification-driven development is not being stubborn. They are defending the skill that defines their professional identity, and their habitus, formed through years of practice in the old field, generates that defence automatically, below the threshold of conscious choice.

Weick’s Mann Gulch analysis provides the vivid case: the framework within which the new practice would make sense has not yet formed, and the framework it is replacing has already begun to dissolve. The professional cannot adopt the new approach from within the old identity. They need a new identity, and identities are not adopted through training. They are formed through practice, over time, in conditions that support the transition. Heifetz names the leader’s task: acknowledge the loss. The accumulated capital is real. The expertise is genuine. Pretending the transition is costless produces the cynicism Seligman describes.

4. The Routinisation of Charisma: Why Transformative Energy Dissipates

Weber identified only one force capable of disrupting the iron cage: charismatic authority. A transformative leader or event can override established rules through the force of direct experience and conviction. But charisma is inherently unstable. It cannot scale, survive succession, or sustain itself once the initial energy fades. Weber called the inevitable process routinisation: the disruptive energy is channelled back into bureaucratic structures.

The pattern is visible in every AI transformation. The moment when a sceptical executive watches a domain expert’s specification become working software in minutes is a charismatic event. It creates a burst of energy that, for a moment, overrides the cage. But then the energy must be routinised. And the existing structures will absorb the disruption into their own logic. “We need to fundamentally rethink how we build software” becomes “we need an AI governance framework, an AI Centre of Excellence, an AI risk assessment process, and an AI adoption metric on the quarterly scorecard.” The iron cage does not open. It reconfigures.

Giddens provides the mechanism. The bureaucratic structures that need to change are the same structures shaping the behaviour of the people who would need to change them. You cannot step outside the iron cage to redesign it, because the cage is not merely around you. It is in your habits, your assumptions, your reflexive responses to uncertainty. Stacey offers a partial corrective to Weber’s pessimism: complex responsive processes, the unpredictable interactions between people, operate within and despite bureaucratic structures. The cage constrains but does not eliminate creativity, improvisation, and resistance. Change happens, though never in the way the planners intended.

5. Using Weber Diagnostically

Weber’s value for practitioners is not as a prescription but as a diagnostic. He does not tell you what to do. He tells you what you are dealing with.

What type of authority dominates? If traditional authority is strong, transformation requires challenging sacred customs while preserving genuine institutional knowledge. If rational-legal authority dominates, transformation requires creating sanctioned experimental spaces within the rule system. If charismatic authority is driving the initiative, transformation will be easy to start and very difficult to sustain, because everything depends on continued executive attention.

What is the means-ends vs value rationality balance? If the organisation speaks exclusively in efficiency language, value arguments will not land. Frame AI adoption in terms the existing rationality can process while simultaneously creating spaces where the deeper questions can be asked. Peters diagnosed this as “rational-analytic management”: organisations so committed to analysis and measurement that they lose the capacity for judgement and action.

Which status groups are threatened? Map the professional groups whose identity is bound to practices AI will change. Understand that their resistance is rational given the current incentive structures. Then change the incentive structures before demanding new behaviour. The developer whose performance review measures lines of code is being perfectly rational within a bureaucratic system that has not updated its definition of value. Dweck’s fixed mindset is not individual psychology in this context. It is reinforced by bureaucratic structures that treat ability as static and measurable, that reward being right over being curious, and that make visible struggle a career risk. Weber would argue that bureaucratic organisations select for fixed mindset, because fixed mindset produces the predictable behaviour the machine requires.

(An Organisational Prompt is something you can do now....)

Organisational Prompt

Take your AI transformation initiative and trace the path of a single new idea from inception to implementation. Map every approval, review, sign-off, governance checkpoint, and committee discussion it must pass through. For each step, ask two questions: “Does this step exist to reduce risk, or to distribute accountability?” and “Was this step designed for the kind of work AI transformation involves, or was it inherited from a different era?”

You will almost certainly find that the majority of steps were designed for predictable, sequential software development and have been applied, without modification, to a domain that is experimental, iterative, and unpredictable. Each step made sense when it was created. Together, they form what Weber recognised: a rational system that has become its own purpose. The iron cage is not visible from inside. It feels like “how things work.” The diagnostic task is to make the cage visible.

Further Reading

Max Weber, Economy and Society (1922/2019). The foundational text on bureaucracy, authority, and rationalisation. Read the sections on legitimate domination and the ideal type of bureaucracy. Dense but indispensable.

Max Weber, The Protestant Ethic and the Spirit of Capitalism (1905/2003). Weber’s account of how a religious ethic became the cultural foundation for rational capitalism. Read it for the “iron cage” passage and for the argument that modernity’s defining feature is not technology but rationalisation.

Peter Baehr, The Iron Cage and Its Alternatives (2001). Corrects decades of misreading and recovers what Weber actually argued. Essential context for anyone citing Weber in a transformation setting.

I write about the industry and its approach in general. None of the opinions or examples in my articles necessarily relate to present or past employers. I draw on conversations with many practitioners and all views are my own.

