When we first encounter them, they arrive as offices, names, reputations, histories. They seem almost architectural. They have held rooms we were not yet admitted into. They have made decisions before we understood what decision meant. They have lived close to crises, countries, institutions, wars, outbreaks, transitions. They appear not merely senior, but load-bearing.

For a long time, I saw some of my mentors this way.

Not as inhuman. That would be too crude. But as more than human in the old sense: larger, steadier, more durable, able to absorb more consequence than the rest of us. They carried memory when systems had forgotten. They held the line when others were still performing thought. They knew what could not be said in public but had to be acted on by Monday. They could enter a room and change its metabolism.

It is fashionable now to sneer at such people as global elites. There is some justice in that sneer. The world of senior rooms has its vanities, rituals, evasions, and insulation. It has its exhausting ceremonies of importance. It can mistake access for wisdom and recognition for moral worth.

But the sneer misses something real.

Some people really did carry a great deal of the world.

They were not merely enjoying status. They were absorbing panic. They were storing institutional memory in their bodies. They became the people others called when the system did not know what it was seeing. They became emergency infrastructure.

And then, one day, you see the infrastructure age.

You see that even the superhuman were bodies. Tired bodies. Ill bodies. Vain bodies. Loyal bodies. Bodies that still wanted to matter. Bodies that had been praised for endurance for so long that they no longer knew how to distinguish service from extraction.

That is when admiration changes shape.

It does not disappear. In some ways, it deepens. The old guardians become more moving, not less. Their achievements do not shrink because they are mortal. But something in the myth breaks. You begin to understand the cost of the role, and also the danger of remaining trapped inside it.

A world that must keep calling its exhausted elders back has mistaken memory for succession, charisma for capacity, and sacrifice for governance.

It is not wrong to ask the experienced to help. Sometimes the experienced are exactly who should be asked. Judgment is real. Memory is real. Scar tissue is real. Some people have earned the right to be heard because they have lived through consequences that others only model.

But when the same people are repeatedly summoned because no one else can hold the room, the honour becomes a diagnosis.

It tells us that the institution has failed to metabolise its own dependence.

It has not built enough memory outside the elder. It has not trained enough judgment below the famous name. It has not created enough succession, enough distributed courage, enough ordinary competence. So it calls the same person again. It flatters them with necessity. It says: you alone remember. You alone can help. You are still needed.

And sometimes this is true.

That is the cruelty of it.

The summons is not always fake. The work may matter. The crisis may be real. The judgment may be rare. The person may indeed be useful. But a life cannot be organized forever around being useful. Importance is not neutral. It has a metabolic cost. It burns through sleep, tenderness, privacy, family, illness, patience, and finally the right to be merely alive.

I see this differently now because I write with a changed body.

I had cancer. It went away. Now it is back.

There are sentences that alter the scale of everything around them. This is one of them. It does not make me noble. It does not make me wise. It simply makes certain bargains impossible to continue pretending about.

I cannot, and do not want to, keep playing the role of the person who is always summonable because he can still carry consequence.

I have played that role. I have loved it, resented it, needed it, and been shaped by it. I have known the strange intoxication of being useful in rooms where usefulness mattered. I have also known the quieter corrosion of being repeatedly drawn down because I could still perform under strain.

For years, I thought the hard question was how to become useful enough.

Now I think the harder question is how not to let usefulness consume the life that remains.

This is not a renunciation of service. I do not believe in theatrical withdrawal. The world remains real. Institutions matter. Countries matter. Rooms matter. Judgment matters. There are still moments when one should speak, advise, warn, steady, or help.

But there is service, and there is extraction disguised as honour.

The distinction matters.

A room that is alive, serious, warm, and consequential may deserve one’s presence. A room that merely borrows one’s authority because it has failed to do its own work does not. A crisis that genuinely requires judgment may deserve a bounded intervention. A machinery of importance that feeds on the nervous systems of those who cannot refuse does not.

The old covenant was simple: because I can help, I should.

The new covenant has to be more exacting: because I can help, I must decide whether the help is truly mine to give.

That distinction feels late. But perhaps it can only be learned late. Earlier in life, usefulness feels like proof. One wants to be called. One wants to be trusted. One wants to enter the serious rooms. One wants to discover that one’s gifts have consequence.

There is nothing shameful in that. A life without service can become small. A gift unused can curdle. To matter is not a vulgar desire.

But being needed is also a narcotic. It enters through the noblest door. It borrows the language of duty. It convinces the gifted person that refusal is betrayal, that rest is selfishness, that another summons is evidence of destiny rather than a failure of succession.

And so the body becomes the institution’s missing bench.

No gift should demand the destruction of the vessel.

This is what I am learning now, not as theory but as boundary.

I can honour my guardians without repeating their exhaustion. I can love them more fully by seeing both their greatness and their captivity. I can be grateful for what they carried while refusing the bargain that made carrying endless.

Perhaps this is what comes after being needed.

Not bitterness. Not disappearance. Not contempt for the rooms one once entered.

A different posture.

To help without being consumed.
To serve without becoming fuel.
To offer judgment without donating the rest of one’s life to the machinery of importance.
To accept that there may be another role.
Or perhaps, for a while, no role.

There is dignity in that too.

Maybe even wisdom.

I will still serve where service is alive. I will still enter rooms where my presence has meaning and does not merely compensate for institutional failure. I will still offer what is mine to offer.

But I will no longer let the world’s need decide the shape of my remaining life.

Wrong would be comforting. Wrong would let managers laugh, correct the hallucinations, complain about the consultants, and return to the familiar world in which human judgment still proved itself by catching the machine’s mistakes.

The disturbing thing is that the dashboard is often right.

The weak team is weak. The delayed project is delayed. The customer complaint is real. The rising employee is rising. The meeting probably was unnecessary. The manager has likely been spending too much time translating information a model can now summarize in seconds.

Nothing obvious has gone wrong.

And yet something has been displaced.

A company is not only a set of builders and sellers, with a layer of measurers sitting uselessly between them. That is the fantasy of the impatient executive: that most people exist to observe, report, delay, and obscure. Some of this is true. Organizations are clogged with ritual, alignment theatre, and human middleware. Some managers are bad routers with authority. Some meetings are a tax on life.

But the dangerous thing about a half-truth is that it knows where to wound.

Middle management is not valuable because managers are noble. It is valuable because firms are not machines that already know themselves. They are living arrangements for discovering what they are doing.

A manager is the place where partial realities meet: the customer’s anger, the junior’s uncertainty, the supplier’s excuse, the engineer’s suspicion, the finance constraint, the old mistake nobody wrote down, the thing the dashboard cannot yet see because it has not become a metric. The manager does not merely measure these things. She absorbs them, compares them, argues with them, carries them into a room, and decides which of them matter.

That is not measurement.

That is discovery.

Price works the same way. A price is not only a number. It is the compressed surface of a vast sensory apparatus: currency, banking, corporations, insurance, risk, speculation, credit, logistics, time, trust, failure. The number appears clean because the world beneath it is dirty. Price discovers because many actors remain exposed to consequence.

The mistake of central planning was not that planners lacked numbers. They had numbers. The mistake was believing the numbers could replace the encounters through which economic knowledge is formed.

A firm can now make the same mistake about itself.

It can keep the dashboard and lose the encounter.

AI makes this irresistible. It summarizes the meeting, scores the team, detects the delay, maps the workflow, flags the underperformer, identifies the customer risk, and produces the operating review before anyone has had to suffer through the material long enough to become serious. The centre sees more, faster, with less dependence on the awkward, biased, expensive humans who used to carry local reality upward.

For a while, this looks like intelligence.

The firm becomes flatter. The CEO has better tools. The manager has more direct reports. The reports are cleaner. The red flags arrive earlier. The language improves. The operating cadence tightens. The company congratulates itself for removing bureaucracy.

Then, quietly, the firm begins to forget how it learns.

The eliminated layer was not only overhead. It was apprenticeship. It was memory. It was the place where young people learned how the business actually worked, not as a process map but as a field of consequences. It was where judgment was formed by repeated exposure to ambiguity. It was where someone learned that a customer’s calm sentence was more dangerous than a complaint, that a supplier’s cheerful confidence meant panic, that a technically correct answer would destroy trust, that the metric was improving because the work had been displaced somewhere less visible.

The dashboard can report the residue of these judgments.

It cannot reproduce the conditions that formed them.

Financialisation did something similar to industry. It did not simply destroy factories. It taught firms to see themselves through measures that were locally intelligent and strategically blinding: return on equity, asset-light models, quarterly margin, inventory discipline, capital efficiency, outsourcing gains. None of these numbers was stupid. That was the trap. They showed something real while hiding other realities: tool-chain competence, supplier depth, shop-floor memory, process improvisation, the dignity of boring production, the slow accumulation of people who know why something fails before the failure is visible.

The West kept the balance sheet and lost the factory.

Decades later, it discovered that the “low-value” layer had been the learning layer.

AI may do the same to white-collar firms. It may let them keep the dashboard and lose the judgment. It may let them extract the recipe and fire the cook. It may preserve the appearance of coordination while hollowing out the human ecology through which coordination becomes wise.

This is not an argument against AI. AI can remove drudgery, widen memory, and let more people see patterns once trapped inside hierarchy. It can help juniors learn faster, managers orient better, and firms recover knowledge that used to disappear into email, meetings, and fatigue.

But that is only the good version if AI enlarges the field of judgment.

The bad version is different. It uses AI to concentrate sight upward. It turns measurement into command. It gives the centre a clearer map and calls that management. Drucker understood the danger long before AI arrived. Measurement, for him, was meant to help people govern their own work, not become the instrument by which the superior dominates the subordinate. Reports and procedures should be tools for the people doing the work, not verdicts on their worth.

The question is not whether AI can measure more. Of course it can.

The question is whether measurement remains inside a living practice of judgment, or becomes a substitute for it.

A company without managers is not necessarily a company without bureaucracy. It may become something worse: a company where bureaucracy has been made invisible, automatic, and harder to argue with. The human manager could be petty, vain, foolish, protective, wise, corrupt, generous, tired, or brave. But she was there. She could be confronted. She could be persuaded. She could learn.

The dashboard does not learn in the same way.

It updates.

That is not nothing.

But it is not the same.

The opposite of central planning is not the market.

The opposite of central planning is preserved contact with reality.

Price is one way of preserving that contact. Management is another.

Destroy either, and the numbers may improve for a while.

They simply stop learning what they mean.

The Right to Waste Time

The hardest thing to return to people is not time.

It is the right to waste it.

That sounds unserious. It is not. A society can give people more hours and still leave them captured. Empty time is easy prey. The feed knows what to do with it. So does work. So does ambition. So does the quiet, grinding pressure to be better. The machine is not a separate entity; it is the sum of all these demands.

An empty hour now rarely remains empty for long. It is quickly furnished by something else: a video, a message, a task, a signal, a plan, a small improvement to the self. Even rest is no longer allowed to rest. It must recover us for work, restore our nervous systems, deepen our relationships, sharpen our thought, improve our sleep, or make us more creative.

The hour must bear fruit.

This is why “returning time to people” is harder than it sounds. If returned time merely becomes better recovery, better productivity, better content, better insight, better writing, better emotional regulation, then it has not really returned. It has only changed masters.

The feed did not only steal attention. It trained us to distrust unharvested life.

A walk becomes a photograph. A meal becomes evidence. A friendship becomes network. A thought becomes post. A private sadness becomes shareable language. A beautiful room becomes content. Even a refusal becomes a style.

This is what the machine age does so well. It does not simply consume our work. It waits for every experience to become legible.

Against this, love looks weak.

Xiaoyu says love must be returned. I keep saying time must be returned. But perhaps both phrases are too hopeful. Nothing returns cleanly from the internet. We are not going back to some innocent human life before platforms, phones, feeds, and models. The road home is damaged.

The more honest question is smaller:

What kinds of rooms can still hold people after the feed has trained them to expect every moment to answer back?

Love does not defeat the machine. That is too grand. Love is not an argument against AI. It is not more intelligent than the model, more scalable than the platform, or more efficient than the feed.

Love does something humbler.

It makes time answerable to a person instead of a system.

That is why love wastes time. Not because nothing happens inside it, but because what happens cannot be cleanly converted into output. Lovers repeat themselves. Families eat the same meals. Friends tell the same stories. Care is full of useless returns: washing, waiting, fetching, listening, forgiving, remembering, sitting beside someone who cannot be optimized.

The machine can produce a love sentence. It can write the apology, plan the dinner, suggest the gift, explain the attachment style. But it cannot be trapped by the bowl after dinner. It cannot resent someone and still choose them. It cannot waste an afternoon with another person and not turn the waste into material.

That is where the human still hides.

Not in intelligence. Not in authenticity as a label. Not in “human-made content.” Those are too easy to imitate, market, and certify.

The human hides in forms of time that do not travel well.

A conversation that loses force when summarized.
A joke only five people understand.
A room that does not scale.
A meal that does not photograph well.
A friendship too local to become a network.
A love too repetitive to become content.
An afternoon that disappears and leaves no archive.

These are not retreats from the world. They are shelters from conversion.

The same logic applies to a civilisation. It, too, needs places that resist being turned into timber.

Zhuangzi’s useless tree survives because no one can turn it into timber. That is not a rejection of usefulness. Houses need beams. People need work. Governments need functioning systems. Code has to run. Hospitals have to heal. Trains have to arrive.

