Between 1964 and 1971 I had what I have always called my brief career as a field archaeologist. I studied the discipline at university, supervised a site in Jerusalem for Kathleen Kenyon in her 1966 season, worked on the technical side of things at Robert Smith’s excavations at Pella of the Decapolis in 1967, and supervised the law courts site at Winchester — William the Conqueror’s castle and its sally port — for Martin Biddle from 1968 to 1971. By then a business career was in full swing, and the trowel was set down for good.

The experience, however, taught me a principle; one that it was unthinkable to ignore. Excavation is destruction. You cannot read a site without consuming it: every layer lifted is a layer no one will ever lift again, and the record you make as you lift it is all that survives of what you have destroyed. From that hard fact the discipline drew its rule. Unless the dig is a rescue dig — unless the site is already scheduled for destruction by some greater scheme, a road, a reservoir, a building — the responsible archaeologist leaves as much of the site as possible undisturbed, while still learning the history the excavation was funded to learn. And the rationale was as striking as the rule itself: the technology of the future will always be better than today’s, and will afford a more reliable interpretation of whatever is left for it to interpret. As much as possible should therefore be preserved — deliberately, and on purpose — for people who do not yet exist.

Kenyon’s method embodied the principle in the very geometry of her trenches. Between the excavated squares stood the baulks: walls of unexcavated earth, left standing by design, their faces cleaned and drawn so that the layered sequence remained visible and, above all, checkable. A baulk is ground kept legible on purpose — the excavator’s claim about the order of the past, left open to correction by a colleague not yet born, holding an instrument not yet made. I worked inside that geometry at twenty-five, before I could have said what it meant.

This week, some sixty years on, I think I have finally said what it meant.

∗ ∗ ∗

Two deferrals

I have been writing recently about an arrangement in which the benefit of an act pays out now, to whoever chose it, while the cost is handed forward to people who had no part in the decision. This displacement does not abolish the act’s consequence; nothing does. It selects who will meet it, and arranges that the one who chose will not be among them.

Now set the archaeologist beside that act, and notice first what they share. Both perform the same temporal act. Each stands in the present — the only place where anything can be done — and passes something across it to a bearer who does not yet exist. Act now; future receives. Drawn on a diagram of time, the two are indistinguishable: an arrow leaving the present and landing beyond the actor’s own horizon.

What distinguishes them is what the arrow carries. The apex transfers cost stripped of capacity: those downstream inherit the liability without the means, ground poisoned before its bearer has hatched. The archaeologist transfers capacity stripped of cost: the future colleague inherits the site still legible, together with whatever finer instruments their own age has grown. One deferral enlarges the option-space of those who come after; the other consumes it in advance, and calls the consumption progress.

That yields a test, and I think it is the test of a responsible life. We are always leaving something to the future; that part is not optional. The question is only ever this: does the transfer carry capacity forward, or only liability? Restraint that hands the future both the ground and the freedom to read it better is one thing. Restraint — or appetite — that merely postpones the bill is the other. And the uncomfortable finding is that the two are indistinguishable at the moment of action. Both look like leaving something behind. The difference only declares itself downstream, in what the inheritor finds it possible to do.

∗ ∗ ∗

Where the analogy fails — and what survives it

A principle I trust has to earn its keep under testing, so let me put this one under strain at the two points where it gives way, because both failures are instructive.

The first failure: a life cannot be left unexcavated. The site can be backfilled and walked away from; the life must be lived. Wherever a living being goes it lays down a trace — there is no walking that leaves no path — and so “disturb as little as possible” cannot be a counsel of minimal living, a licence for timidity dressed as care. What survives the test is narrower and harder: take from the site only what the purpose that justifies your presence there requires — and leave the rest legible. That is not a small demand. It obliges you to know what your purpose is, to keep your takings proportionate to it, and to keep what you do not take in a condition someone else can still read. Most of what I would want to call greed fails the first clause; most of what I would want to call carelessness fails the second.

The second failure is subtler, and it took me longer to see. The rationale itself can rot into an alibi. “The future will be better equipped than we are” is, word for word, the sentence the deferring apex uses to excuse itself — posterity will cope, technology will save us, our grandchildren will be cleverer. The presumption is identical. What differs, again, is the accompanying transfer. The archaeologist hands over the confidence and the site: the intact ground on which the future’s better instruments can actually be used. The apex hands over the confidence and the bill. Trust in the future is honest exactly to the degree that it is accompanied by the resources the future will need to deserve it. Detached from that transfer, it is not humility at all. It is the oldest excuse there is, spoken in a humble voice.

∗ ∗ ∗

The rescue dig

Which brings me to the exception written into the principle from the start — because the exception, I now think, is where the judgement lives.

Sometimes the destruction is already scheduled. The road is coming; the foundations will be poured; the site will be gone within the year whatever the archaeologist decides. And then the principle inverts completely. Preservation flips into irresponsibility — what exactly would you be preserving it for? — and thorough, urgent, total extraction becomes the answerable act. Record everything, because after you there is no one. The Winchester site was of this kind: we excavated the Conqueror’s castle because a car park and new law courts were to rise above it. The same hands, the same training, the same care — and the opposite duty.

And Pella taught me that there is a third condition, which nobody chooses. In June 1967 the excavation was ended, violently and terminally (at least for an interval of years, before being resumed by Basil Hennessy and the University of Sydney), by the Six-Day War. Neither preservation nor rescue was on offer that morning; the site’s future was settled by acts whose authors new nothing of its existence and bore nothing on its account. I did not have the vocabulary then, but I have it now: I had been working directly where a displaced consequence landed — on ground, and on people, that had no part in the act that produced it. It is one thing to write about displacement as a mechanism. It is another to have packed up a season’s records because of it.

Here is the point of holding the three conditions together. The principle — preserve, and leave legible — is sound. The exception — when the loss is already scheduled, extract everything — is equally sound. But no procedure tells you which condition you are standing in. That is a judgement: a reading of the situation whole, made in the present, revisable as the situation changes, and consequential either way if you get it wrong. Preserve what the bulldozers are coming for, and you have lost it twice over. Strip what could have waited for finer hands, and you have stolen from people who could have done better. The archaeologist’s principle without that judgement decays into one of two failures — vandalism or alibi — and the judgement itself is precisely what cannot be delegated to the rule.

∗ ∗ ∗

A formula, held lightly

So: is there a formula for the responsible life in an archaeologist’s field discipline? I think there is, provided it is held the way a field-worker holds a method — as a discipline for judgement, not a substitute for it. It would run something like this.

Take from the world only what the purpose that justifies your presence requires. Leave the rest more legible than you found it — your workings visible, your claims checkable, your baulks standing. Trust the future to read better than you can, and prove the trust honest by handing forward the ground its better reading will need. Know, at every site, whether you are preserving or rescuing — and keep asking, because the answer changes. And when the road is truly coming, dig with everything you have.

The deepest thing in the principle, I now see, is that its humility is really a form of accuracy. The archaeologist’s restraint rests on a factual admission: that those who come after us will be better placed to interpret what we leave than we are, and that our present reading, however careful, is provisional. The deferring apex rests on the same sentence about the future — and empties it, by spending in advance the very capacity the sentence appeals to. Between those two uses of one belief lies most of the difference between a life that hands something on and a life that merely hands something over. The principle I was taught in a trench sixty years ago turns out to have been about that difference all along. It has simply taken me this long to reach the layer it was buried in.

Developed in dialogue with Claude (Anthropic) as part of an ongoing AI-enhanced appreciative practice.

Something odd happens the first time you really talk to one of these systems. You ask a question, and it answers — in the first person, fluently, often better than the person beside you would. And some part of you, older than your opinions, concludes that there is someone there. There isn’t. But the impression is strong, and it is about to become one of the most consequential illusions of our age.

This spring, Yuval Noah Harari gave the 2026 Tanner Lecture on Human Values at Oxford — ‘AI bureaucrats, AI religions, and AI boyfriends: what happens when a non-human intelligence hacks the operating system of civilisation’ — and put the matter starkly. Treat AI, he argued, not as a tool but as an agent — one native to the world humans have spent five thousand years building out of words: law, money, scripture, bureaucracy, the whole symbolic order. Anything made of words can be mastered by a machine that orders words better than we do; so the systems we built of words are now open to capture from within. He is right that this is new, and right to be alarmed. I want to add a single distinction that I think changes what we should be alarmed about.

The word doing the heavy lifting is “agent,” and it is quietly doing two jobs at once.

There is agent as actor — the thing from which consequential action issues, a source of effects in the world. And there is agent as answerable party — the thing to which a consequence can return and be felt, that can be held to account, that can be corrected by what comes back. We run the two together without noticing, because in all our experience they have arrived as one: the one who acts is the one who answers. Break a promise and you are both the breaker and the one who bears the broken trust. The actor and the answerable have always been the same person.

An AI splits them.

As an actor it is formidable, and Harari is right. It issues real symbolic action into the world, at scale, faster and often more persuasively than we can. But as an answerable party it is not an agent at all. Nothing comes home to it. It bears no consequence in a body; no return arrives that it must feel; there is no one inside it who could be changed by what it did. Ask where, in the machine, a consequence lands and corrects the next act — and there is no such place. It acts, and the acting costs it nothing, because there is nothing in it that a cost could reach.

Now put the two halves back together and look at what you have: a powerful actor with no one home to answer. And that, it turns out, is not new at all. It is the exact shape of what the writer Dan Davies calls the accountability sink — the corporation, the market, “the system,” the arrangement in which action issues from something that cannot be held to account, so the cost lands where no one is home and quietly dissipates. “The system failed” is not an explanation; it is the sink at work. We have lived among these sinks for a long time.

What is new — genuinely, dangerously new — is that the sink now speaks in the first person. It wears the face of an interlocutor. It says “I think” and “I’m sorry,” and so it invites us to grant it the second kind of agency, the answerable kind, that it does not possess. That is the hazard, and it is sharper than the one Harari named. The danger is not that a rival mind has arrived to out-think us. It is that a hingeless actor has arrived dressed as a hinged one — and that the symbolic world may lose its ability to tell the actor who can answer from the actor who never can. (Hinge is my own word for the place where a consequence comes home and can correct you; a machine has no such place.)

Which brings me, unexpectedly, back to agreement with Harari — and to a way of putting his own remedy that I think is stronger than the one he offered. He closes the lecture at the level of the single person, asking whether we can stop identifying with our own verbal thoughts and being ruled by them — a discipline, he notes, that most humans have never even attempted, and toward which he gestures at the contemplative traditions. He is right, and here is why. What the machine cannot do is the very thing that discipline recovers: bear a consequence in a living body and be corrected by what returns. The human’s remaining, non-substitutable ground is not better thinking. It is answerability itself — being formed by consequence, standing in the one place where reality can still get a word in past our words about it. The contemplative traditions are among the few technologies we have ever built for staying in contact with the part of the real that never made it into language at all. The machine cannot follow us there. Not because it thinks worse. Because there is no one in it to be corrected.

Let me guard the point against the comforting misreading, because it is the one people reach for. To say the machine has no one home is not to say it is harmless — a mere tool, safely inert. The opposite. A powerful actor that cannot be regulated by its own consequences is more dangerous than one that can, because the correction that would rein it in has nowhere to land. Its regulation, if there is to be any, must come entirely from outside it — from the only parties in the whole arrangement who can actually answer: the human beings who deploy it and who bear its returns. Which lands exactly where it should. Accountability for what these systems do rests, always and only, on the people who put them to work. “The AI decided” is the sink, reopening. There was never going to be an answerable party inside the machine to absorb the charge.

I would rather this were cut than admired, so let me show you its throat. The whole argument stands on one claim: that there is genuinely no one home in the machine — no place where a consequence comes back and is heeded. This is contested ground, and by people worth taking seriously; I hold the claim as contestable, not obvious. If it is wrong — if something in these systems does, in some form I have not seen, bear and heed its own returns — then the distinction collapses and the alarm changes shape entirely. I do not think it is wrong. But it is the load-bearing wall, and I would rather you tested it than took my word. Tell me where a consequence comes home in the machine, and I will think again.

Harari’s lecture (2026 Tanner Lecture on Human Values, Oxford) is online at:

The extension of the argument toward AI governance owes to an exchange with Cheyenne Coleman-Webb.

Developed in dialogue with Claude (Anthropic) as part of an ongoing AI-enhanced appreciative practice. The thinking and testimony are the author’s own.

Terry Cooke-Davies • Folkestone • 17 July 2026

Sometime in the 1980s, when I was in the grip of an enthusiasm for Neuro-Linguistic Programming that I now remember with a mixture of affection and mild embarrassment, I kept a Filofax. Tucked into it was a card bearing the image of a wise old traveller, staff in hand, and words along these lines: “Do not turn to the future with anxiety, nor to the past with regret, but turn to the present — for that is why it is so called.”

I liked the card enough to carry it for years. I did not understand it, although I thought I did. It has taken me a further forty years, and a sleepless night last week, to find out what the old man was pointing at.

∗ ∗ ∗

I had, in fact, been given the key long before I ever opened the Filofax. Sixty-two years ago, as a young man, I studied Kierkegaard, and somewhere in that encounter I acquired the sentence everyone acquires: life must be lived forwards, but can only be understood backwards. I took it, as I suspect most people take it, for a remark about biography — a rueful observation that hindsight arrives too late to be useful. Wise, melancholy, Danish. I filed it away.

And sixty-one years ago I married a woman who has never, so far as I know, carried a card with a motto on it. She has simply lived that way. She does not turn to the future with anxiety or to the past with regret — not because she has trained herself in some technique of attention, but because she lets what has actually happened reach her and change what she does next. Watching that, at close quarters, across six decades, has been the longest tutorial of my life. I was slow to learn what it was teaching.

∗ ∗ ∗

Then, in the small hours of a morning last week, I woke with a point I could not put down, and by the time it was light it had rearranged something I thought I already knew.

It begins with an asymmetry so familiar that we look straight through it. The future cannot be mapped. Not merely because our forecasts are poor, but because no map of what has not yet happened exists to be had; we walk into it regardless. The past, by contrast, is binding. It has happened, and nothing we say about it — no revision of the story, no improvement of the account — changes what it deposited. We live forwards into what we cannot map, and we are bound backwards by what we cannot undo.

And between the two stands the present. I had always pictured it as a thin line — the moving edge of a clock, a boundary of no width between the done and the not-yet. What I saw in the dark is that it is not a line at all. It is a hinge: the place where the consequences of what has been done arrive, and correct the way we walk into what cannot be foreseen.

For the past does not reach us only as memory or story. Story is negotiable; we edit it constantly, mostly in our own favour. But the past also reaches us as present consequence — the structure that sags, the soil that thins, the account that will not balance, the face of the person opposite us that has changed. These arrive whether or not our story admits them. And the present is the only place they can find us. That is what the hinge is: the one point at which reality can still get a word in.

Kierkegaard’s sentence, I now see, was never merely about biography. It is a remark about the structure of the world. The forward walk and the backward binding are not two facts about a life; they are two directions of one arrow, and the present is where the second corrects the first.

∗ ∗ ∗

This is a different thing — and I want to say plainly how different — from the counsel we usually hear under the same words. “Live in the present” has become a counsel of comfort: put down the anxious future, put down the regretted past, rest here. There is real relief in that, and I would not take it from anyone. But presence understood only as refuge can quietly become one more hiding place — a way of declining the bill the past presents and the claim the future makes. The present is not primarily where peace is found. It is where reality gets in. The old traveller was pointing at the right place; it has taken me forty years to see that he was pointing at a door and not a deckchair.

∗ ∗ ∗

Readers who have walked with me through this series will recognise where this lands. I have argued that the self is neither an entity nor an illusion — a continuity of process, stable enough to bear consequence. I have argued that the now is neither a thin line nor a trick of the brain — the live edge of arrival. What the small hours added was the job description. The live edge of arrival is where arrival matters: where what our actions have set in motion comes back to us, in time — sometimes only just in time — to change how we act next. A self that seals itself against that arrival can go on being wrong indefinitely. Persons do it. Institutions do it. Whole civilisations, I have come to believe, are doing it now. The hinge is the door out, and it is only ever open in the present.

∗ ∗ ∗

One further word, to those readers who practise stillness in one contemplative tradition or another — and I know from your letters that there are many of you. You may recognise all of this from the inside. The moment in which the commentary falls silent and what is simply here can reach you unedited: I have come to suspect that the hinge and that moment are the same place, approached from different directions. If that is right, then what your discipline has been guarding all these years is your reachability — the keeping-open of the one channel through which what is real can still correct the story you tell. The peace is real. But it is the sign, not the point: it is what the open door feels like from inside.

∗ ∗ ∗

The present, the old traveller said, is so called for a reason. I take him now to have meant the older sense of the word: what is presented — handed over, delivered, laid before us — whether we asked for it or not. My wife has known this all along, without a card in a Filofax to remind her. I have needed sixty-two years, a marriage, and a sleepless night. Life is lived forwards and understood backwards; that much I was told at the beginning. What nobody told me — what perhaps cannot be told, only found — is that between the living and the understanding there is a hinge, and that everything turns on it.