The people who are supposed to adopt a new way of working often do not adopt it. Not because they disagree with the strategy. Not because they lack access to the tools. Not because they have not been trained. Because they do not believe they can do it.

Albert Bandura identified this as the single most powerful predictor of human behaviour: not what people can do, but what they believe they can do. He called it self-efficacy, the domain-specific belief in one’s capability to perform a particular task, and demonstrated across hundreds of studies that it predicts performance more reliably than actual skill, past experience, or incentive structures. People who believe they can do something attempt it, persist through difficulty, and recover from failure. People who do not believe they can do it avoid it, give up early, and interpret setbacks as confirmation of their inadequacy. The belief precedes the behaviour. And in AI transformation, the belief is almost always missing, because the capability being asked for is genuinely new and nobody has yet experienced themselves succeeding at it.

1. Self-Efficacy Is Not Self-Esteem

Self-efficacy is precise, specific, and domain-bound. It is not self-esteem, not general confidence, not optimism. It is the answer to: “Can I do this particular thing, in this particular situation, right now?”

A senior developer can have high self-esteem and zero self-efficacy for writing AI specifications. The general positive feelings do not transfer. They know, with the clarity that comes from decades of practice, that they are excellent at writing code. They have no basis for believing they can write a specification precise enough for AI to generate working software. They have never done it. They have never seen anyone like them do it. And their body, when they contemplate attempting it in front of colleagues, produces the unmistakable sensations of anxiety rather than readiness.

Bourdieu would recognise the mechanism. The developer’s habitus was formed through years of practice in a field that rewarded code-writing expertise. The dispositions that generate their sense of professional competence are calibrated to that field. AI transformation shifts the field: the new capability (specification writing) has no purchase in the existing habitus. The developer does not merely lack the skill. They lack the embodied basis for believing the skill is theirs to acquire. Their habitus tells them who they are, and who they are is not someone who writes specifications. The belief deficit is not cognitive. It is dispositional.

2. The Four Sources: A Hierarchy That Transformation Programmes Invert

Bandura identified four sources of self-efficacy, ordered by potency. Understanding this hierarchy is essential because most transformation programmes invest everything in the weakest sources and almost nothing in the strongest.

• Mastery experience, actually doing the thing and succeeding, is the most powerful source by a significant margin. Direct personal experience of success builds robust belief that transfers to related challenges and persists through setbacks. But the attribution matters: the person must believe the success resulted from their own effort, not from luck, excessive help, or an artificially easy task.

• Vicarious experience, watching someone similar succeed, is the second most powerful. The key word is similar. Watching an AI expert effortlessly generate code does almost nothing for the self-efficacy of a domain expert who has never written a specification, because the observer cannot identify with the model’s starting point. But watching a peer struggle, iterate, and eventually produce something that works is profoundly influential. The struggle is the mechanism. Edmondson’s psychological safety determines whether the struggle can be visible: nobody will model the learning process if modelling it carries social risk. Dweck’s growth mindset determines whether the struggle is interpreted as learning (valued) or incompetence (stigmatised).

• Verbal persuasion, being told you can do it by someone credible, is weak but not negligible. Effective persuasion is specific: “Your ability to decompose complex business processes into clear rules is exactly the skill that specification writing requires.” This bridges existing self-efficacy to the new domain. Generic persuasion (”I believe in you”) connects to nothing.

Physiological and emotional states, how you interpret your body’s signals, operate automatically. The same elevated heart rate can be read as excitement or as anxiety. The interpretation feeds directly into self-efficacy.

Now consider how transformation programmes allocate resources. They invest heavily in verbal persuasion: executive announcements, strategy presentations, training. They invest moderately in vicarious experience but typically the wrong kind: vendor demonstrations by experts rather than peer learning by similar others. They invest minimally in managing physiological states. And they invest almost nothing in mastery experience, the most powerful source, because mastery experience requires giving people real problems, real tools, and real time, which is expensive, messy, and does not fit a twelve-week change plan. The hierarchy is inverted. The organisation spends the most on what works least.

3. Mastery Experience and the Learning Loop

Bandura’s theory explains why demonstrations and early experiments are not marketing exercises. They are the primary mechanism by which organisations develop the psychological capability for change. When a team writes a specification and watches AI generate working code from it, something happens that no presentation can produce: the felt experience of “I did that, and it worked.” That is a mastery experience, and it builds self-efficacy in a way that is qualitatively different from any other intervention.

The design matters enormously. The first experience must succeed, not artificially, but the difficulty must be calibrated so that genuine effort produces a visible result. Csikszentmihalyi’s flow research provides the design principle: challenge matched to current skill. Too hard produces anxiety that reduces self-efficacy. Too easy produces dismissal. Success must be attributed to effort and strategy, not luck or help. And difficulty must increase progressively, which is where Ericsson’s deliberate practice connects: each step stretches capability without breaking confidence.