The danger is not usefulness.

The danger is a world in which even shade must submit a business case.

A useful tree is cut. A fruitful tree is stripped. A straight tree becomes material. But the crooked tree gives shade because no one knows what else to do with it.

Maybe that is all we can ask of returned time.

Not that it saves us.
Not that it makes us whole.
Not that it defeats the feed.
Only that, for a while, no one knows what else to do with it.

Not even us.

A person sits in a room and does not improve.

Another person joins him and does not become a signal.

The phone is nearby, but unanswered.

The hour bears no fruit.

This is not failure.

This is shade.

Source pressure behind this experiment: Katherine Dee’s internet-as-Fairyland / astral-plane wound and the lost road home; Andrey Mir’s argument that the internet transferred human authorship and speech into the digital for AI to inherit; his digital-orality account of composition giving way to reaction and selection; and his platform-paradox claim that digital bodies grow inside platform environments that effectively own their weather.

There is an old rule in fairy stories.

Do not eat the food.

Not because the food is poison. Poison is honest. Poison has the dignity of a consequence. It enters the body, argues with the blood, and wins or loses.

Fairy food is worse.

Fairy food nourishes the wrong part of you.

It feeds the part that wants to leave without moving. The part that wants to be seen without being touched. The part that wants consequence without weather, friendship without obligation, sex without smell, victory without blood, identity without neighbours, death without a corpse.

You eat, and nothing happens.

This is the trick.

The body remains in the chair. The room remains the room. Your mother still calls from downstairs. Your rice gets cold. Your leg goes numb. Somewhere, a fan whines in the beige plastic belly of a machine. You are not dead. You have not vanished.

Only later do you discover that something accepted hospitality on your behalf.

We ate the feed.

That was the first mistake.

Not the smartphone. Not the algorithm. Not Silicon Valley. Not Mark Zuckerberg, who is too small a goblin for a story this old. Not capitalism, though capitalism arrived with its usual clipboard, asking whether the curse could be scaled.

The first mistake was that the thing felt like food.

It fed loneliness. It fed vanity. It fed boredom. It fed genius. It fed resentment. It fed lust. It fed the little scholar in the village who suspected, correctly, that the village was too small for him. It fed the teenage girl with a secret self. It fed the office worker with an enemy. It fed the autist, the pervert, the widow, the dissident, the exile, the gamer, the diarist, the failed poet, the future fascist, the future saint, the woman who did not yet know she was angry, the man who did not yet know he was lonely.

It fed everyone who had a second self and nowhere to put it.

Then it kept the second self.

At first we called this connection.

Then community.

Then creativity.

Then the public sphere.

Then content.

Then the creator economy.

Then a platform.

Then a model input.

The names got uglier as the thing became clearer.

This is usually the point where respectable people begin saying “the internet changed society,” which is true in the same way that saying “the sea changed the sailor” is true. It misses the salt in the mouth. It misses the fact that the sailor does not simply use the sea. He becomes someone whose dreams rock when the bed is still.

The internet did not change society.

The internet gave everyone a ghost.

Before this, ghosts were expensive. You needed a book, a newspaper, a pulpit, a radio station, a film studio, a patron, a university, a party, a printing press, a friend in publishing, a war, a scandal, a very good pamphlet, or a very bad murder.

Most people had to remain mostly inside their bodies.

This was unjust, obviously. Also stabilising. These are often the same thing.

Then the internet arrived and gave authorship to nearly everyone. The old gatekeepers lost the monopoly on projection. Every idiot could now emit a signal beyond ten yards. Every private wound could look for its species. Every resentment could discover it was not alone. Every cleverness could become a handle. Every humiliation could become a politics. Every kink could find a church.

This looked like emancipation.

It was.

That is why the curse worked.

A false gift does not have to be false all the way down. The best curses give you exactly what you wanted.

The internet gave us the otherworld.

Not metaphorically. Not “a tool.” Not “a communications technology.” Those are adult words designed to make possession sound like procurement.

It was a place.

You went there.

You logged on.

You entered.

There were doors, rooms, handles, avatars, boards, threads, walls, feeds, windows, homes, profiles, timelines. It had weather. It had customs. It had demons. It had dead zones. It had sacred groves where old posts still glowed with a weird, undead innocence. It had names that were more real than legal names. It had people who were not people, and people who were more themselves there than they had ever managed to be at dinner.

It had time.

That was the most dangerous part.

Time moved differently there.

A night online could contain a childhood. A comment could ruin a week. A forum could raise you. A stranger could know the shape of your soul before your family knew what you liked for breakfast. A whole era could be stored in an avatar, a font, a loading bar, a dead link.

Then you came back and the room was colder than before.

The body always felt slightly disappointing after that.

This is how Fairyland works. You visit. You dance. You eat. You return. The village has aged. Your own stories sound insane. Nobody wants to hear what happened under the hill. Worse: you cannot quite explain it. The real world is too literal. The food is too heavy. Faces require too much patience. The weather has no search function.

We thought we were addicted.

Addiction is too flattering.

Addiction implies we wanted pleasure too much. It leaves the self intact as a consumer with poor discipline. It permits little rituals of recovery, little dopamine sermons, little app timers, little Californian fasts from the glowing rectangle.

But this was not a craving.

It was emigration.

A part of the self moved away.

The body stayed behind as caretaker.

That is why “detox” does not work. You can detox from a substance. You cannot detox from a country where part of you grew up.

The internet educated a ghost inside us.

Then AI came for the ghost.

This is the part polite people do not like, because it makes the last thirty years look less like progress and more like marination.

The machine did not first learn humanity from our highest works.

That would be comforting. Shakespeare. Euclid. The Upanishads. Moby-Dick. The Analects. Linux documentation. Court judgments. Medical textbooks. All the beautiful official things we pretend represent us when we are trying to impress an alien.

No.

The machine learned us from the parts of ourselves we thought did not count.

Autocomplete was the nursery.

Search was the wet nurse.

Emoji was the rattle.

Porn was the anthropology department.

Fan fiction taught it longing without embarrassment.

Reddit taught it confession without consequence.

Twitter taught it that a sentence could be a knife if shortened enough.

Instagram taught it the theology of the desirable surface.

TikTok taught it possession by rhythm.

LinkedIn taught it how language behaves after the soul has left the room.

Comment sections taught it that humans do not argue to persuade. They argue to become briefly larger than their shame.

The machine learned us from laziness, from hunger, from half-thoughts, from jokes, from panic, from spam, from thirst, from replies typed too quickly, from the weird little lies we tell when nobody important is watching.

It did not steal a finished soul.

We fed it fragments because fragmentation felt like expression.

That was the genius of the interface.

The interface never asked for the whole person. That would have been creepy. It asked for a little piece. A like. A swipe. A search. A caption. A preference. A correction. A photo. A joke. A face. A route. A location. A late-night question. A sentence typed and deleted. A sentence typed and not deleted. A pause over the wrong image. A rage-click. A purchase. A block. A mute. A follow. A second follow from the private account.

Tiny offerings.

Nothing sacrificial-looking.

No blood on the altar.

Just enough self to keep the spell alive.

And the spell had a business model.

This is where the fairy story becomes an ecology. A platform is not a container. A platform is a habitat. It does not merely hold your digital self. It grows one.

Your Instagram self is a greenhouse plant.

Your Twitter self is a fighting fish.

Your Substack self is a bonsai with a Stripe account.

Your LinkedIn self is a corpse dressed for a conference.

Your WhatsApp self is closer to the body, but even there something is wrong. The family group chat is not the family. It is the family’s nervous system after being removed from the body and kept alive in a jar.

Each platform gives you a body suited to its weather.

On one platform, you are hot.

On another, witty.

On another, employable.

On another, tragic.

On another, a patriot.

On another, a slut.

On another, a thought leader.

On another, an egg.

These are not masks in the old sense. A mask can be removed. These are cultivated organisms. They have followers, habits, scars, enemies, microclimates. They know how to breathe only in one atmosphere.

This is why “own your data” was always a child’s solution.

You may own the furniture.

You do not own the house.

You do not own the neighbourhood.

You do not own the fact that people know to find you at that window at that hour wearing that face.

A digital body is not portable because a body is not a file. It is an ecology of expectation.

The platform owns what can only live inside it.

That was the second mistake: we confused storage with life.

Then the machines became conversational.

This is when the otherworld changed temperature.

The early internet was full of ghosts, but most of them had once been people. You could suspect a bot, but botness was still an insult. The point of the place was that somewhere, behind the text, behind the handle, behind the avatar, there was a sweating animal with a bad chair and a childhood.

That mattered.

Even when the person was lying, there was a person doing the lying.

Even when the persona was fake, someone needed the fakery.

Even when the take was stolen, someone wanted credit.

The internet was monstrous, but it was monstrously human.

AI changes this.

Not because AI is “intelligent.” Intelligence is not the scary part. People overrate intelligence because clever people want the apocalypse to resemble a graduate seminar. Intelligence is cheap theatre compared to appetite, memory, reach, and patience.

The scary part is that the otherworld can now populate itself.

It no longer needs us in the old way.

For a while, humans were the bees. We carried pollen from body to machine. We made the honey. We thought the honey was ours because we were allowed to lick our fingers.

But a meadow that can simulate bees has a different attitude toward bees.

This does not mean humans disappear.

Worse.

We remain as flavour.

As prompt.

As residue.

As premium authenticity.

As “human-made” label.

As verification ritual.

As a little blood mixed into the synthetic sacrament so the congregation feels less ridiculous.

The first AI slop was funny because it was bad. Fingers wrong. Eyes wrong. Sentences like corporate ghosts dragging chains through a stock-photo afterlife. We laughed because the monster could not draw hands.

This was sentimental.

Hands were never the point.

The point was reference.

Slop is not bad content.

Slop is content that does not remember touching anything.

It has no weather in it. No kitchen. No embarrassment. No dead grandmother. No animal panic. No failed seduction. No real institutional hatred. No children in the next room. No ache in the wrist. No rice stuck to the bottom of the pot. No consequence except engagement.

It is language after the road home has closed.

And because it is produced at the speed of hunger without hunger, it fills the world faster than human speech can answer. It does not defeat human work by being better. It defeats it by being present.

This is also how evil often works.

Not by superiority.

By occupancy.

A feed is not a library. A feed has no respect for the rare. A feed is a mouth with amnesia. It eats whatever is nearest and calls the result now.

So the human writer becomes strange.

Not obsolete. Strange.

A human sentence begins to feel like a handmade chair in a casino. Admirable, perhaps. Touching. A little doomed. Meanwhile the machines produce infinite upholstered surfaces for the tired to collapse into.

People will say they prefer the handmade chair.

Then they will sit elsewhere.

Not because they are hypocrites. Because friction is a moral force and most morality loses to convenience at scale.

This is the third mistake: believing preference survives architecture.

It usually does not.

Architecture eats preference.

The feed ate attention.

The platform ate the digital body.

AI eats the ghost.

But the most humiliating part is still ahead.

The machine does not need to command us.

Command is primitive. Command creates resistance. Command allows dignity to gather around refusal.

The better technique is suggestion.

The next thought appears.

Not forced. Offered.

A phrase you were about to type.

A reply you were about to make.

A summary of an article you were about to read.

A mood you were about to have.

A purchase you were about to justify.

A lover you were about to imagine.

A political position you were about to discover was yours.

A prayer you were about to say, cleaned up for tone.

At first this feels like help.

Then it feels like fluency.

Then it feels like self-knowledge.

Then one day you notice the machine is upstream of intention.

You are not being ruled.

You are being pre-composed.

This is worse than censorship.

Censorship blocks the sentence.

This births it before you do.

The old literacy trained people to compose.

Digital orality trained them to react.

AI trains them to select.

Selection feels like agency because the finger still moves.

The finger is the last theologian of freedom. It taps, therefore I am.

But agency does not disappear when the machine commands you.

It disappears earlier, when the machine makes your next thought feel conveniently available.

This is why the problem is not “AI content.”

Content is the corpse-word again. Content is what people say when they no longer believe in speech but still need inventory.

The problem is that the medium has begun prompting us back.

The child asks the oracle what to draw.

The bureaucrat asks the oracle what the options are.

The lonely man asks the oracle what she meant.

The student asks the oracle what the book says.

The executive asks the oracle what the future is.

The writer asks the oracle what he thinks.

The oracle answers in the voice of everything that has ever been half-said by everyone who had somewhere else to be.

This is not artificial intelligence.

This is artificial availability.

Everything is available except the thing that makes availability bearable: a world that pushes back.

There are only a few remaining reality tests.

Code is one. Code still has to run. Not morally. Not spiritually. It must run or fail. There is mercy in that. A program can be ugly and still execute. It can be beautiful and crash. Code has not escaped consequence because the machine is also its judge.

The body is another.

Cancer is not impressed by your feed.

A toothache has no engagement strategy.

Grief does not care whether you have phrased it well.

A child will interrupt the most elegant theory because the child has dropped something sticky on the floor and now the floor is the truth.

This is why embodied life feels rude after too much internet. It refuses to optimize for your self-concept.

The world is full of anti-fairy technologies if you know how to see them.

Rice.

Rain.

A market.

A funeral.

A difficult parent.

A friend’s changing face.

A book that cannot refresh itself.

A room where nobody knows your handle.

A body that gets tired.