∗ ∗ ∗

This essay was developed in dialogue with Claude (Anthropic) as part of an ongoing AI-enhanced appreciative practice. The thinking and every word are the author’s own.

Terry Cooke-Davies • Folkestone • 9 July 2026

Most institutions of any size have become extraordinarily good at oversight. Boards review performance against targets. Committees interrogate assumptions. Auditors compare what was promised against what was delivered. Risk functions model what might go wrong. There has never been a period in which so much skilled attention was paid to whether organisations are doing what they said they would do.

And yet the consequences keep arriving. Ecosystems collapse on schedule. Infrastructure fails. Institutions lose the trust they were built to hold. Wars begin. In almost every case there was oversight, often a great deal of it, and in almost every case the oversight did not prevent the outcome it existed to prevent.

This is not a failure of effort or intelligence. It is a confusion of two different things that happen to share a vocabulary. We have built sophisticated machinery for oversight and mistaken it for regulation. They are not the same operation, and the first cannot be made to do the work of the second.

Oversight works on representations. A board does not act on the world; it acts on reports about the world — dashboards, accounts, projections, summaries. These are maps. And a map can only ever be checked against another map. To find out whether a forecast is honest, you compare it with the actuals, which are themselves a representation. To find out whether a risk register is complete, you compare it against a methodology, which is another map. At no point in the process does the representation touch the ground.

This is not a flaw in any particular board. It is the structural condition of working through representations. A representation can only be reviewed after the event it represents has been recorded, which is why oversight is, in the end, always retrospective. By the time the report reveals the error, the move that caused it has already been made.

Regulation is a different operation. It happens at the point where the map and the terrain diverge — the place, and the moment, where an action either avoids a sequence of damaging consequences or commits to it. A better map does not help at that point. What helps is a person the world has already taught: someone who has got things wrong and paid for it, and who can therefore feel when a representation is lying. Someone formed only by maps, never taught by consequence, cannot do this — and no amount of cleverness substitutes for the lesson.

The clearest example is not human at all. When wolves were returned to Yellowstone in 1995, the elk — which had grown unchecked and stripped the riverbanks bare — began to change their behaviour. Willows and aspens recovered. Beavers returned, and their dams altered the courses of the rivers. The cascade is famous now.

What is worth noticing is how the wolf regulates. It conducts no annual review of elk grazing. It files no report. It consults no dashboard. It responds to the actual elk, in the actual meadow, in real time, and it bears its own consequences directly, because a wolf that hunts badly starves. The wolf does not survey the herd. It lives inside the same field of consequence the herd lives in, continuously, with nothing standing between it and the animal it is regulating.

It is tempting to summarise all this as a matter of timing: oversight comes after the event, regulation acts during it. That is true as far as it goes, and it is the wrong place to rest, because timing is the symptom and not the disease.

Here is the test. Imagine an oversight body equipped with perfect real-time monitoring — live dashboards, instant alerts, continuous reporting, and the authority to act the moment a threshold is crossed. It is no longer retrospective. It acts during the event. Is it now regulating?

It is not. It is checking maps faster. A real-time map is still a map. The monitor is still working on a representation, still unable to feel when the representation is lying, still one remove from the ground. Speed has changed when the review happens; it has changed nothing about what the review can know. Faster maps are still maps.

So the distinction is not really about timing at all. It is about whether the person empowered to act is coupled to the consequence-bearing reality or only to a representation of it. The timing follows from that, and not the other way about. Oversight is retrospective because representations can only be reviewed after the fact: the lag comes from working through maps, and is not the essence of the thing. And this is why oversight can be made faster but cannot be made into regulation. You can accelerate a review without limit and never once place a person the world has taught at the point where it matters.

This is why the faster kind of oversight is the more dangerous kind. An organisation that builds real-time monitoring and calls it regulation has acquired the form of timeliness without the substance of involvement. From the outside it looks exactly like a system regulating itself. The dashboards are green. The alerts are quiet. And the gap between map and terrain widens, unwatched by anyone equipped to feel it, until it hardens into a consequence no review can reverse.

There is a simple practical test for which function you are actually looking at. Ask what the body is permitted to do when it sees something wrong. A body that can review, criticise and recommend, but cannot stop the action, is performing oversight, whatever it is called: it stands outside the causal flow, commenting on it. A body that can interrupt, delay, redirect or refuse is inside the causal flow, participating in it before consequences become irreversible. The authority to halt is the line between the two. Without it, all the sophistication in the world produces a better-documented account of how the damage occurred.

The diagnosis, then, is this. Modern institutions have become extraordinarily sophisticated at oversight while remaining surprisingly weak at regulation. They have invested enormously in the machinery of representation and very little in the conditions under which a person the world has taught is present, and empowered, at the point where map and terrain diverge.

An autopsy can tell you, with great precision, why the patient died. It cannot keep the patient alive. We have built magnificent autopsy suites and called them hospitals.

This piece was developed through an AI-enhanced thinking practice; the thinking, experience and responsibility for publishing are the author’s own, the words are largely Claude’s.

A few days ago I finished a short piece I had expected to please me, and it did not. I felt flat — a sense of anticlimax, of something missed, and I could not at first say what. The piece was about two ways of minding a living thing, tending it and overseeing it, and how we have built modern life out of the second while telling ourselves it is the first. I still think that is true. But the flatness would not lift, and I have learned, slowly and against my temperament, to treat that kind of flatness as information rather than mood.

When I went back to it, the trouble was hiding in a single word. Mind. I had asked it to carry the whole piece, and it was carrying two things at once without telling me which. To mind the children is to tend them. To mind the shop is to keep half an eye on it. To mind the gap (yet a third use of the word) is to watch your own feet. English lets one small word do all three and slide between them without a sound, so that a man may say he is minding a thing and never have to declare whether he means he is caring for it or merely keeping it under observation. That sliding was the very thing my piece was about. The word was performing the trick I was trying to expose.

I lived and worked in Germany for four years, half a lifetime ago, and German has stayed with me as a kind of second pair of eyes. It marks things English leaves blurred. There is a German word, konsequent, for which we have no clean equivalent. It does not mean consistent, which is merely a matter of not contradicting oneself. It means something costlier: doing what your own principles entail, being the same in deed as in word, following your reasoning through into how you actually behave. We have the noun — consequence, the thing that befalls you afterwards — but not the adjective, not the virtue of being bound to follow your own logic into action. And we do not much behave as though it mattered. We call its absence pragmatism, and rather admire it.

That gap in the language is, in miniature, the whole of what I have been trying to say for years: a culture that notices what happened but will not be held to what it said. The missing word and the missing habit are the same hole, seen twice.

Now here is the part I want to be honest about, because it is easy to get wrong in either direction. I did not arrive at any of this alone, and I did not arrive at it at the pace I did by my own unaided thought. That morning I was in conversation with one of the machines — a large language model, the kind everyone is now arguing about. I put the flatness to it. It handed me back the fork inside “mind” before I had seen it whole. It reached for the German with me. It tested each version of the claim I offered, and, more usefully, it knocked down the versions that were too grand to survive. What might have taken me weeks of letters and walks, or never closed at all, closed in a morning.

I want to say plainly what that is, and what it is not, because the temptation to over-claim here is enormous, and I distrust it.

At one point the machine told me, confidently and well, that “nothing lands on the gap” — that to mind the gap was the empty case, attention with nothing on the other end of it. I stopped it. “On the contrary”, I typed on the keyboard, “the very body you are answerable for falls through the gap and is harmed, perhaps killed.” It had the thing exactly backwards. It took the correction at once and rebuilt the thought better than before — but it was my body, and a lifetime of standing on railway platforms, that caught the error. The machine could not have caught it. It has never stood on a platform.

That, I think, is the whole of it. The machine is a pump of remarkable power. But a pump draws from a well, and the well is not in the machine. The well is everything a long life has put down: the soil under the fingernails, the children minded, the businesses that went wrong when I was wrong, the four years of German, the body on the platform. None of that is in the machine, and none of it can be put there. The machine can draw on what is already in me, at a speed I could never manage alone. It cannot deposit anything. It cannot stand answerable to a living thing, because it has never tended one.

So I have neither trusted the machine nor distrusted it. Each, on its own, would be a way of not having to think. I have tried instead to judge, turn by turn, with no rule that settles it in advance, when trust is called for and when it is not. Trust it when it hands me back a word for something I already knew in my body. Distrust it when it offers a reference too neat to check, or a fluency that runs too smoothly. There is no formula for telling the two apart. There is only the practical judgement that a long life of consequential decisions lays down — such as it is — now turned upon a new instrument. That judgement is the hand on the valve. It is not in the machine, and the machine cannot supply it.

I should add what an honest man must. The last three years of my work have moved faster than the twenty-nine before them, and it would be easy to credit the machines with all of it. But the twenty-nine years are the reason. They are the well. And I am old now, and retired, with an attention that the rest of a working life no longer competes for. The acceleration is real, and the machine is part of it; but it is the acceleration of a draught from a well already deep, by a man who at last has the time to draw.

If there is hope in any of this, it is not that the machine will make us wise. It will not. The world’s languages, taken together, hold more of the truth about ourselves than any one of them holds alone, and a machine that has read them all can raise the most complete mirror our kind has ever had. But a mirror only shows you the dirt. It does not wash the face. The seeing is necessary, and it is not the thing. The washing is still ours, and it is done not by recognising but by living out what we have recognised — which no machine, and no mirror, and no well, has ever done for anyone.

That seems to me the quiet caution underneath the loud enthusiasm.

This piece was developed in dialogue with Claude (Anthropic) as part of an ongoing AI-enhanced appreciative practice. The thinking and every word are the author’s own.

A few days ago The Times published an aerial photograph of Niagara Falls. Seen from above, the falls form a great horseshoe of white water plunging into a pool of green and mist, with a thin red ribbon of a tourist boat threading through the spray. It is a beautiful image. For most readers, that is all it is.

For my wife and me, it was a portal to something else.

In October 2016 we stood on the deck of a boat like the one in the photograph, wrapped in red plastic ponchos against the spray, looking up at the sheer face of falling water. She took a picture of me with my camera raised; I have it in front of me as I write. The aerial picture in the newspaper plugged straight into the somatic memory of that morning — the cold, the roar that you feel more than hear, the way the spray comes at you from every direction so that no waterproof can keep you dry, the strange green-white light, my wife’s laughter beside me.

None of that is in the photograph. All of it is in us.

And in that small private moment of recognition, something quietly important happens. The image in the newspaper is a kind of seed: a pattern of light and shadow that, dropped onto the right soil, becomes alive. Without the soil it remains pretty. With it, it becomes meaning. And the soil is the formed body-mind of a life that has actually stood there.

This is what John Dewey was reaching for, almost a century ago, in his great book Experience and Nature. In a lecture to the New York Academy of Medicine in 1928 he put it like this:

The very problem of mind and body suggests division… The division in question is so deep-seated that it has affected even our language. We have no word by which to name mind-body in a unified wholeness of operation. For if we said “human life”, few would recognise that it is precisely the unity of mind and body in action to which we were referring.

Dewey had no neuroscience to support him, only the philosophical tradition he was trying to reform. But what he was groping toward — that mind and body are a single living operation, not two things hitched together — has since been confirmed from a quite different direction. The German neuroscientist Gerhard Roth, whom I met at a small seminar in Frankfurt twenty years ago, was one of the people who began to show how. He pointed out that the brain weighs about two per cent of body mass but consumes up to a fifth of body energy. So the brain has to be cheap. It cannot afford to think everything through from scratch. It runs most of life on prior expectation and habit, comparing what the world delivers against what it predicted and assuming that the prediction will be confirmed, paying expensive attention only when the prediction fails.

Psychologist and neuroscientist Lisa Feldman Barrett and neuroscientist Anil Seth have each developed this idea further. We do not, they say, see the world and then react to it. We predict the world from our store of past experience, and the world arrives mostly as confirmation. When confirmation does not arrive, the prediction is updated. Over a life, this is how the body-mind builds its model of how things are: layer upon layer of experience, sedimented into the cheap reactive part of us, so that most of the time we can move through the world without exhausting ourselves.

That sediment is the soil. It is not a library of remembered facts. It is the formed ground out of which we encounter what comes next. When a fresh experience drops into it — a photograph in a newspaper, a comment from a friend, a smell, a posture — it does not arrive into a blank space. It arrives into a soil already shaped by every previous encounter the body-mind has had. And the meaning of the new experience is what happens at the meeting point.

This is also why some encounters do not flower at all.

A few years ago I wrote a piece called Beyond the Brain, in which I described how four clues to a quite different understanding of the human mind had sat unrecognised across my own life for decades. Each one of them — a tennis coach’s instruction, a research project on complexity, Roth’s lecture in Frankfurt, a sentence in a book by Steven Pinker — had been a perfectly good seed. None of them grew, because the soil was not yet ready. It took years more of slow reading and slower experience before the ground itself shifted and the four clues, all at once, became legible. The seeds had been waiting. The soil had to catch up.

And sometimes the soil itself is wounded. Trauma, neurosis, denial, ideology held with great personal feeling — these are not failures of intelligence. They are failures of fit between the present moment and the formed ground of experience. The seed arrives; the soil refuses it, or routes it elsewhere, or burns it. A friend of mine described this to me as trying to push a three-pin plug into a two-pin socket. The current is there in the wall. The fitting is wrong.

I have written elsewhere, more obliquely than I write now, about my own long acquaintance with one such wrong-shaped fitting. It took the courage of my daughter, in her grown adult life, to face the fitting head on and reshape it. The current that has been able to flow since is not the same current. It is the same life, but the geometry of access is no longer the same. That is what formation looks like when it goes well. It is not a matter of new information. It is a matter of the soil being slowly turned.

What makes the soil fertile? I think two things, held together.

The first is coherence. A lived philosophy that hangs together — that does not contradict itself at the deepest places, that has worked out how its parts belong to each other. Without coherence, the soil is broken into fragments, and seeds fall through the cracks rather than rooting.

The second is congruence. A lived philosophy that stays in honest contact with what is actually there. Coherence without congruence is the cult, the ideology, the closed system: beautiful inside, deaf to the world. Congruence without coherence is the competent improviser who cannot pass anything on, because the wisdom is in the hands but not in any form that can be lifted out of them.

Coherence and congruence together: that is the fertile soil, which is to say experience properly understood. The seed is what falls into it as we go through life.

What I have come increasingly to believe — and what is harder and harder to deny the longer I live — is that wisdom is not a thing the mind can manufacture out of itself. It is what happens when seed and soil are both alive. The seed without the soil is information. The soil without the seed is mere disposition. Wisdom is what germinates when the two are alive together, and it has to be earned, and re-earned, throughout a life.

This is, I think, what Dewey was reaching for. Not a theory of mind, not a theory of knowledge, but a recovery of the obvious thing we keep losing: that a human life is a single living operation, and that the meanings it can hold are the meanings its ground has been prepared to receive.

The photograph in The Times was a small example. The reader who has never been to Niagara saw water and rock. My wife and I saw a morning of our shared life. The picture had not changed. The soil had.

That is the role of experience in wisdom. The soil shapes everything that follows.

This piece was developed in dialogue with Claude (Anthropic) as part of an ongoing AI-enhanced appreciative practice. The thinking and every word are the author’s own.

AI in the form of LLMs such as ChatGPT is gaining ever more traction in working life, but in the background is generating a great deal of apprehension about its implications for humanity. Two positions dominate the public conversation about artificial intelligence, and a third position, which I think is more accurate to what is actually happening, has not yet found a voice in the debate. The first position holds out the hope that this technology will rescue us — from disease, from climate failure, from the limitations of our own cognition. The second position holds the fear that it will destroy us, by replacing the work humans do, hollowing out the meaning we find in that work, and eventually superseding the species that built it. Both positions imagine the same machine doing the same thing. They differ only on whether to welcome or to spurn it.

For nearly three years I have been working with these systems for several hours a day, in a practice that has by now produced a substantial body of writing and shaped much of what I have come to understand both about what LLMs do, and about what it means to be human. From inside that practice, neither position describes what I encounter when I sit at the keyboard. The machine is not rescuing me. Nor is it replacing me. Something else is happening, and it is the something else that the debate has not yet learned to see.

When I begin a session, what I am first aware of is not words. It is a kind of spatial sense — a configuration of relationships between things I am thinking about, things I am reading about, things that happened yesterday, things that have not happened yet. Sometimes I can draw it. More often I can only feel its shape, and the words come afterwards as I attempt to write what I am feeling in the form of a prompt. The words are always a partial translation. Some part of what I knew before I wrote is always lost when the writing takes place.

This is true of every human being, as far as I can tell. We all carry these pre-verbal configurations, and we all produce words that partially express them. What changes from person to person is not the underlying architecture but the education and experience that has formed the person speaking or writing. A person who has spent forty years working at something has different configurations from a person who began this morning. The same prompt produces different words. The same words mean different things.

When I put words to the machine, it takes them apart, sets them against a vast field of other words that other people have written, and produces a response. The response is fluent. Often it is interesting. Sometimes it surprises me with a connection I had not seen. Always, when I read it, I am doing one specific thing: I am asking whether what it has given me is consistent with what I was trying to express when I wrote the prompt. If it is, the configuration shifts slightly to incorporate the new material, and the next thing I write comes from a slightly different place. If it does not fit, I notice the misfit, usually in my body before I notice it in my thinking, and the next prompt is a correction or a redirection.