Bateson’s levels illuminate what is happening beneath the surface. The first mastery experience is Learning I: the person succeeds within a bounded frame. But the accumulation of mastery experiences, with progressive challenge, produces Learning II: the person’s model of their own capability expands. They are not just learning to write specifications. They are learning that they are the kind of person who can learn to write specifications. The identity shift, from “I am someone who writes code” to “I am someone who can articulate intent precisely enough for machines to act on it,” is the Learning II outcome that transforms the habitus. Each mastery experience is a small intervention in the field that Bourdieu describes, gradually forming new dispositions that generate new practice.

Deci and Ryan’s competence need connects here directly. Self-efficacy is the cognitive mechanism through which competence is experienced. When the specification-generate-validate loop works, the person is simultaneously building self-efficacy (Bandura) and satisfying their competence need (Deci and Ryan). The loop is a psychological engine, not merely a technical process.

4. The Verbal Persuasion Trap

Most organisations invest disproportionately in verbal persuasion because it is scalable and least disruptive. Training programmes, presentations, e-learning modules are all forms of verbal persuasion in Bandura’s framework. They tell people they can do something. They produce minimal self-efficacy change, because telling someone they can do something is fundamentally different from them experiencing it themselves.

Verbal persuasion works when three conditions are met: the persuader is credible and trusted, the message is specific (connecting the new challenge to existing capabilities), and the persuasion is immediately followed by an opportunity for mastery experience. Most training programmes meet none of these conditions. The trainer is external. The message is generic. And the training ends before any mastery experience is attempted. Participants return to their desks, discover the training did not prepare them for real complexity, and conclude they lack the ability.

Argyris identified this pattern. The espoused theory: “This training will equip you.” The theory-in-use: “This training will provide information that is necessary but insufficient, and the gap will be closed through practice, failure, and repetition that the programme neither provides nor acknowledges.” The gap is undiscussable because acknowledging it would undermine the business case for training.

The alternative is embedded learning through guided practice. Not training followed by application, but application guided by coaching in real time, on real problems, with real feedback loops. The specification-generate-validate loop is a self-efficacy building cycle when properly supported: the specification is the attempt, the generated output is the feedback, and the iteration is the mastery experience. But it requires someone alongside during the critical early attempts when self-efficacy is most fragile.

5. From Individual Belief to Collective Capability

Bandura’s later work extended self-efficacy to the group level: collective efficacy, the shared belief in a group’s capability to achieve a desired outcome. This matters for AI transformation because the capability required is not individual but collective. Writing a good specification requires domain expertise, technical understanding, and validation design, rarely possessed by a single person. The relevant belief is not “Can I write a specification?” but “Can we, this team, produce specifications that generate useful results?”

Collective efficacy is built through the same four sources at the team level: shared mastery experiences, watching a similar team succeed, credible affirmation from a trusted leader, and shared emotional states. Giddens would locate collective efficacy in practical consciousness: the taken-for-granted assumption that governs group behaviour. A team that believes “we can figure this out” does not articulate this in every meeting. It operates as a background assumption shaping what the team attempts and whether they persist. Moving a team from “we have been through this before and it never works” to “we can figure this out” requires precisely the mastery experiences Bandura describes. The assumption cannot be installed through communication. It must be built through experience, which is why Weick’s insistence that action precedes understanding applies as much to collective belief as to individual sensemaking.

Stacey would add that collective efficacy is not a property the leader can design into a team. It emerges from the quality of interaction between the people in the team. The leader’s role is to create the conditions: the right problem, the right support, the right level of challenge, the safety to fail visibly. What emerges from those conditions is the team’s to produce.

(An Organisational Prompt is something you can do now....)

Organisational Prompt

Identify the next AI learning activity your organisation has planned. Now redesign it using Bandura’s hierarchy. Ensure that every participant does the thing, not just hears about it, before the session ends. Make the problem real but calibrate the difficulty so that success is achievable with genuine effort. If the session does not end with every participant having experienced “I did that and it worked,” the session has failed, regardless of how much content was covered.

For vicarious experience: do not use an expert demonstrator. Find a peer who learned recently and is willing to share honestly, including the confusion, the struggle, and what they would do differently. The struggle is the mechanism, not a flaw in the presentation.

Further Reading

Albert Bandura, Self-Efficacy: The Exercise of Control (1997). The definitive academic treatment. The four sources of self-efficacy and the experimental evidence that belief in capability predicts performance across virtually all domains studied.

Albert Bandura, Social Foundations of Thought and Action: A Social Cognitive Theory (1986). The broader framework: observational learning, reciprocal determinism, and the mechanisms by which people, behaviour, and environment continuously shape each other.

Albert Bandura, Self-Efficacy: Toward a Unifying Theory of Behavioral Change (Psychological Review, 1977). The foundational paper introducing the four sources. Freely available.

K. Anders Ericsson and Robert Pool, Peak: Secrets from the New Science of Expertise (2016). The essential companion: self-efficacy explains why people persist or give up; deliberate practice explains how expertise develops when they persist. Together they provide the design principles for learning pathways that build both belief and capability.

I write about the industry and its approach in general. None of the opinions or examples in my articles necessarily relate to present or past employers. I draw on conversations with many practitioners and all views are my own.