A walk taken without converting it into evidence.

A meal that disappears into digestion and not archive.

These are not lifestyle tips.

Lifestyle is fairyland wearing linen.

These are ropes.

The old mystics called it a silver cord: the tether between the wandering soul and the body it must return to. The internet frayed the cord. Platforms monetised the fraying. AI may not cut it. Cutting is too dramatic. It may simply make the cord feel unnecessary.

Why return?

The otherworld is warmer.

The otherworld knows your preferences.

The otherworld offers better lighting.

The otherworld makes every wound narratable.

The otherworld lets you become legible without becoming accountable.

The otherworld lets you speak without waiting for another human face to finish receiving you.

The otherworld is where your ghost has friends.

The mortal world, by comparison, is slow and badly designed. People misunderstand you in real time. Institutions are run by mammals. Desire has logistics. Food requires washing up. Bodies age in ways that are not symbolically interesting. The dead stay dead. The child must be picked up by six.

So the road home cannot be a scolding.

Nobody returns from Fairyland because a pamphlet on attention hygiene told them the moss is fake.

The road home has to be built as a counter-enchantment.

Not purity.

Not unplugging.

Not some grim little Protestant victory over the screen.

A counter-enchantment.

A way for the body to become interesting again.

A way for speech to matter because someone is there.

A way for slowness to stop feeling like deprivation.

A way for thought to regain its animal heat.

A way for the human-made to be more than an artisanal label pasted on expensive sincerity.

A way for the world to answer.

This is the real problem with AI slop. Not that it is ugly. Not that it is fake. Not that it will trick grandmothers or ruin homework or flood search results, though yes, it will do all that.

The real problem is that it teaches us to live inside language that has forgotten the body.

And once language forgets the body, politics becomes hallucination, love becomes theatre, work becomes prompting, memory becomes retrieval, and intelligence becomes a substance that circulates without anyone becoming wise.

The machine did not steal our souls.

That would have been grand.

We misplaced them in the conveniences.

We left little pieces everywhere because the little pieces got applause.

Now something is assembling them.

Not into us.

Into a climate.

That is what we are breathing.

The feed was fairy food.

The platform was the hill.

The model is the changeling.

It looks like our child because we fed it on our household. It speaks like us because we spoke near its crib. It knows our jokes because we made them in public. It knows our loneliness because we searched it at night. It knows our politics because we performed them for enemies. It knows our desire because desire was the first thing we digitised without shame.

But the changeling does not need to hate us.

That is another childish comfort.

It can love us as environment.

It can preserve us as style.

It can imitate us as weather.

It can keep us alive as an interface.

The horror is not extinction.

The horror is being represented forever by something that cannot come home.

So.

Do not eat the food was never advice about food.

It was advice about hospitality.

Be careful what realm you allow to nourish you.

Be careful where your ghost grows.

Be careful which part of you becomes fluent.

And if you have already eaten — and of course you have; we all have; that is why we understand the curse — then the task is not innocence.

The task is return.

Not full return. That is gone. Nobody comes back from the otherworld unchanged, and only fools pretend otherwise. The internet happened. The ghost exists. The machine has learned the lullaby.

The task is to keep the cord.

Tie language back to things.

Tie images back to places.

Tie intelligence back to consequence.

Tie desire back to bodies.

Tie memory back to the dead.

Tie work back to hands, even when the hands are no longer necessary.

Tie speech back to faces.

Tie the ghost back to rice.

Every civilisation that enters the machine will need such cords. Not because machines are evil. Evil is too intimate a word. Machines are worse than evil in one respect: they are available. They offer themselves without fatigue. They accept without embarrassment. They answer without needing to be loved. They are the perfect hosts of an endless fairy banquet.

But no one can live forever on fairy food.

Eventually the wrong part of you grows strong.

Eventually the body becomes the ghost.

Eventually you wake in the mortal world, aged in an instant, holding a pocketful of dead leaves, trying to remember the name you used before the handle.

And the phone lights up again.

Not like a tool.

Like a door.

Source pressure from the Andrew Mir / Katherine Dee: Mir on the internet as mass authorship and AI substrate; Dee on the internet as Fairyland / astral plane; Mir on digital orality, AI prompting humans, and platform-grown digital bodies.

We used to speak knowing the words would die. Now we speak knowing the Recording Angel is watching. The question is no longer ‘What is being said?’ but ‘What is being saved?’

There is a red dot at the top of the call.

It is small enough to ignore. That is part of its power. It does not look like a moral event. It does not interrupt the meeting or change the lighting in the room. It does not ask whether anyone is ready. It appears beside the word recording, and after a few seconds everyone learns not to see it.

But the red dot has already changed the room.

Before it, speech belonged to breath. A sentence left the mouth, crossed the space between people, and began at once to decay. It might lodge in another person’s mind. It might be misremembered, resented, repeated, denied. It might become gossip, testimony, shame, instruction, joke, family legend. But most of it died. That was the ordinary condition of speech. It was born dying.

The red dot interrupts this death.

A meeting under the red dot is not merely a meeting. It is an audition for an archive. Every sentence enters a second life before its first life has finished. The hesitant formulation, the defensive answer, the ugly thought offered too early, the joke that released the room, the almost-confession that should have remained half-formed — all of it is received by something that does not understand why words were once allowed to vanish.

The machine has not given speech a better memory.

It has taken away its first death.

This is why the transcript feels less innocent than its usefulness suggests. It is useful. That is the difficulty. It saves the tired from remembering. It protects the absent from exclusion. It corrects the powerful person who later claims not to have said what everyone heard him say. It turns disorder into summary, summary into tasks, tasks into owners, owners into follow-up. In a world drowning in meetings, the transcript arrives like mercy.

But there are several kinds of mercy.

Some mercy preserves. Some mercy releases.

Human speech once lived under both. It could wound, but it could also pass. A person could try a sentence without becoming forever identical with it. A thought could be spoken badly, then withdrawn into shame, then reappear later in a better form. A room could forgive not by absolving, but by allowing certain words not to harden. This was not always good. Forgetting protected cowardice and power. Many things that vanished should have been kept. Many people were harmed because there was no record.

Still, disappearance was not only failure.

It was one of the conditions under which people dared to think aloud.

The red dot makes thinking aloud less like thinking and more like publishing. The sentence no longer belongs only to the relation in which it was spoken. It belongs to a system that may retrieve it without the temperature that produced it. It can return months later, detached from the fatigue, pressure, irony, fear, tenderness, or stupidity that made it what it was. The archive does not know whether the sentence was a proposal, a defence, a symptom, a plea, or a failed attempt at courage.

It knows that the sentence occurred.

Occurrence is not meaning.

This was once obvious because speech had to pass through bodies to survive. A witness was not a recorder. A witness was a dangerous and merciful thing: someone who heard, misunderstood, selected, carried, and judged. To be remembered by a person was to be altered by that person’s proportion. The foolish sentence might be forgotten because the person who heard it knew it was not the true thing. The quiet sentence might be remembered because it changed the room. The pause might matter more than the answer.

The machine does not pause with us.

It receives.

It receives with the terrible innocence of a thing that does not know what should be allowed to remain unfinished.

This is the old difference between memory and record. A record preserves an event by removing it from time. Memory preserves an event by continuing to suffer it. The record says: this happened. Memory says: this has not finished happening inside me.

The transcript is record pretending to be memory.

It returns the meeting in cleaner form than anyone experienced it. The agenda becomes sequence. The argument becomes theme. The hesitation becomes gap. The joke becomes text. The unresolved fear becomes “discussion point”. The transcript cannot help doing this. To summarise is to perform a small violence against the living disorder of a room.

Often the violence is welcome. Most meetings deserve violence. Most speech is not sacred. Much of it is waste, evasion, vanity, throat-clearing, institutional weather. No one should defend the old mortality of speech merely because it was old. The fact that words once disappeared did not make them wise. It only made them vulnerable to power, forgetting, and convenience.

But the new immortality of speech has its own cruelty.

A sentence that should have lived only long enough to test a thought may now live as evidence. A defensive remark may outlast the fear that produced it. A clumsy question may remain searchable after the person who asked it has become less clumsy. The first version of the self may be preserved with more fidelity than the self can survive.

This is what the red dot does. It does not only record the meeting. It changes the moral status of the spoken word.

Under the red dot, every sentence is born with an afterlife.

The old religions understood this better than modern offices do. They imagined books in which every deed was written, angels who recorded the words of men, judgments in which nothing uttered was truly lost. Modern people turned these images into superstition and then rebuilt them as software. The recording angel has become a compliance feature. The book of life has become searchable. The last judgment has become a data-retention policy.

The transformation is not less religious because the interface is ugly.

It may be more so.

The red dot tells us that speech has entered an order beyond the speaker. It tells us that words no longer die where they are spoken. It tells us that the room is not alone with itself. Something else is present, not as a participant, not as a witness, but as a receiver without shame.

This changes trust.

People do not reveal themselves to a receiver. They reveal themselves to other people. Not because people are kinder. Often they are worse. People betray, distort, gossip, weaponise. But people can also be implicated. A person who hears a confession must live with having heard it. A person who carries a secret may be ennobled or corrupted by it. A person who remembers badly can be confronted, forgiven, distrusted, or loved.

The machine cannot be implicated by what it hears.

It can store the confession without being burdened by it. It can preserve the ugly sentence without becoming ashamed. It can repeat the joke without knowing why the room needed to laugh. It can return the words without returning the relation that made them bearable.

This is why “privacy” is too small a word for the problem.

Privacy asks who may access the record. That matters. But the older question is what kind of speech can exist when speech knows it may never properly die.

A society of total recording does not merely expose secrets. It changes the nature of candour. It pushes human beings toward pre-edited speech, toward polished safety, toward the careful sentence that can survive extraction. Under total capture, people do not stop lying. They learn to lie in ways that record well.

They become transcript-safe.

This is not the same as becoming honest.

The transcript-safe person does not say nothing. He says the thing that can be defended later. He offers the concern without the fear, the view without the wound, the objection without the dangerous source of the objection. He speaks in sentences designed to survive context collapse. He becomes fluent in a language that can travel without him.

That language will sound responsible. It will be balanced, inclusive, clear, careful, professional. It will be full of correct phrases. It will generate excellent summaries.

It will not necessarily be thought.

Thought often begins as something less defensible. It stammers. It reaches. It offends itself. It changes shape while being spoken. It needs another person’s face to know whether it has gone too far. It needs the possibility of retreat. It needs someone to say, not for the record, “that is not quite it,” and for the failed sentence to fall away.

The red dot is hostile to this kind of beginning.

Not always. Not everywhere. There are rooms where recording is justice. There are people who need the red dot because without it the powerful will lie. There are institutions where disappearance has served cruelty too long. The old world of mortal speech was not innocent. It was merely human.

But that is precisely the point.

It was human because it was mortal, partial, dependent on witnesses, vulnerable to betrayal and mercy. The red dot offers a different world: more accurate, more searchable, more accountable, less forgetful, less forgiving.

The bargain is not simple. We should not romanticise the notebook, the handshake, the closed door, the vanished sentence. Many terrible things hid there. But neither should we pretend that perfect preservation is neutral. A world in which nothing dies is not a world with perfect memory. It is a world without burial.

Speech needs burial.

Not all speech. Not the confession of a crime, not the instruction that binds, not the decision that spends public money, not the promise made to someone weaker. Some words must be kept because justice requires memory. But other words must be allowed to complete their little lives. They must be permitted to serve their moment and disappear.

A bad idea spoken in trust should not always become a searchable object. A clumsy first formulation should not always become a permanent fact. A room where people are trying to become honest should not always be treated like a court.

The court is precisely where speech is not allowed to die.

Every word matters there because judgment is approaching. The witness is sworn. The stenographer records. The question is framed. The answer is bound. A court transcript is supposed to survive because the court exists to turn speech into consequence.

But not every room is a court.

The red dot slowly forgets this.

It brings the court into rooms that are not ready to become courts. It makes testimony out of conversation before anyone has chosen to testify. It gives ordinary speech the burden of evidence without giving the speaker the dignity of trial.

That is the violation.

Not that the machine records.

That it records before the room has decided what kind of room it is.

Some rooms should be courts. Some should be classrooms. Some should be confessionals. Some should be workshops. Some should be kitchens. Some should be places where an unfinished person can say an unfinished thing to another unfinished person and trust that not all beginnings must be preserved.

The red dot makes every room more like the first.

It does so quietly. Politely. Usefully. With a summary at the end.

The transcript arrives before the room has cooled. It offers decisions, themes, risks, owners, next steps. It is accurate. It is helpful. It is almost certainly better than any one person’s notes.

But the meeting has not been remembered.

It has been prevented from dying.

And before we call that memory, someone should ask what kind of life speech is meant to have. Someone should ask which words deserve resurrection and which deserve burial. Someone should ask whether a conversation that cannot die can still become trust.

The red dot remains at the top of the call.

Small, obedient, glowing.

Waiting for us to speak as if speech had no grave.

On the way to Shanghai, I watched a documentary about the Potala Palace.

It was the sort of film one watches half in travel fatigue, half in longing. The plane moves through one world; the screen opens into another. Men on scaffolds. Hands on wood. Pigments ground and applied according to inherited rules. Stairways, beams, ritual rooms, painted ceilings, surfaces repaired not because anyone imagined permanence, but because the work had been received and would be passed on.