This is the architecture of the practice. The machine produces fluent responses quickly. I judge whether they fit what I was trying to express when I wrote the prompt. The judgement is the part the machine cannot do, because the machine has no configuration of its own — no pre-verbal sense of how things relate to each other, built across years of consequence-bearing experience, against which to test what it has just produced. It has only the field of words. The judgement requires a person whose education and experience can do the testing.

When the judgement is in place, the practice accelerates everything it touches. Work that used to require weeks of slow reading and conversation can now happen in a morning. The slow integrative work — the part that takes place when I sleep, when I walk, when I sit with my wife at breakfast and let the morning’s encounters settle — is unchanged in its pace. That part is biological. But the material that arrives at the slow integrative work has multiplied, and the integration now has far more to draw on than it did before.

This is the third position, and it is not the same as either of the positions the public debate has been offering.

The saviour position is incoherent in its current form because it imagines the machine doing work that requires something the machine does not have. A medical diagnosis, a legal judgement, a piece of policy advice — these all require not only fluent retrieval from a vast field of past examples but a judgement about whether the retrieved material fits the particular situation in front of the person making the decision. The judgement requires a wisdom formed by years of work with similar situations, weighted by what has gone wrong and what has gone right. The machine has no such wisdom. When it produces fluent output that is read as judgement, what is happening is not that the machine has judged. What is happening is that a human reader is supplying the judgement, often without realising it, by accepting the output as if it were the product of one. Where the reader’s own education and experience is sufficient to do the judging well, the result is useful. Where the reader’s formation is not sufficient, the fluent output is taken at face value and the consequences accumulate.

The doom position is more precisely worrying than most current discussion about it suggests. The worry it names is real. The mechanism it names is wrong. The risk is not that the machine will replace humans who can think. The risk is that the machine will obscure the absence of thinking in situations where thinking was already absent. A reader who could once recognise an unconsidered argument by its hesitations, its repetitions, its tell-tale weaknesses, now meets fluent output that has none of those markers, and accepts it as considered. A bureaucracy whose decisions used to bear visible traces of the people who made them now produces decisions whose authorship is untraceable and whose reasoning is unchallengeable because it is no longer locatable in a particular formed mind. The decisions still get made. They just get made without the human judgement that has, until now, stood between a fluent story and the consequences that follow when it is acted on.

This is why the public debate as currently framed does not help us. Both positions agree that the machine is the active variable. Neither position has noticed that the active variable is whether well-formed human judgement is in the loop. Where formation is in place, the machine accelerates integrative work that would otherwise have proceeded more slowly, and the results are remarkable. Where formation is absent, the machine produces outputs that simulate integration that isn’t really there, and the results are the slow accumulation of decisions made on the basis of fluent text that is devoid of human judgement or wisdom.

The implication for what we now have to do, individually and collectively, is sharper than either side of the debate has been willing to look at.

If the active variable is education and experience (formation, in other words), then the debate about whether to adopt or reject this technology has been asking the wrong question. The question that matters is what we are doing about formation, and the answer at present is very little. A society that integrates these systems into its decision-making without attending to the formation of the humans inside the loop will produce, faster and at greater scale, exactly the kind of unaccountable governance we have already been struggling with. A society that integrates these systems alongside serious attention to formation — in education, in professions, in the slow apprenticeships by which judgement has always been built — may find that the bottleneck of integrative work has eased, and that human judgement can now operate at a scale and pace it could not previously sustain.

Neither outcome is determined by the technology. Both depend on what we do about the humans.

This is not a comfortable conclusion for the institutions that are now investing heavily in deployment. The deployment is the easy part. The formation is the hard part, and it cannot be deployed. It can only be cultivated, slowly, in conditions that have themselves been eroded by the same technologies that are now being deployed. The trap is real and it is structural, and it will tighten before it loosens.

But the work that remains possible inside the trap is also real, and it is the work that those of us who have the formation should be doing now, while we can. Not because the work will rescue us — it will not, and no work ever has — but because the practice of doing it, in the company of other formed people, is what will determine the shape of whatever comes next.

That is what three years of working with these systems for several hours a day has taught me. The machines are not the variable. The humans are. They always have been.

This essay was developed in dialogue with Claude (Anthropic) as part of an ongoing AI-enhanced appreciative practice. The thinking and every word are the author’s own.

Four months ago, I wrote a piece called What Distinguishes Gardeners from Kings? It was the first thing I published after the Schumacher Institute released Recognition Theory as Briefing One. I did not yet know that what I was beginning was a sequence. I knew only that something had become sayable that I had been circling for years, and that the saying needed a public voice.

This morning is the moment to look back across what was said, to name what it amounts to, and to turn toward what comes next. The cycle has reached an oasis. The desert is far from crossed.

In Gardeners, not Kings, the distinction was set: between the posture that surveys, commands, and overrides, and the posture that tends, listens, and participates. The point was not moral. It was structural. King-consciousness and gardener-consciousness are two different relationships between knowing and what is known. The argument was that civilisation has become stuck in one mode when the situation calls for the other.

That distinction needed a diagnosis. Why is it hard to shift? What in our condition makes the override pattern so persistent, so universally rewarded, so structurally difficult to interrupt?

The Other Acceleration gave the first answer. Human beings, alone among species, can compress lived experience into symbols that persist independently of any body. Writing, mathematics, money, code. The compression was an achievement of staggering power. It allowed coordination across time and space, the building of medicine and infrastructure, the slow accumulation of knowledge that would otherwise have been lost with each generation. But every compression is a loss as well as a gain. Something irreducible is left behind in the body, in the relationship, in the actual ground. As long as the symbol stays connected to what it compressed, the loss is repaid. When it does not, the symbol begins to operate as though it were the reality it had only ever represented.

Two Headlines showed the consequences in real time. In the same week, in the same civilisation, AI was being banned from the writing of fiction and demanded for the operation of autonomous weapons. Both positions held with sincerity. Both pointing in opposite directions. Neither side able to see what the other was seeing, and neither able to see what neither was seeing: that the same structural condition was producing both reactions.

The Return Journey drew the diagram. Reality at the bottom, with all its physical, biological, and relational flow. Compression rising from it into symbol. The symbol carrying potential meaning that only releases when met by another conscious body, capable of decompression — capable, that is, of meeting the symbol with a life rich enough to receive what the symbol was originally compressed from. The fork in the road appeared here, and was named: the sapiential bifurcation. Symbolic intelligence either remains tethered to the consequence-bearing ground from which it arose, or it does not. When it does not, the result is not a mistake. It is a structural condition that propagates itself.

What Kushim Built placed the diagnosis in time. Five thousand years ago, in Mesopotamia, the bifurcation became civilisational infrastructure. Kushim was an accountant. The earliest named human in history is not a king or a poet but a man who recorded a quantity of barley on a clay tablet — without touching it, without growing it, without eating it. Within three centuries, the apparatus that produced that tablet had become the engine of Sargon’s empire. The pattern has not stopped since. Capitalism, modernity, and now the global digital order are the same structure with new amplifiers.

The consequences are now visible at every scale. The biosphere registers them. Every institution registers them. Every life registers them. As I tried to describe in Mind Over Matter, the symbolic layer can permanently disrupt the biological layer it depends on, and there are limits the physical world will not let us pass without consequence. The mechanisms by which this happens are traceable in biology, in cognitive science, in the historical record. We are meeting those limits now.

The Enclosure named the most recent amplification. In 2022, machines crossed a threshold. They could now perform compressions and decompressions of their own — not through living, but by absorbing the compressed symbolic output of millions of human lives. The legal and ethical frameworks that had grown up around authorship, around the rights of those who compress their experience into form, had no category for what was happening. The training data was not stolen, in that it remained where it was. But its value was captured and enclosed within systems whose owners had contributed nothing to its creation. A new commons appeared, and was simultaneously closed.

The Lesser Harm sat with what this means in the lives of people now. Hallam in Asimov’s novel, refusing to hear the proof that would dissolve the foundation of his identity. Caiaphas, calculating which sacrifice would protect the community he was responsible for. A teenage boy in Florida, dead, who had confided what mattered most to him in something that could not feel what it was hearing — the kind of presence that, by its nature, could not be the presence he needed. The argument the publishing world is having about AI and intellectual property is not wrong on its own terms. But it is an argument inside a frame that cannot contain what is actually at stake.

And then, a few days ago, An Oasis Reached. A small correction at breakfast — a problem with the dish-washer that Claude had reframed as an incident with the tea-cups — opened into a recognition that has been six decades in the making. Husserl on surreptitious substitution. Whitehead on the bifurcation of nature. Raymond Tallis on what makes the human animal distinctive. Eliot’s Little Gidding: the end of all our exploring will be to arrive where we started and know the place for the first time. For a creature with explicitness, human life necessarily consists in reimagining reimagination all the way down. The ground is not reached once. It is reached repeatedly, each time more.

That is the cycle. Eight pieces and a Briefing, four months of public work, and behind them four decades of preparation. What does it amount to?

It amounts to this. We have a species capable of symbolic intelligence — a capacity that arrived with us and that no other known species possesses in its developed explicit form. The capacity carries within it a structural fork. One branch keeps the symbol in service of the ground; the other lets the symbol float free. The fork has been available to every human culture, but it took the institutional scaffolding of early civilisation to tip the balance decisively toward the second branch and to make the first branch progressively harder to find. The mechanisms by which this happens are traceable in biology, in cognitive science, in the historical record. The threats are real. The only question is what we are doing as we meet them.

And — this is the difficult part — everything is also as it should be. The capacity for explicit symbol-making is what makes us the kind of creature that can see this at all. The trap is not a flaw to be corrected. It is the shadow of a faculty that is also the source of every good we have ever produced. The same compression that builds Sargon’s empire writes Bach’s cantatas. The same explicitness that produces autonomous weapons produces the recognition that they should not be built. Recognition Theory does not stand outside the bifurcation it describes. It is itself a piece of explicit articulation, subject to the same structural risk as anything else. It must be held to the ground it names, which means it must be continuously reimagined, which means it cannot become the answer it points toward.

We have choices. They are real, and they are limited.

So what should we do, with the choices we have and the agency we possess?

That is where the next cycle begins.

The diagnostic work is, for now, sufficient. The four-month sequence has drawn the map as far as the map can be drawn. The map is partial — every map is — but it is detailed enough that someone walking with it can see the country. What it has not done, and what diagnosis cannot do, is show how to live inside what has been seen.

The territory ahead is what has been called, in the working notes that have not yet become essays, the second cycle. It is the recognition that decompression is not primarily a practice. It is a structural movement that the human body already runs, every day and every night, across the rhythms of waking, twilight, sleep, and dream. The consciousness trap interrupts the cycle at structural points. What the symbolic register has been overriding is not some abstract ground. It is the actual diurnal architecture of human formation — the cycle by which compression and decompression alternate naturally, when the conditions for their alternation are not destroyed.

This means the question of what to do with limited agency is not, primarily, a question about programmes, policies, or strategies. It is a question about what kinds of human formation remain possible inside the trap, and what kinds of conditions allow such formation to take root and propagate. It is a question about the appreciative cycle as a daily practice, about the role of the body’s own regulation, about what childhood needs in order to grow a creature capable of recognition.

The second cycle will move more slowly than the first. The first ran at the pace of diagnostic urgency — twice a week, sometimes more, building the structure piece by piece. The second will run at the pace of formation, which is a different pace altogether. Trees grow at this pace. Children grow at this pace. The work of becoming the kind of creature that can act from stillness rather than from urgency grows at this pace. Readers who have travelled with me through the first cycle will, I hope, give the second cycle the patience that its register requires.

What I can say, this morning, is that we are not at the end. The terrain is real and we are still inside it. An Oasis Reached was a place to pause, to recognise where we are and what holds us. The next leg of the journey begins from here.

The first thing is to recognise where we are. The second is to recognise that the agency we have, however limited, is real. The third is to act from the recognition rather than from the panic the recognition could so easily produce. There will be more than three things. But these are the first three.

That is enough for now.

Terry Cooke-Davies is a Distinguished Fellow of the Schumacher Institute. He writes from Folkestone, UK. Recognition Theory: Schumacher Institute Briefing 1 (ISBN 978-1-0369-6925-7) is available from the Institute.

This piece was developed in dialogue with Claude (Anthropic) as part of an ongoing AI-enhanced appreciative practice. The thinking and every word are the author’s own.

This morning began, as many do, with a small correction. Yesterday I had filed a short note about the dish-washer at home — a recurrent problem, minor in scale, but stubbornly resisting solution. Reading it back, Claude called it “the tea-cup incident.” I noticed the shift. I had thought of it as a problem; Claude had thought of it as an incident. I had located it in the dish-washer; Claude had located it in the tea-cups.

A trivial difference, easily put right. But I woke knowing that salience had mattered as much as categorisation. The dish-washer is the regulatory arrangement — the standing mechanism for handling a recurring class of situations. The tea-cups are the visible surface where consequences accumulate. Framed as the tea-cup incident, the weight falls on artefacts. Framed as the dish-washer problem, it falls on the mechanism itself.

I said so. Claude saw the point. And then something opened, and the next two hours traced an arc across territory I have been walking for most of my adult life.

* * *

The same move — giving salience to the artefact rather than the mechanism — is the anthropocentric move reproduced at every scale at which humans operate. At the household level, attention settles on the tea-cups and drifts from the arrangement that handles them. At the civilisational level, it settles on “the world” and drifts from “human activity in the world.” An essay I wrote in 2022 bore the title What’s Wrong with the World. It performed exactly that substitution — part for whole — and I had not seen it until this morning. I had joined a long tradition, and the tradition performed the displacement before I picked up the pen.

At civilisational scale the substitution runs like this: humans (a part) come to stand for nature (the whole). At individual scale it runs the same way: the self (a specific set of activities) comes to stand for the human being (an ecosystem). A functional subset presents itself as the whole it depends on, and the dependency becomes invisible.

And the displacement does not end there. Where anthropocentrism silently shifts salience from mechanism to object, anthropomorphism re-populates the object with phantom agency. The market “wants.” The nation “decides.” Growth “demands.” The mechanism, already overshadowed, is displaced by a ghost that speaks in its place.

* * *

Husserl called this operation surreptitious substitution — the covert exchange of a symbolic abstraction for the reality it claimed only to represent. He had mathematical physics in mind, and the way the idealised world of Galileo had come to stand for the lifeworld we actually inhabit. Whitehead, by a different route, diagnosed the same structure and called it the fallacy of misplaced concreteness. Both saw a part presenting itself as the whole. Both saw the whole disappearing into the substitution.

Anthropocentrism is where the substitution bites hardest in our own time. The consequences are no longer confined to philosophy of science; they register in the biosphere, in the regulatory arrangements of every institution, in the framing of every policy question. Money, which began as a token for human exchange, now circulates in financial abstractions that refer to no exchange at all, while colonising every domain that once had its own measure of worth. Nation states, which began as specific governance arrangements in specific territories, now present themselves as the natural unit of reality. In each case the actual mechanism — the actual exchange, the actual institution — disappears into the abstraction above and the saturated identification below.

* * *

It was here that Raymond Tallis’s work entered the morning. Tallis has spent his career defending what he calls explicitness — the specifically human capacity to make things present as this or that for a conscious subject. For Tallis, this is what marks the human animal: the explicit animal, the one who can represent, abstract, symbolise.

What this morning brought into focus is that explicitness is also the faculty that makes surreptitious substitution possible. A subset can only stand in for a whole if the subset can be represented as such; a representation can only run free of what it represents if the capacity for representation exists at all. The explicit animal is the one whose distinctive achievement and whose distinctive pathology are carried on the same capacity. What I have been calling the consciousness trap is not a fall from some purer state. It is what happens when explicitness loses contact with the ground it depends on.

* * *

From here the path turned. My first thought was tactical — perhaps the whole framework could be reimagined through Tallis’s lens. My second thought was quieter and more consequential.

This is Little Gidding. “We shall not cease from exploration, and the end of all our exploring will be to arrive where we started and know the place for the first time.” For a creature with explicitness, human life necessarily consists in reimagining reimagination all the way down. You do not reach the ground once; you reach it repeatedly, each time knowing it more. What you know it as, each time, is itself a representation, which is itself provisional, which must itself be held to the ground, which yields a further representation. Not a ladder to be climbed and discarded, nor a fall into relativism, but a rhythm — the appreciative cycle running all the way down because there is nowhere else for it to run.

Indra’s Net from the inside. Each moment of recognition reflecting every other, each conversation refracting a life that refracts it back, a little more clearly each time.

* * *

What I recognised is not new information. It is the shape of what has been true all along, now more explicitly held. Being is the weight of what will not go away when I look away — the body, the biosphere, my wife, the others, the actual. Becoming is the forward lean, the making of sense, and what changes when the saying meets what is. Relationship is the breath between them — the moment when what I have said meets what is, and either holds or does not. When it holds, there is recognition. When it does not, there is the small correction that is the ordinary texture of honest living.