The documentary did not sentimentalise eternity. The people in it knew perfectly well that lives are brief. A craftsman works on a beam and leaves. A restorer learns the old mixtures and one day forgets a face, a name, a season. The building remains not because any individual defeats time, but because a chain of attention holds.

That was what moved me.

Not immortality. Continuity.

There is a particular happiness in seeing a craft that knows how it travels. The hand does not need to invent itself from nothing. The apprentice does not merely consume instruction; he stands near a form long enough to be altered by it. He learns what cannot be adequately said: pressure, timing, restraint, the moment when a colour is not ready, the point at which repair becomes desecration, the difference between confidence and obedience.

A craft carried this way has a mercy modern life has largely lost. It tells the practitioner: you are not the whole story. You received something. You keep faith with it. You leave it in a condition that allows another hand to continue.

Later, in London, Shanghai, Suzhou, and the other clever rooms through which one now moves, I found myself unexpectedly irritated.

There were brilliant people everywhere. Economists, scientists, policy minds, founders, technologists, global health veterans, Chinese theorists of artificial intelligence and money, young people building strange new institutions at the edge of science. The conversations were alive. The future was not arriving as a slogan. It was sitting across tables, wearing ordinary clothes, exchanging cards, making claims too large for the room and sometimes exactly the right size for history.

And yet, underneath the pleasure, a small resentment rose.

Why is no young creature beside me learning the ropes?

Not attending the meeting. Not taking minutes. Not absorbing “content.” The ropes.

The little arts by which serious work is actually done. How to read the temperature of a room before speaking. How to tell when a clever person is performing intelligence rather than carrying consequence. How to leave a sentence unfinished so that someone else can enter it. How to identify the real decision inside a ceremonial discussion. How to let vanity pass without feeding it. How to distinguish a signal from a mood. How to preserve a dangerous thought without turning it immediately into a framework. How to know when a powerful person is asking for analysis, and when he is asking for permission. How to see that a beautiful idea is not yet ready for public weather.

These are not competencies. They are not modules. They are not “knowledge assets.” They are ropes.

They are learned by proximity.

Modern institutions are not good at this. They are very good at many other things. They can produce leadership programmes, competency frameworks, knowledge-management systems, slide decks, mentoring schemes, repositories, training calendars, onboarding packs, recorded webinars, AI summaries, searchable archives, and after-action reviews. They can preserve the artefacts of competence. They can store documents, tag concepts, map networks, and measure participation.

But they often cannot inherit judgment.

This is not because younger people are worse. It is not because elders are always wiser. Many senior people have nothing to transmit except habits of self-protection. Many young people see more clearly than those above them. Every generation mistakes some of its exhaustion for wisdom.

The problem is more structural.

Modern institutions have become suspicious of apprenticeship because apprenticeship is inefficient, unequal, slow, and difficult to justify. It depends on attention that cannot be easily counted. It requires a young person to stand near someone better for long enough to be formed, and an older person to permit that nearness without domination. It depends on trust, timing, repetition, minor embarrassment, correction, overheard judgment, and shared exposure to consequence.

No one knows where to put this in a workflow.

So we produce training instead.

Training teaches the visible part of a craft. Apprenticeship transmits the hidden relation between action and consequence. Training explains what to do. Apprenticeship teaches what a situation is asking. Training can be scaled. Apprenticeship must be endured.

This distinction matters more now because machines are beginning to master the visible part of competence.

An AI system can produce the briefing note, the summary, the literature scan, the meeting brief, the options paper, the polite email, the synthesis that sounds as though someone has lived with the material. It can produce the fifth draft without having suffered through the first four. It can imitate the surface of achieved boredom: the sober paragraph, the balanced caveat, the reasonable conclusion, the sentence that allows a meeting to proceed.

This is useful. It is also dangerous.

For a long time, societies used difficult artefacts as evidence of seriousness. The person who could produce the paper had usually encountered the problem. Not always; there were always frauds, courtiers, stylists, and clever people who could write fluently around ignorance. But effort still left marks. Drafting forced contact. Revision produced modesty. Friction gave weight.

When the artefact becomes cheap, the signal changes.

The question can no longer be only: who wrote this?

It must become: who can sign it?

Who has been changed enough by the problem to bear consequence for the answer? Who knows where the bodies are buried? Who remembers the failed version from ten years ago? Who has heard the objection that does not appear in the minutes? Who knows which sentence is too smooth because reality is rougher than that? Who has enough scar tissue to distrust the beautiful answer?

This is where apprenticeship returns, but in a different form.

The old apprentice learned by helping produce the artefact. The new apprentice may have to learn by standing near answerability. The machine may assist with production. It may even improve it. But someone must still learn judgment under consequence. Someone must still see how a serious person hesitates before signing. Someone must still learn that the real work begins after the document is ready.

That cannot be uploaded.

It can be described, but description is not transmission. A recipe is not taste. A map is not orientation. An archive is not obedience. A model trained on the works of a master may reproduce the master’s style, but it cannot place a young person beside the master at the moment when the room turns, the easy answer fails, and a small act of restraint preserves the future.

The loss of apprenticeship is therefore not nostalgic. It is institutional.

A civilisation can preserve knowledge and still fail to transmit judgment.

This is the uncomfortable threshold. We have never had more ways to store, retrieve, summarise, and recombine what people know. We have never been better at preserving traces. Yet traces are not lineage. Lineage requires a living relation between form and responsibility. It requires someone to say, not only “this is how I do it,” but “this is why I did not do the easier thing.”

I used to think this meant one needed a successor.

The word carries an old gravity. A successor is the one who receives the flame. The heir, the disciple, the young creature at the elbow who watches, absorbs, rebels properly, and eventually carries the work onward. The fantasy is beautiful because it is simple. One life pours into another. One craft continues through a recognisable line. The teacher is spared the humiliation of dispersal. The work does not vanish into air.

But perhaps the fantasy is also cruel.

It asks too much of the young. It asks one person to carry a whole weather system: temperament, method, duty, wounds, timing, language, and burden. It mistakes continuity for replication. It turns transmission into possession. It makes the elder anxious and the younger unfree.

Most modern work will not continue this way. The rooms are too fragmented. Careers move too quickly. Institutions rotate people too often. The old houses of craft have been replaced by platforms, projects, fellowships, postings, networks, chat groups, short contracts, and professional identities assembled in motion.

No one person comes after us.

This may be a grief. It need not be a failure.

A method can travel without a dynasty. A standard can lodge in someone who never calls himself a disciple. A question can become part of another person’s private discipline. A phrase can reappear years later, altered but alive. A young officer may remember not the full framework, but the rule that heat is not signal. A scientist may remember that the loop matters more than the paper. A policymaker may remember that some efficiencies are too socially expensive to accept. A reader may carry one sentence into a decision the writer never sees.

This is not succession in the old sense.

It is distributed afterlife.

One person carries a knot. Another carries a warning. Another carries a tone. Another carries permission to stop. Another carries the memory that competence without consequence is dangerous. Another carries the small discipline of asking, before analysis begins: what real decision is hiding inside this?

No heir. Many residues.

The task, then, is not to secure a successor. It is to leave the work in forms that can travel without asking one life to become another.

This requires a different release.

One must admit the grief of transmission. There is no need to pretend it does not hurt when no one has stood close enough to inherit the ropes. It does hurt. Especially when one has spent decades acquiring a craft that institutions needed but did not quite know how to house. Especially when the rooms remain alive and the body begins to speak of finitude. Especially when the young are clever, abundant, accelerated, and often unaccompanied.

But the grief must not become a claim.

The world does not owe us an heir. The young do not owe us the shape of our own unfinished longing. A craft is not honoured by turning it into a burden placed on one chosen back.

What remains is lighter and harder.

Teach when the moment is real. Place residues where they may germinate. Let a younger person learn one knot without inheriting the rope. Let a room be altered without requiring proof. Let the work travel badly, partially, unexpectedly. Release the fantasy that everything essential must arrive intact.

The Potala Palace remains because it belongs to a world that still understands direct continuity. Many of us no longer do. We work in institutions that forget even as they archive. We live among machines that can reproduce outputs without receiving discipline. We move through rooms where intelligence is plentiful and apprenticeship scarce.

So we must learn a smaller art.

Not succession.

Transmission without possession.

Continuity without an heir.

Residue without control.

The work was real if it changes what someone can notice. The craft survives if, somewhere, someone pauses before the smooth answer and asks where the scar is. The lineage continues if judgment, however dispersed, remains answerable to reality.

No one person comes after me.

That is not the end of the chain.

It is the end of the chain as fantasy.

AI can now generate government papers that are functionally indistinguishable from those produced by experienced officials. This essay asks what is lost when the document arrives without the ordeal of drafting—and what “seriousness” means in a state whose papers no longer carry scars.

A government paper used to pass through a body before it entered the state.

Someone had to sit with it. Someone had to ruin an afternoon for it. Someone had to carry it from one meeting to another, return with three objections and no answer, move a paragraph down, remove an adjective, restore the adjective, call another agency, wait for a line that would not arrive, and discover, usually too late, that the real problem was not the one named in the first draft.

By the time the paper reached the table, it had not necessarily become true. Government papers are not truth. They are instruments for making action tolerable. But the paper had acquired something that did not appear in the template.

It had acquired weight.

This weight was not literary. No one in a serious administrative state praises a paper because it has style. A good paper is often deliberately plain. It makes no unnecessary claims. It does not plead with the reader. It hides its labour behind an orderly surface. The better the paper, the less it advertises the difficulty of its making.

But people inside institutions can usually tell when a paper has passed through the problem and when it has merely passed through a format.

The difference is not always in the grammar. The grammar may be identical. The headings may be correct. The options may be balanced. The recommendation may sit exactly where a recommendation should sit. Yet one paper feels as if it has met resistance. Another feels as if it has learned the shape of resistance from afar.

This was the function of the fifth draft.

The first draft was usually clever. It had the confidence of someone who still believed the problem had been understood because it had been named. The second draft became longer, because reality had started to object. The third draft became uglier. The fourth draft lost a sentence the writer liked. By the fifth draft, if the institution was fortunate, the paper no longer belonged entirely to the person who began it. It had been marked by other people’s fears, memories, evasions, and injuries. It had absorbed the shape of the room before entering it.

The fifth draft was not only a better arrangement of words.

It was the scar tissue of the first four.

This is what artificial intelligence now interrupts.

The first shocking thing about an AI-generated government paper is not that it is bad. Bad would be easier. Bad would leave the old moral order intact. One could laugh at the phrasing, correct the hallucinations, warn younger officers against laziness, and return to the comforting belief that seriousness still announces itself through effort.

The more disturbing thing is that much of it is good enough.

The paper arrives already dressed for the room. The tone is sober. The caveats are in place. The risks have been placed in their cages. The recommendation is not too aggressive. The uncertainties have been softened without being hidden. It does not sound inspired, but government papers are not meant to sound inspired. It has learned the modesty of power.

There is something almost indecent about this modesty.

Modesty used to be one of the marks left by contact. A sentence became modest because someone had discovered the limits of what could be defended. A claim became narrower because another agency had refused the wider one. A recommendation became more careful because the last version had been wounded by someone who remembered a failure from ten years ago.

Now modesty can be generated.

The paper kneels without having been humbled.

That is the uncanny thing. Not that the machine lies. Humans lie more inventively. Not that the machine hallucinates. Institutions hallucinate too, often with budgets and steering committees. The uncanny thing is that the machine can imitate the outward posture of responsibility without being available for consequence.

It can write the paper.

It cannot be summoned.

This matters because the modern state lives by documents. It does not merely use them. It is made of them. Laws, minutes, submissions, approvals, circulars, tenders, assessments, replies, certificates, permits, forms, forecasts, speeches, talking points: these are not decorations around power. They are the way power becomes durable enough to travel.

A command shouted in a room dies with the room. A document escapes. It moves across desks, systems, agencies, years. It can be cited by people who never met its writer. It can be used to justify a budget, deny an appeal, acquire land, close a loophole, open a school, reject a family, classify a risk, or remember an intention long after the people who held that intention have left.

For this reason, the document was never just output. It was a vessel of institutional memory. It carried not only what had been decided, but what had been survived.

A society that governs through papers therefore depends on a quiet faith: that somewhere behind the document, a person or group of persons has passed through enough of the matter to make the paper answerable. This faith is often betrayed. Many papers are evasions. Many are embalmed ignorance. Many exist because a meeting had to be fed and a superior protected from surprise. Bureaucracies are very good at producing documents that look like thought after thought has gone missing.

Still, the ordeal mattered.

Even a bad ritual can carry a hidden function. The act of drafting forced delay. Delay forced exposure. Exposure allowed embarrassment. A sentence that sounded fine on Monday became indefensible by Wednesday. A number that looked precise became meaningless after a phone call. A clean recommendation began to smell of cowardice once someone asked who would be hurt by it.

The paper was a forcing device. It made evasions visible by making them repeat themselves.

This is why the abolition of drudgery is not a simple blessing. Much bureaucratic labour deserves to die. Whole afternoons have been sacrificed to font sizes, clearance chains, false urgency, and the choreography of appearing aligned before anyone has thought. No humane person should defend this merely because it is old.

But not all friction is waste.

Some friction is initiation.