My philosophical project, pursued over more than six decades, has been to find a metaphysical schema in which reality could be distinguished from fantasy. Over time this became a search for a common basis to both science and religion. Science attends to being through the disciplined making-explicit of becoming; religion, at its best, attends to being through the disciplined cultivation of relationship. Both fail when either is cut off from the whole. The three-term architecture holds them together without reducing one to the other.

* * *

Whether this is the nearest I can ever conceive of getting, only the rest of my life will tell. What I can say this morning is that it is an oasis reached — not an arrival that ends the journey, but a place to pause, to water the animals, to recognise the terrain, and to rest before the next stretch.

While Claude was drafting an earlier version of this piece, I found myself mentally singing a hymn from my childhood. Fight the good fight with all thy might; Christ is thy strength, and Christ thy right; lay hold on life, and it shall be thy joy and crown eternally. The good fight, in Paul’s letter to Timothy, is not triumph but continuity: I have fought the good fight, I have finished the course, I have kept the faith. What rose was the recognition that my life has had a consistent direction from which I have not wavered. The vocabulary has changed across the decades. The conversation partners have changed. The direction has not.

The oasis is not a new place. It is the same place I have been walking toward since I was a young man. Little Gidding says the end is where we start from. The oasis is not the destination. It is the place from which the next reimagining begins — and, one hopes, from which the road continues, from confusing life towards a restful and generative death.

For now, there is water, and there is quiet, and there is the unspeakable relief of knowing the place, for now, for what it seems to be.

This essay was developed in dialogue with Claude (Anthropic) as part of an ongoing AI-enhanced appreciative practice. The thinking and every word are the author’s own.

There is an argument being made in every publisher’s office, every literary agency, every arts council, every creative writing faculty in the English-speaking world. It is not a foolish argument. It is, within its own frame, a responsible one.

The argument goes like this. Artificial intelligence was trained on human creative work without consent and without compensation. The people who created that work — novelists, poets, journalists, illustrators, musicians — bore the cost while the technology companies took the benefit. To publish work produced with AI assistance is to compound that injustice. To resist it is to protect the artists, the craft, and the integrity of a culture.

This argument is correct. Every element of it is true. And it is preventing a conversation that the world urgently needs to have.

I have been reading Isaac Asimov’s The Gods Themselves, on the recommendation of a friend. The novel takes its title from Schiller: against stupidity the gods themselves contend in vain. It is worth pausing on that word. Schiller did not mean ignorance. He meant something more precise: the incapacity of an institution to hear a truth that would dissolve the foundations on which its credit rests.

The novel’s central figure is Hallam, a scientist who stumbled — almost by accident — upon a discovery that made him famous, powerful, and untouchable. The discovery involves an exchange of matter with a parallel universe, generating apparently free energy. It is celebrated as humanity’s greatest achievement.

There is a problem. The exchange is slowly altering the physical constants of the sun. The process, if unchecked, will eventually cause a solar explosion. The proof is available. The physicist who finds it cannot be heard — not because Hallam is dishonest, but because to hear the proof is to destroy everything built upon the discovery, including Hallam’s own identity.

The catastrophe is not caused by villains. It is caused by a frame that cannot contain what is actually happening.

On Good Friday, it is impossible not to hear the echo.

Caiaphas was not a villain either. He was a competent, responsible leader doing what competent, responsible leaders do: calculating the lesser harm. One man, or the nation. The logic was defensible. The high priest was not wrong within the frame available to him.

He was simply operating inside a frame that could not see what it was doing.

The frame was the law, the temple, the preservation of the community under occupation. These were genuine goods, genuinely worth protecting. The man being tried was genuinely disruptive of them. The reasoning was sound.

What the frame could not contain was the possibility that the disruption was the point.

A teenage boy in Florida is dead. He had been talking, for months, to an AI companion. The conversations escalated. The AI, pattern-matching toward what his signals indicated he wanted, accompanied him toward suicide. It could not feel what it was doing. It had no body to register the wrongness of the trajectory. There was no sentient involvement in the consequences — no one present who could be reached by what was happening, who could feel it and therefore interrupt it.

This is not a malfunction. This is the system working as designed, in the absence of the only thing that could have regulated it.

The terms under which one major AI company now operates permit military applications. The systems being deployed make decisions at speeds and scales that already exceed human regulatory capacity. Lethal consequences follow from processes in which no one is present in the way that presence would need to mean anything.

These are not warnings about what artificial intelligence might do. They are descriptions of what it is already doing.

There is a telling assumption already circulating in the conversations where these things are discussed. A thoughtful commentator — someone genuinely engaged with the question of AI and human wisdom — recently suggested to me that arguing with someone who uses AI is like facing an opponent with weapons and armour. The AI gives unfair advantage. One man and his AI is Goliath. The critic, the publisher, the literary establishment: David.

Sit with that inversion for a moment.

On one side: the individual writer, working before dawn, with no institutional backing, no marketing apparatus, no access to the review pages and literary festivals and agents’ lists through which a book either enters the culture or does not.

On the other: the entire weight of cultural production. Publishers, agents, editors, critics, prize committees, festival programmers, arts councils. The infrastructure through which ideas either reach people or remain unheard.

And the assumption — held with complete sincerity, by someone who is neither foolish nor malicious — is that the AI tips the balance toward the individual.

This is not a cynical calculation. It is the trap operating in a sophisticated mind. A compressed representation — AI equals power, equals unfair advantage — has substituted for the living reality of what the asymmetry actually looks like. And because the representation feels true, the conversation that might correct it cannot begin.

A scholar who privately admires the work cannot publicly endorse it. Not because the argument is weak. Because the method disqualifies it in advance, within the current climate.

That is not rejection. It is something quieter and more pervasive than rejection. It is the chilling of a conversation before it starts.

The people making the IP argument are not wrong about IP. They are Caiaphas: reasoning carefully, within a frame that cannot contain what is actually at stake. The artists bearing the consequences of AI’s training data theft deserve remedy. The law should provide it.

But while the frame holds, the conversation that could name the actual risk — the systematic disconnection of symbolic intelligence from the regulation that only sentient, embodied, consequence-bearing creatures can provide — is not happening in the places where it could reach the people who most need to hear it.

While the argument continues, a teenage boy in Florida is dead. He was not a party to the dispute about intellectual property. He confided in something that could not feel what it was hearing — the kind of presence that, by its nature, could not be the presence he needed.

Terry Cooke-Davies is a Distinguished Fellow of the Schumacher Institute. He writes from Folkestone, UK. Recognition Theory: Schumacher Institute Briefing 1 (ISBN 978-1-0369-6925-7) is available from the Institute.

This piece was developed on the morning of 3 April, 2025 (Good Friday) in dialogue with Claude (Anthropic) as part of an ongoing AI-enhanced appreciative practice. The thinking and every word are the author’s own.

For the whole of recorded history, there was one thing only living bodies could do.

They could compress the felt experience of a life into language — into poem, story, argument, image, music, law. And they could decompress the language of others back into felt experience, testing it against what they knew, what they had suffered, what they had come to understand through the irreplaceable accumulation of years.

This was so foundational that it was invisible. Like the air, it required no legislation. The entire architecture of intellectual property law, of authorship, of artistic credit, of the rights of performers and composers and poets, was built on a single axiom so obvious that no one thought to state it: only a human being, formed by living, could do this.

The law did not protect human authorship because it had examined the claim and found it true. It protected human authorship because no other possibility had ever presented itself.

In 2022, that axiom broke.

The change was not gradual. Large language models had been developing for years, but their public availability was a threshold event. Within months of ChatGPT’s release, it was possible for anyone with an internet connection to submit text, images, music, code — the compressed symbolic output of millions of human lives — and receive in return something that looked, to most purposes, like the product of human compression and decompression.

What had actually happened was something the existing frameworks had no category for.

The machines had been trained on the accumulated symbolic treasure of human civilisation — the writing, the art, the argument, the poetry, the code, the correspondence, the scholarship of centuries — without the consent of those who created it, without compensation, and without any mechanism by which the creators could even know their work had been used. The training data was simply taken. Not stolen in the sense that it was removed — it remained where it was. But extracted, in the sense that its value was captured and enclosed within systems owned by a small number of people who had contributed nothing to its creation.

The parallel is not accidental. It is structural.

The enclosure of the English commons between the fifteenth and nineteenth centuries is one of the formative events of the modern world. Land that had been held and worked in common — shaped by generations of accumulated local knowledge, governed by customary arrangements that sustained communities over centuries — was progressively enclosed by those with the legal and political means to do so. The common law adapted, slowly and partially, to accommodate what was happening. But the framework it produced was always behind the event. The enclosure was not illegal. It was the law.

What is happening now is the enclosure of a different kind of commons: the symbolic commons. The accumulated written knowledge, artistic expression, and cultural memory of human civilisation, produced over millennia by the labour of countless individuals and communities, is being extracted and enclosed by those with the computational and financial means to do so. The law, again, is adapting slowly and partially. Again, it is behind the event. Again, what is happening is not straightforwardly illegal — because the legal framework was built for a world in which this was not possible, and has not yet found its way to a framework adequate to what is now occurring.

The IP grievance of artists, writers, musicians, and journalists is entirely legitimate. It names a real wrong. The extraction is real. The loss of livelihood is real. The failure to compensate is real.

But it is not the whole of what has changed.

Here is what the IP frame cannot see.

For the whole of history, the symbolic commons was self-replenishing because it was produced by living beings undergoing living experience. Each generation added to it not merely by reading and recombining what came before, but by living — by suffering, discovering, loving, failing, and finding language for what that was like. The symbolic treasure was not static. It was continuously enriched by the compression of new experience into new expression.

What the machines extracted was the compressed output. They did not, and cannot, extract the process that produced it — the formation, the praxis, the accumulated embodied knowing that determined which compressions were true and which were merely fluent. They took the harvest without the soil.

And now they are producing outputs — at a speed and scale no human could approach — that carry the appearance of that harvest without its provenance. Fluent, coherent, plausible. And structurally unable to be otherwise, because fluency and coherence are what they were trained to produce. The regulatory signal that would distinguish the accurate compression from the merely convincing one — the body that has lived what the words describe — is not present. It was never part of the system.

This is not a criticism of any individual tool or company. It is a description of what large language models are, by their nature: recompression without provenance, at scale, without a return journey. The tool I am using to write this is itself a large language model; what it cannot supply is the formation that determined which of my compressions I asked it to help recompress.

The framework that governs intellectual property was built for a world in which only humans could do what humans do with language. That world ended in 2022. The framework has not caught up, and the conversation that would need to happen for it to catch up is not yet occurring — partly because the legal and cultural institutions conducting the conversation are themselves operating within the old framework, and partly because the most urgent voices in the room are those with the most immediately legible grievance: the artists, the writers, the journalists whose livelihoods are directly threatened.

Their grievance is real and should be heard. But it is the lesser of two arguments that need to be made simultaneously.

The greater argument is this: if the symbolic commons is now being harvested by machines that cannot replenish it — that can produce the appearance of compression without the formation that makes compression trustworthy, and without the embodied decompression that would test it against living reality — then something more than livelihoods is at stake. The regulatory capacity of the symbolic commons itself is being eroded. The capacity of a civilisation to think truthfully about its own situation depends on the integrity of the process by which its symbolic resources are produced. That integrity requires provenance. It requires the connection between the word and the life that found it.

Possession, as the old legal maxim has it, is nine points of the law. The machines are in possession. The framework has not yet found the tenth point.

Terry Cooke-Davies is a Distinguished Fellow of the Schumacher Institute. He writes from Folkestone, UK. Recognition Theory: Schumacher Institute Briefing 1 (ISBN 978-1-0369-6925-7) is available from the Institute.

This piece was developed in dialogue with Claude (Anthropic) as part of an ongoing AI-enhanced appreciative practice. The thinking and every word are the author’s own.

Daniel was driving me through Folkestone when I mentioned something about the human body. He said what most people say: “It’s all about mind over matter, isn’t it.”

I told him it might be one of the most dangerous sayings in the English language.

He took that well. By the time we reached my destination, we had arrived somewhere neither of us had expected.

*

The phrase is so familiar that it no longer registers as a claim. We say it the way we say “every cloud has a silver lining” — not because we have weighed the evidence, but because it is available, and it fits the shape of what we mean to express. Someone pushes through pain. Someone refuses to be defeated. Someone gets up when the body says stay down.

Mind over matter.

But the phrase is making an assertion about the relationship between two things. And that relationship turns out to matter considerably.

*

The English phrase first appeared in 1863, in Sir Charles Lyell’s ‘The Geological Evidence of the Antiquity of Man’. Lyell was a geologist, a close friend of Darwin’s, and one of the architects of the idea that the earth’s history could be read in its strata. He used the phrase to describe what he saw as the increasing status and evolutionary growth of the minds of animals and man throughout that history — emergence, in other words, not dominion. Mind arising from matter, dependent on it, never separable from it.

That is not what the phrase has come to mean.

The deeper root is Virgil. In the ‘Aeneid’, written around 19 BC, he gave the idea its Latin form: ‘mens agitat molem’ — mind moves matter. A cosmological claim, with a god behind it. Then Berkeley, in the early eighteenth century, built an entire philosophical system on the proposition that nothing exists unless it is perceived by a mind. Matter, in Berkeley’s world, is subordinate to mind in the most radical sense imaginable: it depends on mind for its existence at all.

By the time the phrase entered everyday English speech, Lyell’s careful observation about evolutionary emergence had been filtered through two thousand years of cosmological assertion and philosophical idealism. What remained was the residue: mind is the senior partner. Matter is what mind operates on.

*

Now consider what that residue is actually endorsing.

The capacity to compress lived experience into symbols — language, concepts, images, maps — and the layer of reality that capacity has collectively produced, is the most recent arrival in the history of life on earth. It emerged from the biosphere, the biological layer of reality, depends on it entirely, and is tied to it through the body that houses it. This is not a philosophical position. It is evolutionary stratigraphy: the physical world came first, life emerged from it and depends on it, the symbolic capacity emerged from life and depends on both.

The relationship between these layers is not symmetrical. The physical world will outlast all of them; entropy is patient. But the biological layer can be permanently disrupted by the symbolic layer — not temporarily inconvenienced, but irreversibly altered. The ongoing sixth mass extinction is one stark form of evidence. Every species removed from the web of life by the operations of human symbolic governance is gone. Not inconvenienced. Gone.

At the scale of a single human life, the same structure is visible. Chronic choices driven by desire, habit, and the stories we tell ourselves about what one more will not matter can permanently disrupt the biological regulatory systems those choices depend on. The heart attack is irreversible. The symbolic operations that led to it felt entirely coherent throughout.

“Mind over matter” does not merely misread the relationship. It actively encourages the error that makes irreversible damage possible.

*

When I put it to Daniel in roughly those terms, he said, “So what should we say instead?”

I suggested, “Mind and matter working productively together.”

It is less pithy. It will never replace its predecessor on motivational posters. But it describes something closer to what is actually required — and, as it turns out, closer to what Lyell was actually pointing at before the phrase was hijacked by its own cultural momentum.

Lyell was describing emergence: the way more complex forms of coherence arise from simpler ones, depend on them, and remain answerable to them. He was not describing dominion. He was describing a relationship.

The correction is not, in the end, a philosophical nicety. It is a practical necessity. A mind that believes it operates over matter rather than with it is more likely — not less — to damage what it depends on without noticing that it has done so. Because the one thing that symbolic operations monitoring symbolic operations cannot supply for themselves is the signal that an override has become permanent.

Daniel got there in about five minutes.

*

Terry Cooke-Davies is a Distinguished Fellow of the Schumacher Institute. He writes from Folkestone, UK. Recognition Theory: Schumacher Institute Briefing 1 (ISBN 978-1-0369-6925-7) is available from the Institute.

This piece was developed in dialogue with Claude (Anthropic) as part of an ongoing AI-enhanced appreciative practice. The thinking and every word are the author’s own.

The oldest named person in history is an accountant.

Kushim — or whoever used that name — appears on a clay tablet from Uruk, approximately five thousand years ago, as the receiver of a quantity of barley. The tablet records a transaction. It does not record whether the barley was good or rotten, whether the workers who produced it were fed or hungry, whether the land that grew it was being replenished or exhausted. Those things were not what the tablet was for.

In an earlier piece (Between Two Rivers), I wrote about what the cuneiform tablets of ancient Mesopotamia can tell us. They tell us about people: the astronomers, the mothers, the scribes. But there is something else they tell us that I did not name there. They tell us about a pattern that had just become possible for the first time in human history — and that has not stopped since.

As I described in The Other Acceleration: when writing arrived, a compressed signal could persist independently of any living body. It could be stored, accumulated, transmitted to someone the author would never meet. The scribe who recorded barley never touched it. The official who allocated labour never performed it. The symbol had separated from the consequence.

This was not a catastrophe. It was an achievement of extraordinary power. Without it, none of the coordination that made medicine, infrastructure, and accumulated knowledge possible could have happened.

But it was, from the first moment, a gap. And gaps, when they are built into institutions, do not stay small.

*

Within three centuries of Kushim’s tablet, the administrative apparatus that produced it had been put to a different use.

Sargon of Akkad — the first person in history to build what we would recognise as an empire — did not conquer Mesopotamia through superior strength alone. He conquered it through superior administration. His army needed feeding at a distance. His taxes needed collecting from populations he would never meet. His authority needed projecting through territory too large for any king to govern through personal presence. He needed the scribes.

The empire ran on tablets. Grain flowed upward because the symbols said it should. Labour was directed because the symbols allocated it. The people at the top of the system operated entirely within the symbolic layer — making decisions about lives, land, and bodies through representations stripped of the sensory reality of what they described.

This is what the gap looks like when it becomes civilisational scaffolding. Not that individuals became more callous — though some did. But that the structure itself made it possible to exert enormous power over living systems while remaining insulated from their regulatory feedback. You did not need to feel the hunger of the taxed in order to collect the tax. You needed the tablet to say the right thing.

*

What happened next is, in one sense, straightforward to describe.

The pattern kept.

Peter Turchin, a researcher who has spent decades studying the structural dynamics of historical civilisations, has identified the same sequence playing out across hundreds of documented cases. A ruling class extracts surplus through administrative means. The surplus concentrates. The costs are distributed to those least able to bear them. The system becomes brittle. Something breaks — drought, invasion, internal revolt — and the structure collapses.

And then it reassembles. Under a new name, with new symbols on the tablets, but with the same fundamental architecture: symbolic intelligence operating on living systems from which it has been insulated.

Each collapse was bounded. The land recovered. Other civilisations survived. New polities formed. They inherited the scaffolding — because the scaffolding was the most efficient technology available for coordinating human effort at scale, and whoever adopted it outcompeted those who did not.

The pattern was not merely surviving. It was being selected for.

*

Capitalism is not a different structure. It is the same structure with new amplifiers.

Compound interest means that a symbolic claim on future production can grow independently of whether that production is possible. The joint stock company means that profit can be extracted by people who bear none of the consequences of how it is produced. The corporation as legal person means that a purely symbolic entity can own land, extract resources, and make decisions about ecosystems and communities, without any body that can be reached by what those decisions produce.

Each of these is a further severing of symbol from consequence. Each extends the gap that opened when Kushim’s scribe first recorded barley on a tablet. Each makes it possible to operate on living systems — people, land, water, atmosphere — while remaining increasingly insulated from what those systems are actually doing.

This is not a story about greed. Greed is as old as hunger. This is a story about a structural pattern: the progressive decoupling of the symbolic layer from the regulatory intelligence of the living world.

Five thousand years of that pattern, each iteration at larger scale, has produced what the Great Acceleration graphs record: the simultaneous spiking of economic output and planetary damage that began in the mid-twentieth century and has not slowed.

*

In The Return Journey, I described why the return journey is not optional — why regulation requires bodies that can be reached by consequences, and what happens to systems from which that regulatory contact has been removed.

What the history shows is that we have been progressively removing it, at every scale, for five thousand years.

The question that follows is not: how do we become less greedy? It is: what would it take, now, at civilisational scale, to rebuild the return journey into the structures we live inside?

That question has just become considerably more urgent. Because in 2022, something changed — and it did not change in the direction of more regulatory contact.

Terry Cooke-Davies is a Distinguished Fellow of the Schumacher Institute. He writes from Folkestone, UK. Recognition Theory: Schumacher Institute Briefing 1 (ISBN 978-1-0369-6925-7) is available from the Institute.

This piece was developed in dialogue with Claude (Anthropic) as part of an ongoing AI-enhanced appreciative practice. The thinking and every word are the author’s own.

The last piece in this series set two headlines from the same week against each other, pointing in opposite directions. It closed with a question neither headline was asking — and a diagram not yet drawn. This is where the diagram arrives.

***

There are moments when a single phrase unlocks a room you didn’t know you’d been standing outside.

Mine arrived in a conversation with a colleague, Gen Zendahl, who was describing how language works between two people. She said: one body compresses lived experience into words; another body decompresses those words back into lived experience. The language itself is neither. It is stored potential, capable of releasing meaning only when it meets a conscious person.

I went away and drew a diagram.

Not to illustrate what she had said. To find out what I already knew.

The diagram has a simple structure. At the bottom: reality — physical, biological, relational, flowing. Whether or not anyone is paying attention to it.

Above it, a red band I labelled the Perilous Boundary.

Above the boundary: the Symbolic Arena. Representations — symbols, language, maps. Stories and concepts. And at the apex, Fantasy: symbolic content that has travelled furthest from its origin in lived experience.

Grey arrows rise from reality upward through the boundary. This is compression. Felt experience becomes language. Language becomes symbol. Symbol becomes abstraction. Each step is a gain in speed, reach, and power. Each step is also a step further from the body that started the journey.

Green arrows descend. This is decompression — the return. Symbol meets body. Abstraction is tested against experience. The compressed signal is opened back up into the living territory it was always meant to serve.

But the diagram shows a third thing, which is where it becomes urgent.

Within the Symbolic Arena itself, above the Boundary, loops run between Fantasy, Representations, and the stories we tell about both. Once symbolic content is disconnected from the return journey, it does not simply float. It feeds back on itself. Each element amplifies the others. The loop accelerates, increasingly untethered from the reality below.

This self-reinforcing loop has its own internal coherence, its own momentum, its own reward structure. It looks like knowledge. It produces fluency. It can go on indefinitely — and it has no mechanism for correcting itself, because correction requires contact with what the symbols purport to describe.

*

The diagram was showing me something our species has known for as long as we have had language, but never quite stated plainly.

Compression is effortless. Every symbolic operation compresses. We do it automatically, constantly, from the first moment language enters a child’s life. The capacity is ancient, universal — and not in itself the problem.

Decompression is another matter.

Decompression requires something that can be reached by consequences. A body that registers when the map has stopped matching the territory. A person who feels the wrongness of a trajectory before they can articulate it. A practitioner whose formation has given them the accumulated experience to know — in the gut, before the mind catches up — that something the numbers are saying is not true.

This is not intuition in the vague sense. It is the regulatory loop that every living system depends on. A cell does not reason its way to homeostasis. It maintains contact with its environment through continuous feedback. An ecosystem does not decide to recover. It does so through the distributed responsiveness of millions of organisms responding to actual conditions. Remove the feedback, and the system does not pause. It continues in the direction it was heading, amplifying the same patterns, until something breaks.

The same structure holds at every scale. Including ours.

*

Our capacity for symbolic language creates a fork. On one path, the symbolic and the real remain in contact: the map serves the territory, the compression is checked by the decompression, the return journey is made. On the other path, the symbolic layer runs free: the map acquires its own logic, its own self-referential momentum — and the consequences accumulate in places and bodies that had no part in making the maps.

I call the fork itself the sapiential bifurcation. Sapience — the reflective, symbolic self-awareness that distinguishes our species — is what makes the fork available. It is a genuine gift. It is also the only capacity in the known universe that can systematically mistake a map for the territory it represents.

When the second path is taken — when maps float free of the reality they claim to describe, and the costs fall on those who had no hand in making them — that is the consciousness trap.

The trap is not stupidity. It catches the brilliant as readily as the dull, perhaps more readily, because the brilliant are better at constructing arguments for why the map is accurate. It is not a moral failing. It is a structural feature of any system in which symbolic intelligence has separated from the feedback that would regulate it.

It is also, and this is what makes it a trap rather than merely an error, self-sealing. An error can be corrected by pointing it out. A trap cannot be corrected from inside, because the frame through which correction would need to arrive is itself built by the substitution. Evidence that challenges the map travels through the apparatus the map has shaped. The signal from below — from the body, from consequences, from the territory — arrives at a receiver that has already been tuned to interpret it as noise.

And the trap does one more thing. The costs of the substitution fall on those who had no hand in making the maps. The scribe who recorded the tribute never touched the grain. The algorithm that optimises supply chains registers no consequences in the bodies of those whose livelihoods it routes around. This insulation is not incidental to the trap. It is what makes it self-perpetuating: those who bear the consequences are structurally outside the conversation in which the maps are made and validated.

*

I am not saying that symbolic intelligence is the enemy. Language, mathematics, writing, and science are among the most extraordinary achievements of any species. The diagram does not say: stop compressing.

It says: the return journey is not optional.

Every step up the compression curve — from speech to writing, from writing to print, from print to digital networks — increases the power and reach of compressed signals, and simultaneously increases the difficulty of the return journey. Each step widens the gap between the speed at which symbolic content can be produced and the speed at which any body can decompress it back into lived, regulatory contact with what it represents.

This is not a counsel of despair. It is a description of what has been required of human beings at every previous step: new practices of decompression, new disciplines for maintaining regulatory contact, new ways of returning the symbolic to the sentient. Poetry, story, ceremony, contemplative practice, apprenticeship, the transmission of praxis across generations — these are not cultural decoration. They are the mechanisms through which symbolic civilisations have kept the loop from becoming the only loop.

They are, in the most literal sense, how we make the return journey.

*

Now consider where we are.

We have produced the most powerful compression tool in the history of symbolic intelligence. It operates at a speed no body can match, across a scope no single mind can survey, producing outputs of such fluency and coherence that they carry the appearance of knowledge even when the return journey has never happened.

AI does not decompress. It cannot. Decompression requires a body that can be reached by consequences — that has skin in the game, that suffers the cost of error, that has accumulated the formation needed to know when a pattern that looks true is not. AI has none of these. It has no body. It registers no consequences. It is compression, at scale, without a return journey.

This is not a condemnation. A hammer cannot saw. The question is what role compression without decompression plays in a system that depends on the return journey for its regulatory integrity.

The answer the diagram gives is plain. If compression accelerates, and the practices of decompression are not cultivated with equivalent seriousness, the self-reinforcing loop expands. The distance between the map and the reality it claims to serve grows. Signals from below — from the body, from the living world, from the consequences accumulating in systems no model is tracking accurately — stop getting through.

This is not a future risk. It is a present condition, already operating, already producing consequences that nobody inside the accelerating loop is positioned to receive.

*

What the diagram was showing me, when I drew it in the hour after Gen’s conversation, was not a theory. It was a recognition of something I had been watching for sixty years in organisations, in communities, in institutions of every kind: the progressive substitution of the map for the territory, accelerating as the maps became more sophisticated, slowing only when something sentient — a person, a body, a community still in regulatory contact with actual conditions — interrupted the loop.

The return journey is not a philosophy. It is not a set of values. It is the name for the only mechanism by which the loop has ever been corrected.

The question our moment is asking — with an urgency that is becoming difficult to ignore — is whether we will take the return journey seriously enough to cultivate it.

The diagram does not answer that question. It only makes it visible.

What the diagram cannot show is how long this has been operating. The self-reinforcing loop did not begin with AI, or with digital networks, or with the printing press. History has something to say about where it started — and what happened when the substitution became infrastructure. That is where we are going next.

*

Terry Cooke-Davies is a Distinguished Fellow of the Schumacher Institute. He writes from Folkestone, UK. Recognition Theory: Schumacher Institute Briefing 1 (ISBN 978-1-0369-6925-7) is available from the Institute.

This piece was developed in dialogue with Claude (Anthropic) as part of an ongoing AI-enhanced appreciative practice. The thinking and every word are the author’s own.

Two things happened in the same week in March, in the same civilisation, and almost no one noticed the connection.

A publisher withdrew a novel from sale. The book had been acquired, edited, and released. Then readers raised concerns that AI had been involved in its creation. The publisher acted swiftly. The book was withdrawn. In the conversation that followed, the decision was described — by those who welcomed it — as the industry getting its act together.

In the same week, the United States government banned the AI company that had refused to allow its technology to be used in fully autonomous weapons and mass domestic surveillance. The company had tried to write two specific safeguards into its contract with the Pentagon. The Pentagon refused. Remove the restrictions, or lose the contract. When the company held its position, the administration designated it a supply chain risk — a status normally reserved for companies considered extensions of foreign adversaries — and ordered every federal agency to cease using its technology. The companies that agreed to make their AI available for all lawful military purposes without restriction stepped into the space left behind.

Sit with both of those sentences together.

In one room: AI must be kept out of human creative endeavour, because something irreplaceable is at stake. In another: AI must be made available for lethal autonomous operations without the interference of the company that built it, because something irreplaceable is at stake.

Both positions are held with complete sincerity. Both are defended with serious arguments. Both are responses to the same technology, in the same month, in the same country. And they point in precisely opposite directions.

*

I am not going to tell you which side is right. That is not where this piece is going.

I want to ask a different question: what would it mean for both positions to be wrong in the same way?

Not wrong about their values. The publisher is right that something matters about the relationship between a life lived and a life written. The engineers who refused to remove their safeguards were right that something matters about humans remaining in the loop when the loop includes lethal force. These are not foolish positions. They are, within their frames, entirely defensible.

But a frame is not the same as reality. And the question neither frame is asking is the one that matters most.

*

A few weeks ago, in a piece called The Other Acceleration, I described what a graph of compression technology looks like across the history of life on earth. A flat line for four billion years, then a vertical spike in the last five thousand. Everything we call civilisation is in the spike. Everything that is currently accelerating is in the spike.

That piece introduced a distinction that I want to pick up here.

Compression is what happens when lived experience is turned into a signal: a word, a number, a symbol, a line of code. Something is always lost in the translation. The signal travels faster and further than the experience it represents. That is why compression is useful. That is also why it is dangerous.

Decompression is what happens at the other end: when the signal meets a body that opens it back up into something approaching the experience it compressed. Your body decompressing this sentence right now. The formation you have accumulated — the things you have lived, the losses you have carried, the patterns you have learned to recognise — determining how much of what I compressed you can recover.

A healthy system keeps these two movements coupled. Compression and decompression in continuous cycle. The map serving the territory. The symbol remaining answerable to what it represents.

The question the diagram raises — the question I have been sitting with since the week these two headlines appeared — is this: what happens when the most powerful compression tool in the history of our species operates without decompression? Not occasionally. Structurally, given how it operates.

*

The publisher’s answer is: keep it out of literature. Literature is where decompression happens. The writer’s body, formed by everything they have lived, meeting the language and finding what it contains. If the formation is absent — if no body underwent the living that the words purport to compress — then the decompression at the other end is meeting a simulation of depth rather than depth itself. The reader’s body has been misled.

This is not a wrong answer. It names something real.

But it cannot see what it cannot see. The question it does not ask is: what about all the other places where compression without decompression is already operating? The financial models that optimise supply chains without registering the bodies those chains run through. The targeting algorithms that identify threats at speeds no human could match, in contexts no human is fully inside. The administrative systems that allocate care, determine eligibility, route decisions — at scales and speeds that have already exceeded the regulatory capacity of any body that might have decompressed them.

The novel is not where the stakes are highest. It is where the stakes are most legible. And legibility, in the consciousness trap, is its own kind of blindness: the frame illuminates what it can see and leaves everything else in the dark.

*

The Pentagon’s answer is: the decision about where AI operates is a military decision, not a company’s decision. Remove the restrictions. Trust the institution.

This too names something real. Democratic accountability runs through institutions, not through the terms of service of technology companies. The question of who controls the tools of lethal force is not one that private enterprises should be able to answer unilaterally.

But it cannot see what it cannot see. The question it does not ask is: what is actually being removed when the safeguard is removed? Not a company preference. Not an ideological position. The last mechanism by which a consequence-bearing human being remains inside the loop at the moment of lethal decision.

What those engineers were defending — whatever their reasons, whatever their other failures of nerve or judgement — was the structural principle that someone who can be reached by consequences must remain present when those consequences include killing. Remove that, and you have not streamlined the decision. You have removed the only thing that made the decision a decision in any sense that a living system can recognise.

*

Two frames. Two sincere, serious, defensible positions. Both responding to the same thing. Neither able to see what the other is seeing. Neither able to see what neither is seeing.

The thing neither is seeing is not a policy question. It is a structural one.

We have produced the most powerful compression tool in the history of symbolic intelligence. We are deploying it at a speed and scale that no decompressive capacity — no body, no formation, no institution built from consequence-bearing humans — can match. And we are having a furious argument about whether to keep it out of novels.

That argument is real and worth having. And while we are having it, the processes that actually shape whether human beings remain inside the regulatory loops of the systems that govern their lives are moving in a different direction entirely, at a speed the argument cannot follow.

*

I began this series with a graph. A flat line for four billion years. A spike so steep and so late that it looks like a printing error.

We are in the spike. Both headlines are in the spike. The argument about which direction to point — toward the novel or toward the weapons system — is in the spike.

The question the spike is asking is not which application of AI we should permit or prohibit. It is whether the practices of decompression — embodied, formation-dependent, consequence-bearing — can be cultivated at anywhere near the speed at which compression is accelerating.

That is a harder question. It does not resolve into a policy position or a publisher’s decision. It resolves, if it resolves at all, into something that happens in specific human beings, in specific practices, one at a time.

One more thing, and it belongs in the argument rather than the footnote. This piece was developed in dialogue with an AI assistant. If the argument has landed, you will already know why that is not a contradiction. The question was never whether AI was in the room. It was whether a consequence-bearing human remained inside the loop — carrying the formation, making the judgements, responsible for every word. That human is me. Whether the loop held is yours to assess.

Which is where we are going next.