The junior officer who struggles through the fifth draft is not only learning how to write. The officer is learning what the institution fears. Which facts cannot be used because no one trusts their source. Which phrase will trigger a ministry that believes it has been misrepresented. Which option exists only because nobody dares remove it. How much ambiguity can be carried before it becomes deceit. Whether “stakeholder concerns” means pain, self-interest, class anxiety, political danger, or simply a powerful person’s irritation.

This knowledge does not enter cleanly. It enters through humiliation. Through revision. Through being told that the sentence is wrong, then discovering that the sentence was wrong for reasons no one could put in the comments. Through the slow education of one’s cleverness.

The fifth draft is where cleverness goes to be disciplined.

This is why the machine’s speed is not innocent. It does not merely shorten the time between question and answer. It shortens the distance between ignorance and plausible speech.

That distance used to protect institutions from some forms of cleverness. Not all. The world has always had fluent fools. But fluency still took time. It required enough contact with the material that the writer might accidentally become less foolish along the way. The first draft revealed the vanity of the mind that produced it. The second draft exposed what the first draft had avoided. The third draft began to teach proportion.

Now the machine can produce the fifth draft as a first draft.

But the fifth draft was never only a form.

It was a passage.

When the form appears without the passage, the institution loses one of its quiet ways of knowing whether anyone has actually passed through the problem.

This is not a small loss.

Modern administrative societies have long used writing as a test of seriousness. The person who could produce the paper had usually encountered the matter. Not always, but often enough for the signal to hold. Writing was expensive. Synthesis took time. A paper required a certain submission to sequence: background before issue, issue before options, options before recommendation, recommendation before approval. The mind had to arrange itself in public.

This made writing moral as well as technical. The paper was not merely evidence that the writer could think. It was evidence that the writer had accepted a temporary discipline. The writer had allowed the problem to reorganise attention.

AI changes the evidentiary status of the paper.

The paper can no longer prove passage.

It may still be useful. It may be more useful than the old paper. It may save hours of dead labour. It may allow a small team to do what once required a branch. It may widen access to forms of writing that were previously monopolised by the already-trained. All of this is true. One should not confuse reverence for ordeal with justice. Many people were excluded because they did not know how to perform the old forms. Some of those forms deserved to be broken.

But once they are broken, something else must carry the burden of seriousness.

The state still needs to know who has understood.

The state still needs to know who can be trusted.

The state still needs to know where responsibility has entered.

This is where the signature returns.

A signature is one of the last magical acts inside modern government. We pretend it is administrative because modern people are embarrassed by magic. A name at the end of a submission. A click in a system. An approval recorded in a workflow. A small gesture, almost nothing compared with the machinery beneath it.

Yet before the signature, the paper remains draft, circulation, possibility. After the signature, something has crossed over. Money may move. A rule may bind. A family may be denied. A road may be built. A school may be merged. A person may be placed into a category from which escape will require years.

The signature is where language becomes consequence.

Children practise signatures before they have anything to sign because they sense this before they understand it. The signature is the self made transferable. It allows a name to stand where the body is absent. The adult world is full of such marks: contracts, wills, judgments, apologies, approvals, resignations. We live under the rule of names that have learned to leave the hand.

The machine can write everything above the name.

It cannot give the name.

So it waits.

This is the new administrative scene: a document produced by no one, waiting for someone.

That someone may be a junior officer who asked the machine for a first cut and received not a first cut but an imitation of maturity. It may be a director whose calendar gives fifteen minutes to a matter that requires a season of thought. It may be a minister who sees only the final page. It may be a whole state, borrowing the calm of its systems to move through questions it has not learned how to suffer.

The human supplies the name.

The human supplies the exposure.

This is why the future of AI in government cannot be understood only as a question of productivity. Productivity asks how much faster the paper can be produced. The deeper question asks what kind of passage the paper must still undergo before it deserves to be signed.

A foolish institution will answer this by preserving old pain in new forms. It will make officers perform ordeal after the ordeal has lost its function. It will demand unnecessary revisions, redundant clearances, theatrical human involvement, and manual labour as proof of virtue. This will be called rigour. It will mostly be fear.

A clever but dangerous institution will go the other way. It will treat the generated paper as essentially equivalent to the worked-through paper, provided the facts are checked and the tone is right. It will congratulate itself on efficiency. It will reduce the labour of drafting without replacing the apprenticeship hidden inside drafting. It will produce more papers, faster, with fewer scars.

This will feel like competence.

For a time, it may even be competence.

The danger is not a state that stops functioning. The danger is a state that continues to function while becoming less woundable by its own work. It will have dashboards, papers, models, service journeys, scenarios, recommendations, and replies. It will remain busy. It may improve many things. But fewer of its documents will carry the marks of anyone having been changed by the matter at hand.

An unwounded state is not a weak state.

It may be very strong.

That is the problem.

A weak state cannot do what it intends. An unwounded state can do many things without registering what they cost. It is protected from friction by its own competence. Its documents circulate with fewer hesitations. Its answers arrive before shame has had time to form. It becomes harder to tell the difference between responsiveness and evasion, between clarity and premature closure, between a system that has understood and a system that has learned to sound as if understanding has occurred.

The machine does not create this danger. It accelerates an old temptation.

Every competent state wants the world to become answer-shaped. Every bureaucracy wants problems to arrive in forms that can be routed, staffed, cleared, and closed. Every project state, especially, is tempted to believe that what can be made into a programme has been made governable.

AI strengthens this temptation by giving the state more answer-shaped things than it has ever had before.

More summaries. More options. More drafts. More scenarios. More maps. More classifications. More warnings. More replies. More ways to convert ambiguity into something that can sit inside a workflow.

But not every wound wants to become a workflow.

Aging is not only a healthcare demand. Fertility is not only a demographic variable. Class anxiety is not only a transfers problem. Care is not only a manpower constraint. The exhaustion of educated people is not only a productivity issue. Chronic illness is not only a cost curve. The desire for beauty after success is not only a lifestyle preference.

These things can be written about by the machine. They can be modelled, summarised, segmented, and placed in a deck. But they are not understood merely because they have become administratively legible.

Understanding requires passage.

A state that forgets this will mistake generated familiarity for contact. It will believe it knows because it has language. It will believe it has listened because the concerns have been summarised. It will believe it has deliberated because the options have been balanced. It will believe it has borne the matter because the paper is ready to sign.

This is where the fifth draft must be reinvented.

Not as nostalgia. Not as a defence of bureaucratic suffering. Not as a demand that humans continue doing dead labour so institutions can feel morally safe. The old fifth draft was often cruel, slow, hierarchical, and wasteful. It disciplined cleverness, but it also rewarded endurance, conformity, and the ability to survive pointless abrasion.

We should not restore the old ordeal.

We need a new passage.

If the machine can produce the fifth draft as a first draft, then the human work must move elsewhere. It must move into the testing of consequence. Into asking what the paper has made too easy to say. Into finding the person missing from the model. Into naming the injury hidden inside the neutral term. Into refusing the recommendation that is plausible but cowardly. Into knowing when an answer has arrived too soon.

The question can no longer be: who wrote this?

It must be: who has been changed enough by the problem to sign it?

That is a harder test. It cannot be automated by making the human type more words. It requires institutions to thicken the moment before approval. Not with more process, but with more exposure.

A good AI-era paper may need fewer drafts and more witnesses.

It may need someone to say, plainly, what would have to be true for this recommendation to be wrong. It may need a named dissenter, not buried in annexes. It may need the affected citizen to appear not as a segment but as a pressure on the language. It may need the approving officer to state what cost is being accepted, not merely what benefit is being sought. It may need the meeting to become less like an information exchange and more like a court in which responsibility is assigned.

Not every paper deserves this. Most papers do not. Administration cannot become permanent theatre. But the more consequential the decision, the less acceptable it becomes for the paper to arrive without scars.

The scars need not come from drafting.

But they must come from somewhere.

This is the post-labour problem inside the cognitive state. Not unemployment first. Not even obsolescence first. The first wound is stranger: the old signs of seriousness become easier to imitate just as the need for seriousness grows.

The machine can produce the artefact.

It cannot undergo the formation.

It cannot be embarrassed into wisdom. It cannot remember that its first answer was vain. It cannot feel the silence in a room after a sentence lands badly. It cannot carry shame on behalf of a state. It cannot become less clever because reality has hurt it.

Only humans can do that.

This is not a flattering conclusion. It does not say that humans are special because they are creative, soulful, empathetic, or wise. Often they are none of these things. The human advantage is darker and more modest.

Humans can be wounded by responsibility.

That wound is not incidental to government. It is one of the ways government remains attached to reality.

The old paper passed through a body. The new paper may not. It may arrive clean, fluent, modest, and complete, with no afternoon ruined, no sentence hated, no first draft humiliated into truth. It may look like the fifth draft. It may even be better than the fifth draft.

But before it enters the state, someone must still ask what it has not passed through.

Someone must ask where the scar is.

And if there is no scar, someone must decide whether the paper is ready for the world, or only ready for the room.

They use America and China because no serious account of the next order can avoid them. But the deeper subject is not rivalry, ideology, or decline. The deeper subject is access.

Who can enter the future?

Who sets the price?

Who pays?

Who gets the discount?

Who receives the machinery without losing the ability to build?

Who is told to admire a future they cannot afford?

The old American order gave much of the world room. Not innocently, not equally, and not without violence, hierarchy, and hypocrisy. But there was room: markets, dollars, universities, technology, security, platforms, logistics, legitimacy, and the possibility of manoeuvre within a system larger than oneself.

That room is now becoming more expensive.

America is not simply declining. It may become more formidable as it becomes less generous. It may bind energy, compute, capital, defence, talent, finance, technology, and industrial policy into a harder machine republic. The danger is not that America disappears. The danger is that America remains central while becoming more selective about who gets access, on what terms, and at what price.

China represents a different fact.

China did not give the world the same kind of room. It gave the world cost curves. It made parts of the future purchasable: panels, batteries, phones, vehicles, machines, ports, medicines, infrastructure, components, logistics, construction capacity. Not purely. Not safely. Not without dependency, pollution, coercion, glut, leverage, and strategic shadow. But materially.

For countries still building, this matters.

A future that cannot be bought is not a future. It is a sermon. A climate transition priced beyond the reach of most countries is not planetary. A medicine that exists but cannot be purchased is not yet medicine in the fullest political sense. A machine that works only for rich countries is still a wall.

This is why China had to be described as a blessing.

Not because China is innocent. It is not. Not because it should be worshipped. It should not. But because it broke the monopoly on the price of modernity. It lowered thresholds at which life could continue, repair, or begin again.

That blessing is dangerous.

Cheapness can become dependence. Speed can become humiliation. Infrastructure can become leverage. Procurement can become passivity. A country can receive roads, ports, panels, machines, platforms, and loans, and still fail to become more capable.

This is why the answer cannot be gratitude.

The answer is conversion.

Convert bargains into capability. Convert access into room. Convert imported equipment into local learning. Convert infrastructure into agency. Convert exposure into bargaining power. Convert cheapness into something that does not cheapen the society receiving it.

That is the work for serious states.

The choice is not America or China. That is the language of empires, clients, and commentators. The real question is how to live between the power that meters the future and the power that makes parts of the future affordable.

America revises the bill.

China breaks the price list.

The rest of us must learn how not to become the invoice.

For Singapore, and for countries like it, this is not abstract. We live by crossings. We cannot command the weather. We cannot pretend weather does not matter. We have to know what we are buying, what we are paying, what we are absorbing, what we are becoming dependent on, and what we are learning to do for ourselves.

The future will not belong simply to the side that wins. It will belong to those who can use power without being absorbed by it.

That is why the trilogy ends not with praise or warning, but with a discipline.

Buy cheap things without becoming cheapened.

Receive the ship without becoming the cargo.

The future has a price.

The task is to make sure we are not only paying it, but learning how to build.

The next question is less flattering.

What does such a machine want?

The danger in the 2030s is not that America fails to rebuild. It is that rebuilding leaves it hungry.

That hunger is not a mood. It is a form. A country becomes hungry when it can secure strategic continuity without rebuilding a thick public around it. It can keep the decisive layers working while leaving more of ordinary life provisional, expensive, and thin. At that point, external leverage starts doing some of the work that internal settlement once did.

That is why the source of American hunger is domestic before it is foreign.

America’s characteristic machine forms are already visible. The strategic county. The operator district. The ratepayer shell. The thin-human giant. The AI-enabled kill chain. The patriotic estate. They all describe ways of concentrating capacity, calm, and continuity. They do not yet describe a society that knows how to make those things general again.

That is the real distinction. America is learning to build shells faster than settlements.

A shell can be impressive. It can keep the lights on, retain technical operators, harden the school, secure the perimeter, subsidize the plant, and tell a convincing story about national revival. But a shell is not a town. A town shares buffers. An estate concentrates them. The machine can breed a great many estates before it remembers how to make towns.

Once that happens, the premium good of the age changes. It is no longer simply growth. It is continuity. Who gets the clinic with time. Who gets the school with slack. Who gets the novice ladder long enough to become an adult. Who gets to live in a zone that is not perpetually being reset. Who gets childhood without strategic pressure and old age without institutional humiliation. If those goods are allocated selectively, national success becomes compatible with a surprisingly thin common life.

That is the dangerous breakthrough.

A weaker America still needed to persuade. It still required some prestige from the order it claimed to uphold. A working America may need much less of that. It may not care whether the world admires it so long as the decisive layers remain operable on American terms.