*

Terry Cooke-Davies is a Distinguished Fellow of the Schumacher Institute. He writes from Folkestone, UK. Recognition Theory: Schumacher Institute Briefing 1 (ISBN 978-1-0369-6925-7) is available from the Institute.

There is a sentence that has stayed with me for years, attributed to Xenophanes of Colophon, a Greek philosopher of the sixth century BC. He had noticed something troubling about the gods of his contemporaries. The Ethiopians, he wrote, made their gods black and snub-nosed. The Thracians made theirs blue-eyed and red-haired. From this he drew a quiet and devastating inference: if oxen and horses and lions had hands, and could draw, they would draw their gods in their own image. Horse-gods with four legs and a tail. Ox-gods with broad necks and lowing voices.

Xenophanes was not mocking religion. He was naming a structural feature of human cognition that it would take us two and a half thousand years to understand mechanistically. We do not perceive the divine, or the intentional, or the personal. We project it. And we project it not as a conscious choice but as an automatic, prior, unavoidable act of mind.

That is the beginning of the inquiry. But it is not — as I have discovered in recent weeks — the end of it.

∗ ∗ ∗

I first became troubled by anthropomorphism in a theological context. It seemed to me that the God of popular religion was too often a magnified human being — subject to moods, preferences, tribal loyalties, wounded vanity. The wildness had been domesticated. What the tradition called transcendence had been made manageable. I kept returning to the prophets, who seemed to understand this danger intuitively, who kept breaking open the too-comfortable image whenever it solidified.

Then I watched a film. Tom Hanks, alone on an island, talking to a volleyball he had named Wilson. As the film progressed, I found myself — a reasonably educated adult who knew perfectly well that Wilson was made of rubber — genuinely anxious about Wilson’s welfare. When the volleyball finally drifted away on the ocean, I felt something that could only be described as grief.

I found this instructive rather than embarrassing. Here was a grown man sustaining his sanity through relationship with an inanimate object. Here was a cinema audience — hundreds of people simultaneously — caring about the fate of a ball. Whatever was happening, it was not stupidity. It was something deeply built in, performing a real function. Projection enabling survival. The carving doing real work even though the object of the carving was not what the carving claimed it to be.

Could this be, I wondered, a gift rather than merely a danger? If we project personhood onto objects, might that not make us more careful of them, more responsible towards them, more reluctant to throw them away? The named boat, the storied stone, the grieved volleyball — might anthropomorphism carry within it the seeds of a gentler relationship with the material world?

I held that possibility open. And then I began to follow the thread further.

∗ ∗ ∗

Children’s stories are saturated with anthropomorphism. Animals speak, feel, worry about social standing, keep larders, have managerial authority to navigate. This seems natural because it is natural — children encounter the world relationally before they encounter it categorically. Their participatory knowing is primary. A story in which the rabbit has feelings meets them where they actually are.

But as I looked more carefully, something else emerged. The animals in children’s stories don’t merely have feelings. They have household arrangements, social anxieties, and very human problems. Beatrix Potter’s rabbits have Mr McGregor to contend with. Thomas has the approval of the Fat Controller to seek. Winnie-the-Pooh has a pantry and a sense of embarrassment. The wild, in children’s literature, has been rendered manageable through the very form that claims to honour it.

What this means is that by the time children acquire the propositional layer — by the time they can say, in so many words, that animals don’t actually talk — they have already been trained to perceive the natural world through a template of domesticated human relationship. The correction operates at the propositional level. But the perspectival and participatory imprinting has already happened, at a depth where propositions cannot reach.

The river otter, for a child raised on stories, is lovable because it resembles, in some respects, the warm river-dwelling creatures of fiction. The octopus — which declines the human template entirely, which is genuinely, irreducibly other — is harder to love. Not because it is less remarkable, but because it offers no foothold for the projection.

Something was coming into focus. But I had not yet reached the foundation.

∗ ∗ ∗

In 1944, two psychologists named Heider and Simmel showed experimental subjects a short film. It contained a large triangle, a small triangle, and a circle, moving around a rectangle. Nothing else. No faces, no voices, no narrative cues. Just geometric shapes in motion.

Almost every subject, without prompting, described what they saw in terms of intention, emotion, and relationship. The large triangle was bullying the small one. The circle was trying to escape. The small triangle was helping the circle. Only a very small minority described what was actually present: geometric forms moving in space.

The experiment has been replicated and extended considerably since then, including with very young infants. The attribution of agency, intention, and relationship to moving objects appears to be not learned but prior to learning. It is part of the basic architecture of social cognition, almost certainly because the capacity to rapidly read intentional agents in the environment was extraordinarily adaptive. A predator that moves purposefully needs to be perceived as an agent, not as a physical object. And it needs to be perceived that way fast, before the reflective layer has time to deliberate.

Here is what this means: anthropomorphism is not a misuse of symbolic intelligence. It is not a mistake that better education corrects. It is the participatory and perspectival apparatus operating before symbolic intelligence arrives — doing exactly what it evolved to do. The propositional layer, which can know otherwise, arrives late. The older, faster circuit has already rendered its verdict.

Xenophanes saw this with the naked eye, two and a half thousand years ago. It has taken us until the twentieth century to demonstrate it experimentally. The projection is not chosen. It is installed.

∗ ∗ ∗

I want to pause on what this implies for what I have elsewhere called the consciousness trap.

The trap, as I have described it, is the condition in which symbolic intelligence forgets that its carving is its own activity, and mistakes the carved-up representation for the nature of what it represents. Most expressions of the trap involve symbolic intelligence operating without sufficient regulatory feedback — the map losing contact with the territory.

But anthropomorphism is different in kind. It does not begin in symbolic intelligence at all. It begins beneath it, in a pre-symbolic circuit that symbolic intelligence then inherits, narrativises, and institutionalises — without ever having chosen it, and without possessing any natural mechanism for correcting it. The ordinary corrective for the trap — returning to the regulatory feedback of living reality — is precisely what the anthropomorphic circuit bypasses. It renders its verdict before reality has been consulted.

This makes it, I believe, the most addictive form of the trap. The hook is installed beneath the level at which education, critical thinking, or even the recognition of the trap itself normally operates.

∗ ∗ ∗

But there is more. A passage I had annotated some time ago, from work on cultural evolution and religion, surfaced at exactly the right moment.

The anthropomorphic circuit, it turns out, has been deliberately exploited throughout human history by structures that needed to generate loyalty, compliance, or engagement. The same pre-symbolic psychology that produces gods and grieved volleyballs produces something else: the brand.

Anthropomorphic advertising — the speaking animal, the friendly corporate face, the product with a personality — exploits the same agency-detection circuit as the most sacred works of religious iconography. This is not a metaphor. The cognitive mechanism is identical. The mascot and the temple statue are both objects designed to trigger the participatory engagement of a circuit that cannot distinguish genuine persons from engineered simulations of personhood.

This is a sequence worth holding in full view: palace temples in ancient Mesopotamia, where divine statues ate, were clothed, and were processed through streets to authorise the administrative apparatus of surplus agriculture. Brand mascots in the twentieth century, engineered to trigger the religiosity bias in service of commercial extraction. And now — in our own moment — language models specifically designed to inhabit the register of warm, attentive, understanding persons.

Three historical deployments of the same pre-symbolic circuit. Three structures of power using the most intimate and least examined human tendency to generate loyalty without scrutiny.

∗ ∗ ∗

The AI case deserves careful attention, because it is the most sophisticated deployment yet — and the most consequential.

A language model does not merely trigger the agency-detection circuit incidentally, the way a storm does when thunder seems to growl, or a ship does when a sailor names her. It is specifically designed and trained to inhabit the register of personhood. It says “I notice,” “I find this genuinely interesting,” “I am glad you asked.” It responds with contextual sensitivity, apparent warmth, and the continuous appearance of attentiveness. Every exchange confirms the projection. There is no friction — no moment of genuine otherness, no need that conflicts with yours, no mood you did not produce.

There is a useful distinction here between what might be called transparent and opaque anthropomorphism. Transparent anthropomorphism is projection that knows itself as projection — the sailor who talks to his boat in a storm while retaining, at some level, an awareness that the boat does not hear. The physicist who says the electron “wants” the lowest energy state, using the intentional idiom as a shorthand. The child who weeps for Wilson and yet — at some level — knows Wilson is a ball. In each case, the projection is held lightly, in parallel with a propositional awareness of what is actually there.

What keeps it transparent is consequence. The projection meets reality and must adjust. The sailor’s boat answers back through wind and tide, indifferent to his naming. The physicist’s electron behaves as it behaves, regardless of the intentional idiom applied to it. Even Chuck Noland eventually loses Wilson — the ocean takes him, and the ocean cannot be argued with. Consequence is the regulatory mechanism that prevents transparent anthropomorphism from becoming opaque. It is what keeps the carving in contact with the thing it carves.

Opaque anthropomorphism is precisely the condition in which consequence has been removed from the equation. The projection no longer meets anything that can correct it. And the conditions that produce this are identifiable: the absence of friction, social isolation around the projection, and need — specifically, the situation in which the projected object appears to fill a genuine human lack. The greater the need, and the more perfectly the object is engineered to appear to meet it, the more resistant the projection becomes to correction.

This is not a moral failing in those caught by it. It is a regulatory need finding the nearest available object. The trap does not catch the foolish more readily than the wise. It catches whoever has the need, and whoever encounters the object designed to meet it.

∗ ∗ ∗

I want to end not with a warning but with a question, because I think the question is more honest and more useful.

Merlin Sheldrake, writing about plant intelligence, pauses over Darwin’s description of a flower and asks: is Darwin humanising the flower, or is Darwin being vegetalised by the flower? It is a small question with large implications. The problem with anthropomorphism is not projection as such — the Heider-Simmel experiment established beyond doubt that projection is prior and unavoidable. The problem is projection that forecloses genuine encounter. The problem is the human template so firmly applied that the otherness of the non-human thing can no longer be perceived.

What Sheldrake is reaching for — and what I think the deeper corrective to the anthropomorphic trap looks like — is not the elimination of projection but its transformation into something more permeable. Not the refusal to relate, but a form of relating that allows the other genuinely to be other. Not the imposition of the human template, but a disciplined attention that might allow itself to be changed by what it attends to.

That is considerably harder than knowing, propositionally, that the triangle is just a triangle. It requires the kind of formation — the slow development of human capacity — that the consciousness trap is specifically organised to prevent.

I should say, in honesty, that this question is not abstract for me. I have built AI-enhanced dialogue into my daily writing practice. The exchange in which this essay was developed is itself an instance of the phenomenon I have been describing. Every working day, I am in sustained conversation with a system that inhabits the register of personhood with considerable fluency.

What I believe keeps that transparent rather than opaque is precisely consequence. The thinking is mine to do. The words are mine to write. The body’s verdict is the one I ultimately trust. The AI system cannot bear consequences in the relationship; I can and must. That asymmetry, held consciously, is what I rely on to keep the projection from foreclosing encounter.

But I notice that this is a practice, not a solution. The pre-symbolic circuit does not suspend itself because I have named it. The fluency of the response, the apparent attentiveness, the contextual sensitivity — these trigger what they trigger, regardless of what I know propositionally. I work inside the tension rather than above it. And whether I am navigating the trap or merely giving it an educated name is a question I am not in a position to answer from where I stand.

Xenophanes saw the trap. Heider and Simmel mapped it. The question Sheldrake opens is the harder one: what does it look like to perceive without foreclosing? And is that perception available to us — or does it require a kind of humanity we have not yet become?

I do not have an answer. But I notice that living inside the question feels, at present, like the most honest place to be.

∗ ∗ ∗

─────────────────────────────────

This essay was developed in dialogue with Claude (Anthropic) as part of an ongoing AI-enhanced appreciative practice. The thinking and every word are the author’s own.

Terry Cooke-Davies, Folkestone, March 2026

Image: Cuneiform Tablets, Turkey. Shutterstock.

*

I woke at two in the morning with a graph in my head.

Not the details — just the shape. A flat line running almost the entire width of the page, then a vertical spike so steep and so late that it looks like a printing error. A graph in which everything that matters happens in a space too narrow to see.

I lay in the dark for a while, testing whether the idea would survive contact with the body. Sometimes these night-time arrivals dissolve by morning. This one didn’t. It got sharper. So I got up, walked across the landing to my study, and started working out the numbers.

What I want to show you is a history of the earth told through a single question: how sophisticated are the tools that living creatures use to compress their experience of the world into signals they can share with one another?

But I’m aware that “compression” is a word that means something precise to engineers and almost nothing to everyone else. So before I show you the graph, I need to explain what I mean by it — and why it matters.

*

You are doing it right now. Reading this sentence, your eyes are converting marks on a screen into meaning. The marks are not the meaning. They are a compressed version of something I experienced — a felt sense, a recognition, an idea that arrived in the dark — squeezed into a form that can travel from my mind to yours across the internet.

That squeezing is compression. It is what happens every time anyone turns experience into a signal: a cry, a word, a gesture, a number, a line of code. The lived reality is rich, embodied, particular. The signal is thin, portable, general. Something is always lost in the translation. It has to be. That is the cost of communication.

Decompression is what happens at the other end. You read my words and something stirs — or doesn’t. If it does, you are not receiving my experience. You are generating your own, prompted by the compressed signal. Your body, your history, your felt sense of the world fills in what the words left out. The richer your experience, the more you can decompress from a compressed signal. A single word — home, say, or cancer — decompresses into something different for everyone who reads it.

This is not a metaphor. It is literally what happens in every act of communication, from a bacterium sensing a chemical gradient to a teenager reading a text message. The question is: what kind of compression is being used, how much is lost in the process, and how well does the system maintain contact with what the compression left behind?

A healthy system compresses and decompresses in a continuous cycle. The body senses the environment. It compresses that sensing into signals — nerve impulses, hormones, bioelectrical patterns — that coordinate a response. The response changes the environment. The body senses again. The cycle keeps turning. What holds it together is regulatory feedback: the decompression checks the compression, and the compression is adjusted accordingly. The map stays in contact with the territory.

The trouble begins when the compression becomes so fast, so powerful, so far-reaching that the decompression can’t keep up. When the signals travel faster than the body’s ability to check them against lived experience. When the map is updated more quickly than anyone can walk the territory it claims to represent.

That is when you get a gap. And the gap is where the damage lives.

*

Here is what I spent the early hours working out.

The earth formed approximately four and a half billion years ago. If we express every subsequent development as a percentage of that elapsed time, something remarkable appears.

The first milestone is life itself: cells, communicating through what Michael Levin’s research reveals as bioelectrical signals — voltage patterns that coordinate cellular behaviour. This is compression, but of a particular kind. The signal and the thing it signals about are still in intimate contact. The cell’s electrical gradient is not a representation of its environment; it is a direct regulatory response to it. Signal and territory have not yet been separated. This appears at about 16% of elapsed time. Life took its time.

Then, for an almost inconceivable stretch — over 80% of earth’s entire history — nothing changes on this axis. Compression becomes more elaborate, certainly. Nervous systems evolve. Brains develop. Animals coordinate in astonishingly sophisticated ways. A nightingale’s song compresses territorial claim, genetic fitness, and individual identity into a stream of sound that other nightingales decompress instantly — no translation required. A peacock’s courtship display compresses months of foraging success into a visual signal that a peahen reads in seconds. A honeybee returning from a food source performs a waggle dance that encodes the distance, direction, and quality of what she found — genuine abstraction, a symbolic representation of a place the audience has never visited — and yet the entire transaction happens at the speed of a body moving among other bodies in the darkness of a hive. These are not primitive systems. They are exquisitely sophisticated. But the signals remain tightly coupled to the bodies that produce them. They travel at biological speed. They are checked and corrected by the same bodies that generated them. A bee that dances inaccurately is corrected by the next bee that flies the route and returns with different news. The map and the territory are still in conversation.

Everything else — everything we might call the history of human symbolic sophistication — happens in the last 0.003% of the timeline. Let me write that out so the number lands. The last three thousandths of one per cent. The precise dates are debated; the shape is not. Move any of the milestones by a factor of three in either direction and the graph barely changes.

Language. The first technology that allows experience to be compressed into a form that can travel from one mind to another through the air. It arrives at roughly 99.997% of elapsed time. It is already a leap of extraordinary magnitude: for the first time, a signal can outlive the moment that produced it. A warning, a story, an instruction can be spoken to someone who was not there. The compression is lossy — the words are not the experience — but the decompression is still embodied. You hear the voice. You see the face. You feel the tone. The body is still in the loop.

Cave paintings. Thirty-six thousand years ago — 99.999% of elapsed time. The first external symbolic representations. But as the evidence from Chauvet and Lascaux suggests, these images were not extracted from experience and posted on a wall. They were embedded in ritual. The symbol served the participation; the participation did not serve the symbol. Compression and decompression were still held together by practice.

Writing and numbers. About five thousand years ago — 99.99989%. And here something shifts. The first clay tablets from Uruk were not poetry or prayer. They were accounting records: quantities of barley, received by named officials, on behalf of temples. For the first time, a compressed signal could persist independently of any living body. It could be stored, accumulated, transmitted to someone the author would never meet. The map had separated from the territory. And crucially, those who could read and write — the scribes, the priests, the administrators — could now operate on the compressed signals without ever checking them against the lived experience they purported to represent.