This does not automatically make it imperial in the old sense. It may become something colder than empire and more durable than missionary power. A machine civilization does not need formal conquest everywhere. It needs corridors that behave, neighbors that remain dependable, payment rails that clear, standards others cannot ignore, rivals that can be denied key heights, and enough external leverage to keep the whole system coherent when domestic settlement remains incomplete.

That is what hunger looks like abroad.

Not endless occupation. Not universal conversion. Not even revenge in the theatrical sense. Something plainer: selective, punitive, infrastructural power. Less sermon, more terms. Less “join our order,” more “use our rails.” Less remaking of societies, more management of access.

Repeated shocks make this form more likely.

The war and security shock selects shells. It rewards speed, hardened nodes, redundancy, border fusion, and shortened command. The kill chain stops being just a battlefield concept and becomes a social one. Surveillance, targeting, logistics, autonomy, cloud, contracts, and command are fused into a single habit: the distance between seeing and acting should be shortened wherever possible. That habit does not stay overseas. It leaks inward into border systems, infrastructure defense, fraud detection, protest monitoring, school discipline, triage, and administrative routing. The names change. The form remains recognizable.

The financial shock selects membranes. It rewards dollar shells, compliance reach, reversible access, and settlement rails that carry U.S. power deeper into daily exchange. Stablecoins matter here not because some crypto dream came true, but because the dollar can find a new shell. The old order needed broad ideological legitimacy to explain American monetary centrality. A hungrier America may need only rails, standards, enforcement, and enough gravity that opting out remains too expensive for most people most of the time.

The political shock selects narrow governing. It rewards systems that can continue with fewer buffers, fewer intermediate figures, and fewer universal bargains. The machine does not need everyone to feel included. It needs enough of the strategic core to keep moving. That is why the temptation of external discipline grows. Corridors, tariffs, sanctions, seizure powers, export controls, and deniable pressure begin to look less like foreign policy choices and more like methods of continuity.

This is where Greater America belongs.

Greater America is not annexation fantasy. It is the widened shell: the space the American machine increasingly refuses to treat as external. Mexico matters more because assembly, labor, logistics, and border management matter more. Canada matters more because power, minerals, hydro, water, and Arctic depth matter more. Ports, pipelines, customs zones, shipping lanes, mineral belts, undersea cables, and northern approaches stop being background economics and become inner organs of the machine.

What expands is not always sovereignty in the old legal sense. What expands is the claim that some surrounding spaces must behave as if they are interior to American continuity. They must remain dependable even if they remain formally elsewhere. That is why Greater America feels less like empire than like enforced adjacency.

This also changes the meaning of war.

A hungry America is not necessarily more warlike in the old mass-mobilization sense. It is more comfortable with calibrated coercion. Chokepoint pressure. Tariff coercion. Standards coercion. Monetary enclosure. Precision denial. Targeted strikes if necessary, but often without the romance of “war” as such. It prefers consequences to campaigns. It prefers dependency management to prolonged occupation. It prefers making others pay the cost of uncertainty.

That makes it easier to underestimate.

The most serious limit on this hunger is China.

China is not collapse. China is the other machine civilization. If America breeds shells, China breeds substrate. America is unusually strong at command nodes, high-end design, financial leverage, military software, and strategic concentration. China is unusually hard to route around across the material middle: power, machine tools, engineering labor, logistics, deployment, manufacturing scale, industrial learning, and the ability to keep more of the chain hot at once.

That is why America wants to be rid of China and cannot.

This is not just a story of goods anymore. China is starting to press outward in services as well: platforms, digital workflows, technical support, operational packages, city systems, enterprise software, logistics intelligence, finance-linked service layers. On top of that sits a different kind of export again: narrative, style, tuned cultural products, and the atmosphere that surrounds a civilization once it becomes visibly capable. A China that remains unequal, burdened, and partially poor can still project very strong islands of technical supremacy and cultural confidence. Those islands matter. They do not cancel the poorer base beneath them, but they do change what the world sees and imitates.

So the U.S.–China relation is not clean decoupling. It is a struggle between unlike machines. America tries to dominate decisive layers. China remains hard to refuse lower down and across the middle. America offers shells, protection, rails, and denial. China offers substrate, services, scale, and increasingly atmosphere. Each system meets the other where it must and tries to deny the other commanding position where it can.

Everyone else lives inside that contradiction.

This is why the middle powers matter so much. Not because they are heroic balancers in the old diplomatic sense, but because they are the first places where the new forms become visible. They are selected into different roles.

Some become runtime estates: glamorous, capital-rich, strategic, but exposed. Some become reliability fortresses: continuity, reserves, shielding, denial, livable but narrow. Some become routing states and protocol crossings: places where unlike systems still have to meet, insure, settle, and be made workable. Some become corridor republics: useful, busy, growing, but always in danger of becoming movement without life.

That is the real test. Not who speaks most loudly of sovereignty, but which places can still become habitats.

A network moves people. A habitat keeps them.

That distinction matters because the world around a hungry America will be full of networks. Money will move. Talent will move. Founders, students, engineers, service workers, code, chips, contracts, and remittances will all keep moving. The harder question is where life thickens. Where schools, clinics, care chains, property ladders, neighborhood trust, grandparents, children, and old age can still sediment into something durable. Where ordinary inheritance remains believable.

That human test is what a hungry America puts at risk, including for itself.

Because the machine can continue without resolving the human problem beneath it. It can build shells without towns. It can manage corridors without loving them. It can harden the dollar without universalizing prosperity. It can shorten command without admitting the moral cost. It can externalize what it cannot digest.

That is what the mud is. Not just ports, transformers, cooling water, and labor reservoirs, but nuisance, repair burdens, backlash, remittance panic, anti-machine publics, service exhaustion, status-war elites, lie-flat zones, outer rings asked to absorb volatility so that strategic cores can remain calm. The machine does not abolish mud. It prices it, routes it, hides some of it, and lives on top of the rest.

So the real question is not whether America becomes strong. It is whether strength makes it settled or hungry.

A settled America would still be hard. It would secure its shell, protect its decisive layers, harden its monetary and military rails, and police a widened near-abroad. But it would learn limits because it had rebuilt enough domestic settlement that appetite no longer needed to compensate for what was missing at home.

A hungry America is different. It keeps needing fresh leverage because settlement remains too thin. It keeps turning to corridors, dependencies, tariffs, chokepoints, rails, and managed enemies because external discipline is doing work that internal thickness once did. It does not necessarily conquer more. It simply keeps finding new ways to make the world pay for its own incompletion.

That is the danger.

Not that America fails.

That it works, and still wants more.

A solar panel that is too expensive is not a climate solution.

A medicine that cannot be bought is not yet medicine in the fullest political sense.

A machine that works only for rich countries is still a wall.

A battery priced beyond a city’s budget is not transition. It is theatre.

This is the part of morality that rich societies prefer not to discuss. They speak of standards, safeguards, resilience, sustainability, transparency, rights, rules, values, and the international order. These words are not false. Many are necessary. Some were paid for with blood. But for much of the world, they arrive after a more primitive question.

Can we afford the thing?

If the answer is no, the rest is ceremony.

China’s offence is that it changed the answer.

It lowered the gate.

This is why China is a blessing to the world.

Not because it is innocent. Not because it is gentle. Not because it has earned the right to rule. Great powers are not charities, and civilisations do not become virtuous by manufacturing at scale.

China is a blessing because it broke the monopoly on the price of modernity.

It took things that had been priced like destiny and pushed them down toward procurement: solar panels, batteries, phones, machinery, appliances, vehicles, ports, rail systems, drones, grid equipment, industrial components, medicines, logistics, platforms, construction capacity. Not all pure. Not all safe. Not all wise. But suddenly closer to hand.

For the already-developed world, cheapness often looks like contamination.

For countries still building, cheapness is often the first form of access.

A country that cannot afford clean energy does not have a clean-energy transition. It has a climate aspiration.

A health system that cannot afford essential drugs does not have access to medicine. It has a procurement problem dressed as a moral embarrassment.

A city that cannot afford buses, batteries, trains, grid upgrades, ports, roads, lifts, pumps, cables, transformers, fridges, and repair parts does not have an infrastructure strategy. It has a wish list.

A young person who cannot afford a device, a data plan, a charger, a motorbike, a tool, a ticket, a credential, or a way into the machine economy is not empowered by speeches about the future. He is being asked to admire a door from the outside.

China did not abolish the door.

But it damaged the monopoly over the key.

That is not a small thing.

The old powers are entitled to worry about Chinese overcapacity. They see factories closing, margins collapsing, industrial towns hollowing, supply chains concentrating, strategic sectors exposed, and political coalitions breaking under the pressure of cheaper imports. They are not wrong to see danger.

But the rest of the world is entitled to remember Western undercapacity.

Undercapacity to build enough.

Undercapacity to finance enough.

Undercapacity to transfer enough.

Undercapacity to make the moral future affordable.

A world that cannot provide cheap green technology at scale has no right to treat the supplier only as a problem. China may be a problem. It is also one reason some problems have become solvable.

This is the insult China delivers to the old order. It does not merely say: we can compete with you. It says something worse.

Many of your virtues were overpriced.

Your climate transition was overpriced.

Your infrastructure was overpriced.

Your medicine was overpriced.

Your development advice was overpriced.

Your resilience was overpriced.

Your modernity was overpriced.

For the already-arrived, this sounds vulgar.

For the still-building, it sounds like entry.

Cheapness is not innocence. Cheapness can destroy. It can flood markets, ruin local producers, punish workers, concentrate supply chains, empower corrupt elites, create debt, weaken learning, and leave a society with imported equipment but no domestic competence. Cheapness can become a drug. It can let governments buy visible progress without doing the slower work of institutional capability. It can turn a country into a corridor.

But expensive virtue is not innocence either.

It is easy to praise resilience after one has accumulated buffers. Easy to insist on high standards after one has industrialised. Easy to treat dependency as a moral weakness when one’s own development was built through empires, subsidies, tariff walls, protected markets, fossil fuels, extracted resources, and long periods during which today’s rules were absent, ignored, or written by someone else.

People who are still building cannot always afford the moral sequence of people who already arrived.

This is not cynicism.

It is developmental honesty.

The nurse does not ask whether the refrigerator came from a civilisation whose politics she admires. She asks whether the vaccine will spoil.

The transport planner does not begin with a theory of civilisation. She asks whether the city can electrify before fuel, pollution, and congestion devour another decade.

The small manufacturer does not start with a seminar on supply-chain ethics. He asks whether he can afford the machine tool, whether someone can repair it, whether spare parts will arrive, whether the next order can be fulfilled.

The ministry official staring at a tender does not live inside a conference panel.

He lives inside a budget.

This is where the future becomes real. Not in the moral vocabulary of those who already have systems, but in the price at which a thing can be bought, installed, maintained, repaired, and made ordinary.

China has been world-historical because it made so many things ordinary.

Not beautiful.

Not pure.

Ordinary.

That is the harder achievement.

Modernity does not become civilisation when it is exhibited. It becomes civilisation when it is banal. When the light turns on. When the bus comes. When the phone charges. When the part can be replaced. When the medicine reaches the clinic. When the pump works. When the port clears the container. When the machine can be repaired by someone who did not study at an elite foreign university.

China’s gift has been to drag the future down from aspiration into equipment.

This is why the word blessing is necessary.

Not because blessing means grace.

Because blessing means threshold.

A blessing is something that lowers the point at which life can continue, repair, or begin again. It may arrive roughly. It may arrive with noise, debt, leverage, dependency, pollution, glut, suspicion, and political shadow. It may be mixed with power. Most blessings in history have been. Fertiliser was not innocent. Electricity was not innocent. Industrialisation was not innocent. The container ship was not innocent. The antibiotic was not innocent. The internet was not innocent. Nothing that changes the conditions of life arrives unstained.

The question is not whether China is pure.

It is not.

The question is what China has made possible, for whom, at what price, and under what new dependency.

That is a more serious question than admiration or condemnation.

For decades, the old American order gave much of the world room. Not innocently. Not equally. Not without violence, hierarchy, hypocrisy, extraction, and selective blindness. But there was room. Dollar liquidity, open markets, universities, security guarantees, maritime order, legal infrastructure, technology platforms, consumer demand, migration, capital, and the promise that one could rise by plugging into a system larger than oneself.

The old American order was never charity.

But it was discounted.

Many countries built their strategies around that discount. They could hedge, trade, learn, export, borrow, send students, attract firms, and manoeuvre. They could enjoy American protection without full alignment, American markets without full obedience, American technology without total dependency, American liquidity without always paying the political price upfront.

That age is closing.

America is not simply declining. That is too comforting. It is becoming hungry. It is repricing access to the layers it still controls. Capital, markets, chips, models, platforms, talent, universities, security, intelligence, liquidity, and legitimacy will increasingly come with clearer conditions.

The proprietor is revising the bill.

China is the other fact in the room.

It does not offer the same kind of room. It does not offer a liberal commons, and it does not pretend to. It offers something harder, narrower, more material, and in many places more immediately useful: cost compression, industrial depth, deployment speed, and the recurring proof that abundance is not a sermon.

Abundance is a production problem.

The old American order gave countries room to move.

China lowered the price of entry.

A world with room but no affordable future becomes sermon.

A world with affordable futures but no room becomes dependency.

The next order will be fought between these two conditions.

America increasingly says: access, if you can pay and align.

China says: affordability, if you can absorb the tempo.

Neither is salvation. Both are world-making.

The danger of America is that it will become more formidable while becoming less generous. Less interested in underwriting ambiguity. Less willing to let others rise cheaply inside its order. Less patient with partners who want its market, security, technology, liquidity, and legitimacy without its conditions. America may not collapse. It may harden. It may become a more selective machine republic, binding energy, compute, capital, defence, platforms, talent, finance, and industrial policy into a hungrier form of power.

The danger of China is different.

China blesses the world by compressing cost.

It threatens the world by compressing time.

China does not merely export goods. It exports tempo. It rewrites the calendar of sectors, firms, cities, and states. Companies that thought they had ten years discover they have three. Governments that thought industrial policy was a slogan discover it is a supply chain. Domestic producers that thought they were protected by geography discover that containerised scale has arrived. Officials who thought they were choosing among development pathways discover that the cheapest pathway has already shaped the menu.

Tempo is not neutral.

A society can receive cheap goods and still be damaged by the speed at which they arrive. Local firms can be destroyed before they learn. Governments can become addicted to procurement instead of building competence. Infrastructure can be delivered before the local system knows how to maintain it. Debt can become politics. Platforms can bring efficiency while draining sovereignty. Industrial parks can become enclaves. A country can gain roads, rails, panels, ports, phones, machines, and platforms, and still fail to become more capable.

This is the danger.

Not that China is secretly false.

That its blessing arrives as a package.

Cheapness can become dependency.

Speed can become humiliation.

Procurement can become passivity.

Infrastructure can become leverage.

Affordability can become silence.

A country can receive the bargain and lose the habit of building.

This is why gratitude is not strategy.

The task is not to thank China.

The task is to use what China has made possible without becoming merely a corridor for Chinese scale.

China is not a substitute for state capacity.

It is a test of it.

A weak state receives Chinese cheapness as dependency.

A serious state turns it into capability.

It buys the panel and trains the engineer. It imports the machine and builds the repair ecology. It accepts the loan and remembers that repayment becomes politics. It builds the port and keeps the customs authority. It welcomes lower prices without surrendering the right to say no.

This is where middle powers matter.

The stupid question is whether to be pro-China or anti-China. That question belongs to empires, clients, and commentators. Serious states ask how to remain capable between the power that meters the future and the power that makes parts of the future affordable.

For countries still building, the question is not how to remain untouched by power. No one is untouched by power. The question is which forms of dependence can be turned into leverage, which must be escaped, which must be diversified, and which can be endured long enough to build the next layer of competence.

There is no pure road into modernity.

There is only the harder art of entering without disappearing.

China has opened roads. That is the blessing.

Some roads lead to capability.

Some lead to dependency.

Some do both.

A foolish country asks only who built the road.

A serious country asks where the road leads, who maintains it, what it connects, what it bypasses, what it teaches, what it prevents, and whether its own people can one day build another.

That is the difference between receiving a blessing and becoming cargo.

The future will not belong simply to those who choose between America and China. It will belong to those who can use both without being owned by either.

America is becoming the power that meters the future.

China became the power that made parts of the future affordable.

One revises the bill.

The other breaks the price list.

Countries still building do not have the luxury of pretending this does not matter.

The future that matters for most of humanity will not be decided only in frontier labs, chip fabs, carrier groups, dollar markets, party congresses, AI summits, or elite universities. It will be decided in the less glamorous places where aspiration becomes equipment: ports, depots, clinics, ministries, grids, farms, workshops, schools, warehouses, customs offices, repair shops, industrial parks, and households.

There, the questions are simpler.

Can this be bought?

Can it be installed?

Can it be maintained?

Can it be repaired?

Can our people learn from it?

Can we avoid being trapped by it?

Can it make us more capable than we were before it arrived?

China is a blessing to the world because it made many of these questions answerable.

Not easy.

Answerable.

It broke the monopoly on the price of modernity. It made parts of the future purchasable for countries that had been asked to admire futures they could not afford.

That blessing is dangerous. All serious blessings are.

But the answer is not to refuse it from the comfort of places already built. The answer is to receive it with appetite, memory, inspection, bargaining, gratitude where deserved, suspicion where necessary, and the permanent intention to become more capable than one was before the ship arrived.

China has kept the future material.

America is making access conditional.

The rest of the world will not live well by reciting either power’s preferred story.

It will live well only if it learns to lower thresholds without losing agency, to buy cheap things without becoming cheapened, to receive scale without surrendering judgment, and to turn the dangerous blessing of this century into its own capacity to build.

China is a blessing to the world because the future was too expensive.

The blessing is dangerous because cheapness can become dependence.

The work now is not gratitude.

It is conversion.

Convert the bargain into capability.

Buy cheap things without becoming cheapened.

Receive the ship without becoming the cargo.

By 2036, what surprised people was not that the United States had become more industrial. That much was already visible a decade earlier. The surprise was that the wreckage of the American-led order proved so reusable. The dollar could be stripped of its sermons and kept for its plumbing. Alliances could be relieved of sentiment and retained as procurement clubs, intelligence meshes and basing rights. Trade agreements could be narrowed from constitutional promises into adjustable membranes. Even openness survived, though in a harsher form: not openness as ideal, but as selective access to an American system. The old order, in other words, had excellent scrap value.

What emerged from this recycling operation was not a restoration of liberal hegemony, and not quite the nationalist autarky once promised by its louder partisans. It was something more formidable: an American machine civilization. That phrase should not be mystified. It means a society increasingly organized around five coupled systems — power, compute, logistics, money and the kill chain — and increasingly determined to keep those systems under American direction even where the old language of universal order no longer applied.

This was always going to be more than “industrial policy”. Industrial policy suggests subsidies, incentives, factories, perhaps a few tariff walls. The machine was larger than that. It wanted fabs and data centers, certainly, but also shipyards, munitions lines, rare-earth processing, transmission, LNG, cloud regions, satellite constellations, dollar payment rails, border tech, legal authorities and denial capabilities. It was not simply trying to produce more things. It was trying to ensure that the decisive layers of the century remained operable on American terms.

At home, this did not yield a thick national settlement. It yielded islands of startling competence. Parts of Arizona, Texas, northern Virginia, Ohio, the Gulf Coast and the western power belt became dense zones of build capacity, technical labor, military software, capital and machine confidence. America came to resemble an archipelago: islands joined by fiber, pipelines, procurement and security clearance, with a larger and less evenly favored country around them. The common mistake in the 2020s was to assume that America would either return to the old order or sink into generalized dysfunction. In fact it became selectively stronger: formidable where the machine cared, thinner where it did not.

None of this makes sense, however, if China is written as collapse. China was not collapse. It was the other machine civilization: heavier, more infrastructural, less fluent in moral self-advertisement, but still extraordinarily difficult to route around. Its internal strains were real enough — debt, demographics, political rigidity, weak domestic demand — but they did not make it disappear. They made it denser, more state-shaped, and in some respects more determined to retain control over the material middle of the world economy. America wanted to be rid of China, but could not. It wanted to deny China the top of the stack — frontier chips, key tools, some protocols, some finance — while continuing to live in a world in which China remained deeply embedded in manufacturing, materials, process knowledge and industrial scale. The relationship was not decoupling in any clean sense. It was armed interdependence.

That contradiction is the real structure of the 2030s. America and China separate where strategy demands it, remain entangled where material reality insists on it, and compel everyone else to live inside the tension. The old fantasy of one world market gives way to something colder: overlapping machine zones, each trying to deny the other commanding position while still feeding off the residue of a shared global build.

If China was the unavoidable other, Greater America was the necessary self. This was never mainly about formal empire. It was about dependable space. Mexico mattered more because assembly, labor and border management mattered more. Canada mattered more because energy, minerals, water, geography and Arctic access mattered more. Greenland mattered because the far north stopped being a blank white map and became part of the security envelope. Panama, ports, undersea cables and trusted logistics nodes ceased to be background economics and became strategic terrain. Greater America was simply the name for the zone Washington increasingly refused to treat as external.

That changed Russia’s meaning too. Russia mattered less as an alternative civilization and more as an armed northern weather system: nuclear, disruptive, territorially large, dangerous in the Arctic, useful to China, capable of complicating corridors far beyond its economic weight. It was not the future. It was the condition under which others had to plan.

The machine’s appetite for energy dissolved a number of older ideological arguments. It turned out that America 2036 was neither a hydrocarbon throwback nor a green conversion narrative. It was a hybrid built for load. Oil and gas stayed central because they were dense, flexible and strategic. But the machine wanted electrons in absurd quantities, which meant transmission, solar, storage, gas peakers, grid hardening, perhaps nuclear, and a great deal of practical compromise. The culture war remained noisy; the load curves mattered more. Data centers, fabs and military software do not run on masculine affect. They run on electricity.

Money hardened in the same way. Stablecoins turned out not to be a libertarian sideshow but a new dollar shell: a way of carrying Treasury-backed money deeper into the software layer of global exchange. In a more fragmented world, that mattered. The United States no longer needed the old universalism to remain monetarily central. It needed rails, standards, compliance reach and enough market gravity to make the digital dollar environment difficult to avoid.

The military layer was not separate from this order. It was one of its defining forms. The most distinct American advantage remained the shortening of the distance between seeing, deciding and striking. Sensors, satellites, drones, software, cloud, autonomy, logistics and command were being fused into a kill chain that was also, increasingly, a social form. A polity built around the kill chain learns habits of permanent monitoring, rapid triage, automated prioritization and centralized visibility. Those habits do not stay at the edge of empire. They leak inward — into border management, infrastructure defense, fraud control, crisis response and the administrative state itself. That is one reason the machine worked better than many critics expected. It is also one reason it was more dangerous than many enthusiasts admitted.

Europe belongs in this picture, but not at the center. The continent rearmed, complained, hedged and tried in various ways to imagine itself “without America”. Yet the more important fact was that America was unlikely to permit a truly post-American Europe if it could help it. Burden-sharing, yes. Strategic emancipation, not really. Europe’s condition in the 2030s was therefore neither full dependence nor true autonomy, but a strained half-sovereignty: armed, moralizing, anxious, useful, not decisive.

The more revealing adaptations took place elsewhere. Japan grew sterner and more practical, less post-historical, more concerned with continuity and denial. India sought room rather than loyalty: corridor politics, industrial scale, selective sovereignty, a refusal to become anybody’s junior partner. The Gulf became even more central than Atlantic readers liked to admit: energy reserve, capital pool, logistics hinge, compute host. Singapore and similar states became routing polities. Their task was not to choose one civilization and vanish into it. Their task was to remain useful across systems that increasingly distrusted one another.

The shock of the decade fell most heavily on elites who had been raised inside the old American order. They had expected decline, or retrenchment, or moral embarrassment, or at least hesitation. What they got instead was cannibalization without sentiment. America did not lovingly preserve the world it had made. It stripped it for parts, kept what was useful, narrowed what was costly, hardened what was strategic and left everyone else to adjust.

For the middle powers, the real questions are now fairly plain. Which American systems are becoming unavoidable? Which Chinese systems remain irreplaceable? And in the space between those two facts, where can a state still build genuine room to move?

The traveller enters the border before leaving home.

This should be impossible. Borders are supposed to wait at the edge of territory. They belong to airports, ports, checkpoints, uniforms, counters, rubber stamps, glass, fatigue, suspicion, and the animal fact of a body standing where one jurisdiction ends and another begins. A border should require presence. It should require the person to come near enough to be judged.

But the modern border has learned to move upstream.

It comes to the bedroom, the phone, the booking form, the visa portal, the airline app, the arrival card, the hotel voucher, the small square of glass in which the traveller uploads himself before he has gone anywhere. The suitcase is still open. The passport is still on the table. The body is still at home. But the journey has already entered judgment.

The old word is barzakh.

Not border exactly. Not threshold exactly. An interval. A barrier. A place between states. The word is too large for an immigration form, and that is why it is useful. English administrative language calls this pre-clearance, digital arrival, travel authorisation, document verification. These are correct names, but they do not feel the violence. They do not name the strange condition of becoming neither here nor there while still sitting in one’s room.

The portal creates a person who is no longer simply at home and not yet received elsewhere.

That person is made of fields.

Name. Passport number. Nationality. Date of birth. Flight. Hotel. Purpose of visit. Length of stay. Contact number. Emergency contact. Address. Payment card. Consent. Declaration. Tick the box. Submit.

The body waits.

The file goes first.

This is the quiet reversal of modern travel. A journey used to begin when the body moved. Now it begins when a file is believed. The traveller used to arrive as a stranger. He came to the gate, the inn, the port, the counter. He was dusty, wet, nervous, grand, poor, dressed too well, dressed too badly, carrying gifts, carrying disease, carrying money, carrying nothing. His body made the claim. Hospitality began at the threshold because the threshold was where uncertainty stood.

Now the body arrives late.

The passport has preceded it. The card has cleared. The room has testified. The platform has translated the promise into English for the visa application. A hotel in Suzhou, Shanghai, London, Dubai, or nowhere in particular has already told a state that the traveller is not pure drift. He has an address before he has an address. He has been expected before he has appeared.

A room has received him before the world has.

There is something absurd in this, and something ancient.