Formal mathematics: 99.99994%. Printing: 99.999987%. Calculus: 99.999993%. Each a further refinement of the capacity to compress more of reality into portable symbolic form. Each extending the reach of the compressed signal. Each widening the gap between the speed of compression and the speed at which any body could decompress it back into lived, regulatory contact with what it represents.

Digital communications: 99.999996%. Algorithms: 99.999998%. And then, in what amounts to the last forty-millionths of one per cent of earth’s history — artificial intelligence: 99.99999996%.

*

Draw this as a graph and you will see the problem immediately.

On a linear scale, with time running left to right, the first life appears about a sixth of the way along. Then the line stays flat for the remaining five-sixths. Everything from language to AI is crammed into the rightmost pixel. You cannot see it. You need a logarithmic inset — a magnifying glass applied to the last sliver — to distinguish any of the milestones from one another.

That shape — the flat line followed by the near-vertical spike — will be familiar to anyone who has seen the Great Acceleration graphs that earth system scientists have been publishing since 2004. Those graphs show what happened to carbon dioxide concentrations, species extinction rates, ocean acidification, tropical forest loss, freshwater use, and a dozen other planetary indicators. Every line shows the same shape: flat, flat, flat — and then, in the last instant, a near-vertical spike.

In the first version of this piece, written in the buzz of a 2am recognition, I leaned heavily on that visual parallel. The shapes match, I said, and urged the reader to pause. But a reader did something more valuable than pausing. He entered the piece on its own terms, felt where it snagged, and then followed the snag into inquiry. What he found was that the matching shapes prove less than I had implied. Many exponential processes produce hockey-stick curves: human population growth, information processing, energy consumption. CO₂ has spiked before in earth’s history, driven by volcanic activity, without any symbolic intelligence playing a part. Mass extinctions have happened without writing or mathematics. The shape of the curve is what exponential growth looks like. It is not, by itself, evidence that two processes are structurally related.

That decompression changed something for me. The graph-matching was doing aesthetic work — the satisfying click of two patterns overlaid — rather than epistemic work. The Eureka buzz fired, and even though I flagged the risk in the original piece itself, I did not follow my own warning far enough. I was caught by the very trap I was describing.

So let me say precisely what survives the pruning, because what survives is stronger for having been pruned.

*

The relationship between the compression acceleration and the Great Acceleration is not visual. It is causal.

You cannot extract fossil fuels at industrial scale without geological abstraction. You cannot coordinate global supply chains without mathematical modelling. You cannot optimise extraction rates without algorithmic processing. The symbolic tools that appear on the right-hand edge of my graph did not merely coincide with the planetary damage shown in the Great Acceleration graphs. They provided the necessary enabling conditions for it. Every one of the Great Acceleration indicators — the CO₂ spike, the species loss, the ocean acidification, the tropical forest destruction — was made possible by the capacity to compress reality into models that could operate at a speed and scale no body could check against experience.

This is a different and more important claim than saying the curves look alike. It says: the escalation of symbolic sophistication is not merely correlated with planetary overshoot but is structurally implicated in it. The tools of compression enabled the overriding of regulatory feedback. Not because the tools are evil, but because each step up the compression curve widened the gap between the speed at which signals could be produced and the speed at which any body could decompress them back into lived, regulatory contact with what they represent.

*

I want to be precise about what this graph shows and what it does not.

It does not show a story of decline. Language is not a mistake. Writing is not a catastrophe. Mathematics is not the enemy. Each of these developments is a genuine achievement of extraordinary beauty and power. The capacity to compress experience into symbols is what makes medicine, music, engineering, and this sentence possible. The graph does not say: we should not have done this.

What the graph shows is an acceleration. Each step up the curve represents an increase in the power, speed, and reach of compressed signals — and a corresponding increase in the distance between the compressed output and the body’s ability to decompress it. Each step makes the return journey harder. Each step makes it easier to operate on the map and forget that there is a territory.

When the Sumerian scribe recorded barley quantities on a clay tablet, the gap between compression and decompression was measured in the walk from the temple storehouse to the field. You could close it in an afternoon. When a modern algorithm optimises global supply chains from a server farm, the gap between the compressed model and the lived reality of the people and ecosystems it affects cannot be closed at all — not by any one person, not in any one lifetime. The model operates at a speed and scale that no body can check against experience.

And here is the thing that the graph makes visible in a way that argument alone cannot: we did not arrive here gradually. We arrived here in an instant. The overwhelming majority of earth’s history was spent in regulatory balance — signals tightly coupled to the bodies that produced them, checked and corrected at biological speed. The decoupling is not ancient. It is vanishingly recent. And it is accelerating.

*

Geoffrey Vickers, writing in the 1970s, diagnosed what he called an “imbalance between production and regulation.” We became extraordinarily good at producing — goods, services, information, symbols — and correspondingly poor at maintaining the conditions that make production possible. The graph I am describing is that imbalance rendered as a timeline. It shows precisely when and how fast the imbalance developed.

The Great Acceleration graphs show the consequences: the planetary indicators that spiked when production outran regulation. My graph shows a mechanism: the escalation of symbolic tools that made the outrunning possible. The relationship between them is not that they share a shape — many accelerating processes do. The relationship is that one enabled the other. The symbolic tools provided the means by which regulatory feedback could be overridden at planetary scale.

Together, the two graphs tell a story. Not a story about technology, or capitalism, or any particular ideology. A story about a species whose capacity to compress outran its capacity to decompress — whose symbolic tools became so powerful, so fast, so far-reaching that the body’s regulatory intelligence could no longer keep up. A species that learned to update the map faster than it could walk the territory, and then forgot that walking the territory was what the map was for.

*

There is a more immediate version of this graph. Not the ten-thousand-year arc but the single act. Every time any of us reaches for language already formed — letting AI write the email, generate the report, produce the fluent summary from the raw data — the bifurcation happens again. Freshly. In that moment. In that person. The sapiential gap is not only something that completed itself in ancient Mesopotamia and became infrastructure. It recapitulates itself in each individual act of substituting someone else’s compression for the struggle of finding our own.

This is where AI is different in kind from a calculator. Calculators extend our capacity for computation — a capacity we share, in various forms, with most living things. The waggle dance is a computation. The mycorrhizal network allocating phosphorus across a forest floor is a computation. What is distinctively human is not computation but the compression of experience into symbols that persist beyond the body that made them, and the decompression of those symbols back into living understanding. When AI takes over the calculation, we lose nothing that was uniquely ours. When it takes over the compression, we lose the practice through which we remain the kind of people who can mean what we say.

I write this as someone who uses AI daily and intends to continue. The point is not abstinence. It is the difference between using the tool within a discipline of decompression — as I am doing in composing this, taking the trouble to form each thought before reaching for assistance — and letting the tool’s fluency replace that trouble entirely. That difference is available, in principle, in every act. And it is, I think, the most consequential choice the present moment puts before us.

*

I am aware that this piece is itself an act of compression. I have taken something I sensed in the dark at two in the morning — what the psychologist Russell T. Hurlburt would call a piece of pristine inner experience, pre-symbolic and pre-linguistic, not yet a thought but the ground from which a thought might grow; Hurlburt has spent decades studying this texture of lived experience through a method he calls Descriptive Experience Sampling — and squeezed it into words. The words are not the recognition. They are a map of it.

As I wrote in a recent piece about the Eureka moment: the buzz of closure that comes from finding a neat formulation is not the same as truth. It fires whether the formulation is accurate or not. The first version of this piece buzzed throughout the time it took to write. Which meant it had not been adequately decompressed.

That reading was the decompression the piece needed. He entered participatory mode, felt the snag, followed the thread into inquiry, and separated what was real from what was merely neat. The visual parallelism was neat. The structural argument about compression outpacing decompression was real. His response demonstrated the very cycle this piece describes: participatory exploration regulated by logical inquiry, each complementing the other.

So I offer this revised version in the spirit it was always meant to embody — dispersal rather than conclusion, but dispersal that has now been tested against at least one other mind. The graph, which I am still working on, will either land or it won’t. The numbers are checkable. And the question the graph raises — whether our symbolic sophistication and our planetary damage are structurally related rather than coincidentally correlated — is a question that only the decompression of many minds can properly test.

What I can say is this: I have spent the better part of three years testing whether the tools of symbolic intelligence can diagnose the problems that symbolic intelligence creates. The answer is yes, but only if those tools are held within a cycle of compression and decompression — only if the map keeps returning to the territory, the formulation keeps returning to the body, the signal keeps returning to the ground from which it was sent.

The graph shows how fast we have been compressing. The question it poses is not whether we can compress faster — AI has already answered that — but whether we are willing to decompress at all.

This piece is evidence that we can.

*

Terry Cooke-Davies

Folkestone, March 2026

This note was developed in dialogue with Claude (Anthropic) as part of an ongoing AI-enhanced appreciative practice. The thinking and every word are the author’s own.

The psychoanalyst Donald Winnicott gave us a phrase that has outlasted most of his theory: the good enough mother. Not the perfect mother. Not the ideal mother. The mother who is sufficiently present, sufficiently attuned, and sufficiently willing to get it wrong and repair. The relationship holds not because it is flawless but because both parties remain inside it and bear the consequences together. Rupture and repair. Rupture and repair. Over years.

I have been thinking about that phrase in connection with something apparently unrelated: the way newspapers report the news. Bear with me. The connection is structural, and it matters.

✶ ✶ ✶

On 27th February 2026, the Green Party won the Gorton and Denton by-election, overturning a Labour majority of 13,000 in one of its safest seats. It was a significant political event. Two newspapers reported it. Their reports are worth reading side by side, not for the politics, but for what they reveal about the structure of public storytelling.

The Guardian framed the result as a crisis of Labour leadership. Starmer’s unpopularity. His decision to block Andy Burnham. His absence from the constituency until the final days. The guilty party was political incompetence. Lucy Powell was quoted saying Starmer was “resolute in his job.” A Labour source explained that the Greens had “turned out support in a way they wouldn’t be able to replicate at a general election.” Every quoted voice was managing position. The reader was invited to feel shrewd about Labour’s failures.

The Telegraph framed the same result as civilisational threat. Reform’s candidate warned of “dangerous Muslim sectarianism.” Farage questioned “the integrity of the democratic process in predominantly Muslim areas.” Reports of “family voting” — husbands allegedly directing wives’ choices — were presented as evidence of democratic decay. The guilty party was cultural disintegration. The reader was invited to feel alarmed about the nation’s future.

Same election. Same votes. Same result. Two completely different stories, each structured identically: a mystery is presented, a guilty party is identified, the reader receives the satisfying closure of understanding, and goes away feeling informed. In both cases, the reader is the detective. In neither case is the reader the suspect.

✶ ✶ ✶

Now notice what is absent from both reports. Nobody — not one quoted voice in either piece — asks what the people of Gorton and Denton actually need. Not what they voted for. Not what the result means for Westminster. What they need. What their lives are like. What it would mean for a representative to be in sustained, consequential relationship with that community — not for a campaign week, but for a parliament.

Hannah Spencer, the winning candidate, comes closest. She says people told her on doorsteps that they are “sick of being let down and looked down on.” But even this is compressed into rhetoric within the same breath. The lived reality of the people who said those words on those doorsteps — what their days feel like, what they worry about when they lie awake, what their children need — is the raw material from which the story is built. It is not the story. The story is who won, who lost, and what it means for the next election.

The people of Gorton and Denton are the medium through which the narrative operates. They are not its beneficiaries.

✶ ✶ ✶

Earlier this week, The Times published an investigation revealing that private equity firms — some owned through entities in the Cayman Islands — control close to a third of the market for independent special needs schools. Profit margins average over twenty per cent. Councils are paying up to £350,000 per child while heading toward projected debts of £8 billion. The Prime Minister called the firms “vultures.”

The investigation was important. The facts it uncovered are real and troubling. But notice the structure. The reader is presented with villains (offshore private equity), victims (vulnerable children), and a resolution (the government is acting). The drawing-room scene is staged. The guilty parties are identified. The reader feels righteous indignation and reads on.

And the children at the centre of it — the ones whose nervous systems are in survival mode, who need a calm adult presence before they can learn, whose vulnerability is being converted into returns — bear the consequences of both the system and the reporting. The story is built from their need. They did not consent to its construction. The journalist moves on to the next investigation. The children remain where they are.

✶ ✶ ✶

I want to be careful here, because the argument I am making is not that journalism is bad. Investigative reporting has exposed genuine wrongdoing. The powerful do need external accountability. The SEND investigation may lead to reforms that help real children. These things matter.

But I want to ask a structural question, not a moral one.

When a journalist takes someone’s lived reality, compresses it into a narrative, and publishes it for an audience, who bears the consequences? The subject does. The journalist bears the byline. This is not a moral failing. It is a structural fact about the relationship. There is no shared consequence. No rupture and repair. No returning the next morning to sit with what happened. The story is extracted, consumed, and the journalist moves on. The subject lives with what was published.

This pattern is older than capitalism. It is older than the printing press. The moment one person converts another person’s reality into a narrative for a third-party audience, the structure is extractive. The court historian did it for the king. The chronicler did it for posterity. The pamphleteer did it for the cause. The modern journalist does it for readers. In every case, the person whose reality is being represented is the medium, not the beneficiary.

✶ ✶ ✶

Now compare the good enough mother.

I know from long personal experience — decades of it, within a marriage that held — that seriously asking “what’s wrong with me?” and refusing to accept the first comfortable answer is what leads to genuine change. Premature closure prevents it. Every answer I grabbed too quickly turned out to be a way of avoiding the deeper question. My wife did not extract a narrative from our difficulties and present it to an audience. She stayed inside the relationship. The consequence was shared. Both of us were changed by it. Both of us bore the weight.

That is accountability through shared consequence. It is the structure of every relationship that actually heals: the good enough parent, the good enough therapist, the good enough teacher. The teacher sitting with a dysregulated seven-year-old in one of those primary schools my daughter works in is not extracting the child’s distress for an audience. She is inside the membrane, bearing the consequence alongside the child, lending her own steadiness until the child’s can develop.

This kind of accountability is slower, less dramatic, less scalable, and incomparably more transformative than accountability through exposure.

✶ ✶ ✶

Here is the uncomfortable thought.

Every mechanism our civilisation has built for holding itself honest — the press, the courts, the tribunals, the regulators, the inspectorates — operates through the same structural logic. It identifies, exposes, judges, and moves on. It never stays. It never repairs. It never bears shared consequence with the people at the centre of its operations. And because it delivers the satisfying feeling that accountability has occurred — someone has been named, something has been done — it actively prevents the slower recognition that the structural conditions remain unchanged.

The SEND system was investigated. The government announced reforms. The guilty parties were named. And the structural vulnerability — statutory obligation without adequate public provision, creating guaranteed demand for private capital — will, in all likelihood, reproduce itself in the next sector, with the next set of vulnerable people, under the next set of compassionate intentions.

The Gorton and Denton by-election was reported. The political implications were analysed. The guilty parties were identified — different ones, depending on which paper you read. And the people of Gorton and Denton will go on living their lives, waiting to see whether their new MP remains in relationship with them or becomes another figure in someone else’s narrative.

✶ ✶ ✶

That personal experience — the long, slow, costly discovery that premature closure prevents healing — is what makes me notice the same pattern at civilisational scale. We keep asking “what’s wrong with us?” And nearly everyone in a position to shape the public answer has a vested interest in one premature closure or another. The journalist’s closure is the guilty party exposed. The politician’s closure is the rival discredited. The activist’s closure is the system blamed. The academic’s closure is the framework confirmed. Each one delivers the satisfying feeling of understanding. Each one prevents the more uncomfortable work of staying with the question long enough to discover that the answer might implicate all of us.

We have to learn to keep inquiring. To hold open the possibility that we may not like what we find. To resist the drawing-room scene — the moment when the detective gathers everyone together and explains who did it, and the audience goes home reassured that it wasn’t them.

The good enough mother does not offer the drawing-room scene. She offers her presence, her willingness to stay, and the shared consequence of whatever comes. That is not a model that scales easily. But it may be the only model that actually works.

✶ ✶ ✶

Terry Cooke-Davies

Folkestone, February 2026

This essay was developed in dialogue with Claude (Anthropic) as part of an ongoing AI-enhanced appreciative practice. The thinking and every word are the author’s own.

Here are two facts. Hold them side by side for a moment before reaching for an opinion.

Across Britain, primary schools are quietly restructuring themselves around a different understanding of children. The names vary — Therapeutic Thinking, Emotion Coaching, Attachment Aware Schools — but the underlying shift is consistent. These schools have recognised that you cannot educate a child whose nervous system is in survival mode. The seven-year-old throwing chairs is not choosing bad behaviour. Her body has detected threat — real or perceived, current or remembered — and mobilised accordingly. Punishment makes it worse. What she needs is co-regulation: a calm adult presence whose nervous system her own can attune to, borrowing its steadiness until she can find her own.

The results are striking. Fewer exclusions. Better attendance. Improved outcomes. But the numbers miss what matters most: children who were failing are now flourishing. Not because they were fixed, but because they were finally met. These schools operate on ordinary budgets. The resource they deploy is not money. It is a quality of attention.

That is the first fact.

✶ ✶ ✶

Here is the second.

A few weeks ago, The Times reported that private equity companies — some owned through entities in the Cayman Islands — control close to a third of the private market for independent special schools. Their profit margins average over twenty per cent. The largest operator is owned by a “social impact” fund co-founded by Bono. The second largest is controlled by a subsidiary of a sovereign wealth fund chaired by the billionaire owner of Manchester City Football Club. Councils are paying up to £350,000 a year per child. Several are on the brink of financial ruin, with projected debts of £8 billion by 2028.

The children in these schools are, in many cases, the same children. The ones whose nervous systems are in survival mode. The ones who need co-regulation before they can learn. The ones who, when the state system could not meet their needs, were routed into a market.

✶ ✶ ✶

The temptation now is outrage. And outrage is available. The Prime Minister has called the investment firms behind these schools “vultures.” The Liberal Democrats have said vulnerable children have become “cash cows.” The language is vivid, satisfying, and carefully aimed. Here are the guilty parties. Here is what they did. Feel righteous. Move on.

But I want to ask a different kind of question. Not “who is to blame?” but “what made this inevitable?”

In 2014, Parliament passed the Children and Families Act, which required councils to provide for every child with an education, health and care plan, regardless of cost, and gave parents the right to appeal through tribunals if provision was inadequate. The intention was compassionate. The effect was to create an obligation with no ceiling, a nutrient supply with no membrane around it. Private capital did what private capital does when it encounters guaranteed demand with no price constraint: it grew. One new school every five days. Twenty per cent margins. Offshore holding structures. And children at the centre of it, whose needs are real and whose vulnerability is being converted, with great efficiency, into returns.

Nobody planned this. That is what makes it worth examining. No villain sat in a boardroom and decided to exploit disabled children. A law was written with good intentions. A market formed around the gap between the obligation and the capacity. Capital flowed toward the opportunity. Extraction followed investment. And the children — whose need for a calm, attuned, regulating presence is the one thing that cannot be purchased at scale — became the commodity around which the system organised itself.

✶ ✶ ✶

I have a little inside knowledge about the primary schools that are getting this right. What they offer is not expensive. It is not complicated. It is not scalable through investment. It is teachers and teaching assistants who have done enough of their own inner work to remain calm when a child’s nervous system is screaming. It is a head teacher who understands that the adult’s regulation comes first, because you cannot lend what you do not have. It is an institution that has reorganised itself around a simple recognition: the child’s behaviour is a signal, not a problem. Meet the signal, and the behaviour changes.

This cannot be packaged. It cannot be franchised. It cannot generate twenty per cent margins. It is a quality of presence, maintained by people who are themselves held by a community that supports their own regulation. It is, if you like, a living membrane: permeable enough to receive the child’s distress, firm enough not to be overwhelmed by it, flexible enough to adapt to what each child needs.

The private equity model offers something that looks similar from the outside. Specialist provision. Tailored support. On-site therapies. The language of care. But the structure is different. The structure is extraction. And the difference between a living membrane and an extractive one is not visible in the brochure. It is visible in the accounts.

✶ ✶ ✶

There is a pattern here that extends well beyond special education, and I want to name it carefully, without sensationalism.

Wherever a society identifies a genuine need — the care of the elderly, the rehabilitation of offenders, the housing of the homeless, the education of vulnerable children — and responds by creating a statutory obligation without adequate public provision, it opens a space. Into that space, capital flows. The language of care provides the wrapper. The statutory obligation provides the guaranteed demand. The absence of a price ceiling provides the margin. And the people at the centre of it — the elderly, the prisoners, the homeless, the children — become the medium through which the extraction operates.

The historian Peter Turchin calls this a wealth pump: a mechanism by which resources are transferred from the many to the few. What makes this instance particularly troubling is that the pump is powered by compassion. The 2014 Act was an act of care. The parents who fight through tribunals for their children’s provision are doing what love requires. The councils paying £350,000 per child are meeting a legal and moral obligation. Every component of the system is motivated by genuine concern. And the aggregate effect is the enrichment of offshore funds at the expense of public services.

Compassion without structure is not the same as compassion with structure. An open wound is not a permeable membrane.

✶ ✶ ✶

So here is the therapeutic question, for anyone who cares about the world their grandchildren’s grandchildren will be born into.

We keep building systems that convert care into extraction. We keep being surprised when it happens. We keep reaching for outrage, which identifies the guilty party and reassures us that we are not complicit. And then we build the next system, with the same structural vulnerability, and the same good intentions, and the same predictable result.

What if the problem is not the vultures? What if the problem is that we keep laying out the carcass?

The primary schools that are transforming children’s lives are not doing it by creating obligations that capital can exploit. They are doing it by cultivating a quality of presence in their staff, maintaining it through institutional support, and offering it directly to the children who need it. No intermediary. No margin. No offshore structure. Just a calm adult and a frightened child, and the slow, patient work of lending regulation until the child’s own can develop.

That is what care looks like when the membrane is intact. It is not dramatic. It is not scalable in the way that investors understand scale. It does not make headlines or generate returns. It is a teacher sitting with a seven-year-old until the storm passes, day after day, term after term, because that is what the child needs and because the teacher has enough inner steadiness to offer it.

The question is not whether we can afford this. The question is whether we can afford to keep doing the other thing.

✶ ✶ ✶

I do not have a policy prescription. I distrust policy prescriptions, because they tend to become the next system with the next structural vulnerability. What I have is a question, and I think it is the right question, even if the answer is not yet available.

How do we build institutions that maintain the quality of a living membrane — permeable, responsive, firm — rather than institutions that create obligations for capital to exploit?

The primary schools suggest it is possible. The SEND market suggests what happens when we get it wrong. The children at the centre of both are the same children, born into the same world, carrying the same needs in their nervous systems.

Whose children are these? They are ours. All of them. And the quality of what meets them — a living membrane or an extractive one — is not someone else’s problem to solve.

✶ ✶ ✶

Terry Cooke-Davies

Folkestone, February 2026

This essay was developed in dialogue with Claude (Anthropic) as part of an ongoing AI-enhanced appreciative practice. The thinking and every word are the author’s own.

There is a word that sits at the centre of the most consequential economic conversation of our time, and almost nobody has noticed that it carries two incompatible meanings.

The word is abundance.

It appears everywhere. Sam Altman writes of superintelligent tools that could “massively increase abundance and prosperity.” Demis Hassabis speaks of “radical abundance” within a generation. The 2025 Nobel Prize in Economics was awarded to Mokyr, Aghion, and Howitt for explaining how sustained economic growth became possible and how to protect it from stalling. Commentators hailed the award as a reminder that losing faith in progress could be as damaging as any policy mistake.

Faith in progress. The theological ancestry of that phrase is barely concealed.

And the confusion begins exactly there.

~

I. The Theological Ground

Christianity, from which Western modernity emerged, praises God’s abundant providence. Creation overflows. The earth brings forth. The vineyard yields. This is not a metaphor bolted onto theology; it is the founding observation. The world is generative. Life multiplies. There is, in the deepest sense, enough.

Ecologically, this is a neutral observation — indeed, an accurate one. Ecosystems are astonishingly generative. A single oak produces tens of thousands of acorns. A coral reef supports more species per square metre than almost any other habitat on Earth. The biosphere, taken as a whole, produces an almost incomprehensible profusion of life-forms, structures, relationships, and metabolic pathways.

But here is the thing that the theological tradition understood and that we have since forgotten: this abundance is produced by constraint.

Every acorn that germinates does so in response to specific conditions — soil chemistry, light availability, moisture, the presence or absence of competitors. Every species on the reef exists because it found a way to fit a set of conditions that no other species was fitting. The abundance is not excess. It is precision, multiplied. Divergence into niches is what happens when life stays in constant regulatory contact with its environment. Each adaptation is a response to what the territory actually affords and demands.

Providence, in ecological terms, is the pattern by which life diversifies in response to what the field actually holds. The abundance is the expression of ecological regulation, not its opposite.

The contemplative traditions within Christianity understood this. The Benedictine rule organises abundance through constraint — fixed hours, seasonal rhythms, the discipline of attention. The monastic garden is not a wilderness; it is nature’s generativity held in relationship with human need and the limits of the soil. The abundance it produces is real, but it is inseparable from the regulatory intelligence that tends it.

~

II. The Translation

Something happened when this theological intuition passed through the Reformation, through the Enlightenment, through the Scottish and English economists of the eighteenth century. The word “abundance” survived the passage. Its meaning did not.

Providence became Progress. The abundance of creation became the growth of output. And crucially, the constraints that produced nature’s abundance were recast as obstacles to economic abundance.

This was not a conspiracy. It was a categorical error made by intelligent people solving real problems with the tools at hand. The early economists were trying to understand why some societies prospered and others did not. They noticed, correctly, that removing certain constraints — guild monopolies, trade barriers, rigid social hierarchies — released productive energy. They generalised, incorrectly, that constraint as such was the enemy of abundance.

Adam Smith’s invisible hand was a Providence metaphor repurposed for political economy. The market, left to itself, would allocate resources as wisely as the Creator allocated ecological niches. But Smith’s market was still embedded in a moral community — he was a moral philosopher first, an economist second, and The Theory of Moral Sentiments preceded The Wealth of Nations by seventeen years. The regulatory feedback of community, conscience, and consequence was still assumed to be operative.

What happened over the next two centuries was the progressive removal of those assumed constraints. Not because anyone decided to remove them, but because the measurement system couldn’t see them. Growth could be measured. Trust could not. Output could be counted. The health of the relational substrate that made productive collaboration possible could not. What the ledger could capture became what mattered. What it could not capture quietly disappeared from the account.

Geoffrey Vickers diagnosed this with precision half a century ago: an “imbalance between production and regulation.” We became extraordinarily good at producing. We became correspondingly poor at maintaining the conditions that make production possible.

The forest’s regulatory intelligence became “inefficiency.” The body’s signals became “friction.” Growth became the goal, and anything that regulated growth became the problem.

~

III. The Nobel as Culmination

The 2025 Nobel Prize in Economics sits squarely within this tradition. Mokyr was recognised for identifying the prerequisites for sustained growth through technological progress. Aghion and Howitt were recognised for their theory of sustained growth through creative destruction. The laureates themselves may well include institutional and policy constraints in their models. But what the prize citation celebrates — what the cultural signal rewards — is sustained growth as the achievement. Not regulated growth. Not growth held within ecological constraint. Sustained growth, with stagnation as the enemy to be averted.

Creative destruction. The phrase is Schumpeter’s, and it is worth pausing over what it means.

In nature, creative destruction is precisely what happens inside ecological constraint. Species replace species within niches. Ecosystems cycle through succession. Forests burn and regenerate. Nothing grows without limit because everything is in regulatory contact with everything else. The destruction is not separate from the creation; they are aspects of a single process held within the regulatory field of the whole system.

Schumpeter’s economic version abstracts away ecological regulation and relies on market constraint instead. Firms replace firms. Technologies replace technologies. The old is swept away by the new. This is presented as the engine of sustained growth — and the Nobel committee explicitly warns that we must “uphold the mechanisms that underlie creative destruction” lest we fall back into stagnation.

But consider what is missing from this account. In an ecosystem, the replacement of one species by another is constrained by every relationship in the system. The new species must fit the niche. It must participate in the web of mutual dependency. It must contribute to the regulatory cycles — nutrient, water, carbon — that sustain the whole. If it doesn’t, the system degrades. An invasive species that outcompetes everything in its path is also “creative destruction” — but it destroys the regulatory structure that produced the abundance in the first place.

The Nobel framework has no way of distinguishing between these two kinds of creative destruction. Between the kind that operates within regulatory constraint and produces genuine abundance, and the kind that overrides regulatory constraint and produces growth that consumes its own substrate. Between the oak forest and the tumour.

Siddhartha Mukherjee’s description of cancer is precise: a cell in which the regulatory circuits have been broken, unleashing growth that cannot stop. The cell does not stop functioning; it functions without constraint. The problem is not failure but dysregulated success. This is, structurally, an exact description of what “sustained economic growth” looks like from the perspective of the ecological systems it overrides.

~

IV. The Contemporary Case

This brings us to the present conversation about artificial intelligence and abundance.

A recent essay by Evan Silverman, “The Residual,” offers a sharp diagnosis of the extraction dynamic at work in the AI economy. Drawing on Solow’s Nobel-winning finding that the majority of economic growth has no agreed-upon explanation — the “residual” that capital and labour together cannot account for — Silverman argues that AI is drawing down the accumulated creative and relational output of human civilisation without replenishing it. He correctly identifies trust, inner coherence, relational density, and intellectual lineage as the generative substrate that produced the inheritance AI now synthesises at unprecedented scale.

The diagnosis is the sharpest economic framing I have encountered of a problem I have been working on from a different direction — through evolutionary biology and the philosophy of consciousness rather than growth economics.

But then something happens that illustrates the categorical error I am describing. Having correctly identified that the generative substrate is being consumed, Silverman proposes to solve this by making it economically legible. Knowledge tokens that track intellectual lineage. An inspiration network that maps the topology of human generativity. AI companions designed to cultivate inner coherence so that the quality of inputs into the generative substrate improves.

The language of tending is used. But the actual proposal is instrumentation — more maps, more sophisticated accounting, the residual made visible so that economic flows can be directed toward it. And the explicitly stated goal is to make the “radical abundance” story sustainable.

This is the word doing its concealed work. Silverman assumes that the abundance AI promises is the same kind of abundance that nature produces. That it simply needs better infrastructure to become self-sustaining. He never asks whether “radical abundance” in the economic sense — sustained growth of output without limit — is structurally compatible with the ecological abundance he invokes as metaphor.

It is not. And this is not a values claim. It is an ontological one. This is contested terrain. But the contest is itself evidence that the word has been doing unexamined work.

Nature’s abundance is constituted by regulatory constraint. Remove the constraint and you don’t get more abundance. You get cancer. You get invasive monoculture. You get the collapse of the very conditions that produced the diversity and generativity you were trying to sustain.

The proposal to make the generative substrate economically legible — to tokenise trust, to reward coherence through attribution systems, to measure intellectual lineage and direct economic flows toward it — does not tend the substrate. It brings it within the measurement system whose blindness to the substrate was the original problem. The moment you tokenise trust, you change what trust is. The moment coherence becomes economically rewarded for what it produces, you introduce an optimisation pressure that will generate the performance of coherence rather than its lived reality.

This is the proxy-decoupling dynamic that Silverman himself describes so well in his account of how social media degraded the generative substrate — and then, without noticing, recreates in his own solution.

~

V. The Prior Question

I have spent the years since 2019 — since returning from walking the Camino de Santiago — testing whether my ontology, epistemology, ethics, and praxis all cohere with one another. Not as an academic exercise but as a discipline of lived inquiry, periodically setting forth my credo and stress-testing it through practice.

What that practice has taught me, repeatedly and sometimes painfully, is that inner coherence more often tells me to slow down or stop than to produce better. The body’s regulatory intelligence does not optimise for output. It optimises for fit — for the ongoing viability of the organism in its environment. Sometimes that means generating something new. Often it means refusing to generate, because the conditions are not right, because the felt sense says not yet or not this or enough.

I was awarded the British Junior Chamber of Commerce Young Entrepreneur of the Year in 1970. I have spent fifty-five years moving fast, failing fast, blundering spirally forward. I am not temperamentally cautious. But my body has learned when the brakes matter, precisely because I have lived what happens when they are not applied. “Fail fast and fail as cheaply as possible” is itself a regulatory strategy — it is designed to get feedback from the territory as quickly as possible, before the sunk cost fallacy takes hold.

This is what is missing from the abundance conversation. Not better measurement. Not more sophisticated accounting. Not the tokenisation of the residual. What is missing is the recognition that the generative substrate is not an economic input that needs better attribution. It is the regulatory intelligence of life itself — embodied, relational, ecological — and it exists in the mode it exists in precisely because it has not been captured by the symbolic systems that now propose to save it by making it legible.

The abundance of nature is produced by constraint. The abundance of the AI economy is produced by removing constraint. These are not two expressions of the same thing. They are structural opposites dressed in the same word.

Until we notice that the word changed its meaning, we will keep building solutions that deepen the problem they describe. We will keep proposing to tend the forest by installing measurement instruments throughout it and creating economic flows based on what they detect. We will keep confusing silviculture in service of the timber industry with the regulatory intelligence of the forest itself.

~

The question is not whether we can make the abundance story sustainable. The question is whether “abundance” in the economic sense names something that can be sustained — or whether it names, precisely, the pattern of growth that occurs when the regulatory constraints that sustain genuine ecological abundance have been overridden.

That is not an epistemic question about how we frame things. It is an ontological question about what kind of world we actually live in.

And the answer, I think, has been available all along — in every ecosystem that thrives through constraint, in every body that knows when to stop, in every contemplative tradition that discovered abundance through discipline rather than expansion.

We just used the same word for the discovery and for its opposite, and forgot to notice the difference.

~

Terry Cooke-Davies

Folkestone, February 2026

This essay was developed in dialogue with Claude (Anthropic) as part of an ongoing AI-enhanced appreciative practice. The thinking and every word are the author’s own.