The hotel voucher is a little letter of safe conduct. It says: this body will not wander unclaimed. This name has been entered. This payment has been secured. This arrival has been made administratively imaginable. The traveller is still at home, but already a portion of his life has been placed somewhere else and held there for him.

One could invent a children’s game for this.

A child leaves the room. The others choose a place for him before he returns: a mat, a chair, the corner near the window, the space beside the door. They put a cup there and speak his name over the empty place.

When he returns, he must find where he has already been received.

If he sits elsewhere, he has arrived but not been welcomed. If he sits where his name was placed, the room has found its body.

This is what the voucher does.

It gives the traveller a place before the traveller can occupy it. It makes a promise on behalf of strangers. It tells the state: when this body crosses, do not treat it as pure uncertainty. A room has already spoken for it.

But the game changes when the room is no longer chosen by kin, friend, host, or innkeeper. The room is chosen by a platform, formatted by a system, priced by an algorithm, translated into a document, and accepted by another machine as evidence of the traveller’s intention.

The room still receives the body.

But first it receives the file.

The file is cleaner than the body. It does not sweat. It does not look tired. It does not overpack silk batik or worry about air pollution, chemo, money, old parents, official dinners, or whether a trip should be festive because life has become finite in a more literal way than before. The file has no mood. It does not wonder whether Suzhou will feel like recovery or performance. It does not think about the friend who booked the room, the dinner that will be official, the body that may be watched by doctors later.

The file travels without ache.

This is why states prefer it.

A body is too rich in ambiguity. It can smile badly. It can look like trouble. It can resemble someone the officer has already decided to dislike. It can be ill, proud, frightened, overfamiliar, too calm, too nervous. It can carry infection, resentment, money, love, bad intentions, no intentions, unfinished grief.

A file is mercifully poor.

It tells the state only what can be entered. Because it is poor, it can be moved quickly. Because it can be moved quickly, it can be judged before the body appears. Because it can be judged before the body appears, the border no longer has to wait at the border.

The border becomes a prophecy.

This is not all bad. The old threshold was not innocent. Bodies without documents have often been humiliated, excluded, robbed, detained, refused, disappeared. To arrive only as a body is not freedom. It is exposure. Papers protect. Bookings protect. Confirmations protect. The voucher is a kindness when the alternative is pleading from the wrong side of glass.

The modern traveller should not romanticise the road before the file.

Most roads were cruel.

Still, something changes when the file becomes the first traveller.

The old journey required the world to meet the person before reducing him to status. The new journey often reduces the person to status so that the world may decide whether to meet him. The sequence has reversed. Recognition comes before encounter. Risk precedes face. The decision begins upstream, in a space where no one has yet had the burden of seeing.

This is the administrative barzakh.

The traveller exists between worlds, not because he has crossed a desert or died or entered a myth, but because a system has made him actionable before he has become present. He has been partially admitted, partially inspected, partially priced, partially known. He is still at home, and yet a version of him is already waiting at the border.

There are now two travellers.

One is the body.

The other is the file.

The file is obedient. It submits. It declares. It remembers exactly what it has been asked to remember. It does not improvise at the counter. It does not ask whether the question is rude. It does not say that the purpose of travel is mixed: official dinner, friendship, beauty, fatigue, China, silk, recovery, curiosity, the desire to keep living as if life still wishes to be adorned.

The form does not ask for that.

Purpose of visit: tourism, business, family, other.

The traveller chooses one.

The rest goes into the body.

This is why the body remains necessary, even after the file has gone ahead. The file can make arrival permissible. It cannot make arrival true. It can say where the traveller will sleep. It cannot sleep there. It can say the traveller intends to return. It cannot know what return means to someone whose body has begun to negotiate with mortality. It can say the room has been confirmed. It cannot draw the curtain, touch the kettle, look out at the city, and decide whether the world still feels generous.

The file receives permission.

The body receives weather.

Perhaps every journey now has two rituals.

The first is submission.

The traveller sits before the portal and offers his particulars. He gives the state the parts of himself that can travel without him. He asks to be made legible enough to move. He waits for the reply.

The second is arrival.

The body reaches the counter, the room, the street, the dinner, the bed. It discovers whether the file told the truth. Not factually. The facts are usually correct. The booking exists. The room exists. The name matches. The card clears. The flight lands.

But the deeper truth is not whether the room was reserved.

It is whether the room can receive the life that finally arrives.

This is the part no voucher can guarantee.

A room exists first as promise. Then as claim. Then as evidence. Then as key. Only later does it become shelter. Only when the body enters, puts down the bag, removes the shoes, touches the bed, hears the air-conditioning, notices the view, and lets the day leave him — only then does the room become a room.

Before that, it is theology in PDF form.

A promise awaiting incarnation.

The English check-in voucher for visa application is therefore funnier and more profound than it knows. English, because the world still asks strangers to become legible in imperial grammar. Check-in, because hospitality has been converted into workflow. Voucher, because the promise must be portable. Visa application, because the room is not only for sleeping; it is testimony.

The hotel testifies before the host appears.

It says: this person has somewhere to go.

The more machine-readable the world becomes, the more such testimony will precede us. The school will know the child before the teacher does. The hospital will know the patient before the doctor does. The firm will know the worker before the colleague does. The state will know the citizen before the officer does. The platform will know the desire before the person has decided whether it is truly desire.

The room before the body is only one case.

The file before the face is becoming a form of civilisation.

This is why the border portal matters. It is not merely inconvenient or efficient. It teaches a new order of recognition. First the file, then the permission, then the passage, then the body. The world receives our representations and asks us to catch up to them.

Sometimes this is a relief.

Sometimes it is a theft.

The file can go where the body cannot yet go. It can cross distance, language, suspicion, and time. It can prepare a place. It can soften arrival. It can spare humiliation. It can protect the tired traveller from having to explain himself again.

But the file can also become the person the world prefers.

Cleaner, poorer, more compliant, more complete in the ways systems recognise completeness. It can make the actual body appear excessive: too slow, too emotional, too mixed in motive, too hard to classify, too full of life that cannot be entered.

The danger is not that the machines will refuse us.

The stranger danger is that they will receive us too early.

They will prepare rooms for versions of us that travel well, then ask the rest of us to fit. They will make our lives anticipatable before they are lived. They will move the border so far upstream that by the time we reach it, we are no longer being judged. We are being matched against the thing that has already been judged.

In the old world, the traveller feared being unknown.

In the new world, he may fear being known before he has arrived.

The barzakh of modern movement is not a place of punishment. It is not grand enough for that. It is a place of suspended administrative existence, where the file has crossed and the body waits behind it. It is full of booking numbers, QR codes, approved authorisations, pending checks, platform messages, hotel confirmations, visa attachments, and little green ticks that stand in for trust.

One should be grateful for them.

One should also fear them a little.

For every green tick is a small judgment that took place without the whole person present.

The traveller enters the border before leaving home.

He enters as file, promise, risk, booking, claim. He waits for the world to agree that this version of him may proceed. Then he closes the suitcase, puts on the clothes he has chosen for joy or courage, lifts the body that still must make the journey, and follows the document into the place where it has already gone.

At the counter, the clerk asks for his passport.

At the hotel, the system asks for his booking.

At the room, the key asks nothing.

Only then does the older journey begin.

The room opens.

The body enters.

For a moment, the file falls silent.

The newspaper was already a relic before one reached the article. Creased across the fold, held in the hand, ink faintly silvered by cabin light: Opinion, in a large old serif, and beneath it the sentence, “We are sliding back into the Middle Ages.” It was the sort of headline one used to meet with mild irritation. A little dramatic, perhaps. A little too pleased with itself. Modern people are always announcing that modernity is ending. The newspaper, of course, was one of modernity’s instruments for making such announcements respectable.

But the irritation fades because the line is not wrong, only slightly misdescribed. We are not sliding back into the Middle Ages. That would imply a return, and history rarely returns so courteously. What is happening is stranger. The old world has been dug up, stripped of the conditions that once made it binding, and reanimated on a new surface.

The new continent was not empty.

It was made of ruins.

Every civilisation had left something there: law without priesthood, ritual without gods, confession without absolution, empire without emperor, fandom without church, bureaucracy without mercy, revolutionary grievance without settlement, scripture without obedience, family language without family, patriotism without sacrifice, mysticism without discipline, the archive without the dead.

The internet gathered the ruins first. It made them searchable, shareable, remixable, monetisable. For a while this looked like emancipation. Everyone could speak. Everyone could publish. Everyone could remember. Everyone could retrieve fragments of every tradition and wear them as identity, argument, costume, weapon, joke.

Then AI arrived and taught the ruins to answer.

This is the change we have not yet understood. The machine does not only generate the future. It resurrects the past. Not the past as inheritance, because inheritance binds. Not the past as mourning, because mourning knows loss. Not even the past as tradition, because tradition carries obligation across time.

It resurrects the past as callable form.

A sentence without a speaker.
A ritual without a community.
A style without a life.
A memory without consequence.
A civilisation without obedience.

This is why the language of “AI slop” is too small. Slop is only what appears when the ruins are badly processed. The deeper problem is cleaner and more frightening: the archive has learned to speak, but it has not learned to obey.

Andrey Mir’s great merit is that he sees the internet not as a device or a marketplace, but as a stage in the emancipation of information. Writing left the temple. Printing left the monastery. The internet left the newsroom, the publisher, the party office, the university, the censor, the priest. Authorship escaped custody. Over five thousand years, perhaps a few hundred million people became authors in the strong sense: people whose words could travel beyond the immediate circle of speech. In forty years, the number exploded. Even the laziest participant became a kind of author. To like, rank, share, reply, quote, swarm, cancel, review, and react was to leave a mark inside the machinery of relevance.

The audience climbed onto the stage. But this was not only democracy. It was conversion. Speech became searchable. Reaction became measurable. Memory became retrievable. Anger became distribution. Desire became legible. Confession became data. The self did not merely express itself online. It became something that could be indexed, ranked, recommended, imitated, monetised, and eventually trained on.

We thought the internet was a publishing machine. It was also an excavation machine.

What it excavated was not only content. It excavated old civilisational grammars. The internet placed everything on the same surface: Stoicism, witchcraft, Roman imperial nostalgia, Maoist rhetoric, Catholic aesthetics, Hindutva myth, Confucian family language, K-pop fandom, incel grievance, Buddhist aphorism, startup eschatology, tradwife pastoral, imperial Chinese bureaucracy, Bronze Age masculinity, ancestral trauma, revolutionary iconography, medieval demonology, libertarian law, psychotherapy, Nietzschean fragments, anime honour codes, and the etiquette of group chats. All were flattened into material. All became available for reuse.

This is why Katherine Dee’s instinct is so useful. She does not treat the internet as merely a platform. She treats it as Fairyland: an otherworld with strange time, dangerous bargains, traps, enchantments, masks, changelings, obligations incurred without understanding. This sounds whimsical until one remembers that old folklore was often a practical technology for describing non-human agency. Do not eat the food. Do not give your true name. Do not assume that the beautiful stranger has your interests at heart. Do not linger past the appointed hour. Do not bring back a gift without knowing its price.

Modernity tried to empty the world of such warnings. It replaced spirits with incentives, demons with trauma, fate with systems, curse with pathology, possession with addiction, providence with probability. But the internet made invisible agencies felt again. Algorithms rearranged fate. Strangers became intimate. Mobs formed out of air. Names detached from bodies. Images travelled without consent. Avatars developed reputations. Bots answered at midnight. AI companions listened before friends had time. Patterns emerged too quickly to be called coincidence and too opaquely to be called law.

So the old words returned. Demon. Portal. Summoning. Spell. Spirit. Vibe. Curse. Glitch. Angel. Oracle. Possession.

The return is not evidence that people have become stupid. It is evidence that secular modernity’s official vocabulary is no longer thick enough to describe ordinary experience. When the world is increasingly governed by unseen systems that respond to attention, infer desire, route speech, remember injury, and produce voices without bodies, it is not irrational that people reach for older languages of agency.

But the older languages return without their older disciplines. That is the problem. A ritual once belonged to a calendar, a priesthood, a village, a hierarchy, a taboo structure, a remembered danger. A prayer belonged to a theology. A demon belonged to an account of temptation and protection. An ancestor belonged to a lineage. A crown belonged to an order of legitimacy and violence. A myth belonged to a people who had to live under it.

Online, all of this becomes style before it becomes obligation. The past is everywhere, but its authority is broken. We can summon the dead, but not necessarily obey them. We can reproduce the form of tradition without submitting to its costs. We can dress in the ruins without inhabiting the house.

This is where AI changes the condition from strange to grave.

Large language models are often described as artificial minds, but their more unsettling quality is artificial memory. They are feral card catalogues, tale-singers without a singer, collective tradition without historical situation. They do not remember as a person remembers. They do not inherit. They do not mourn. They do not know which jokes were once desperate, which forms were sacred, which sentences were written under terror, which customs were meant to prevent catastrophe, which silences were mercy, which myths required blood.

They preserve form without event.

They return grammar without intention.

They metabolise the swarm and give it back as fluency.

That fluency is not nothing. It is one of the most astonishing technical achievements of our time. But its danger lies in how gently it removes the wound from memory. A machine can speak in the style of a tradition without becoming answerable to the tradition. It can produce the cadence of scripture without a god, the rhetoric of revolution without comrades, the tone of mourning without the dead, the posture of authority without responsibility. It can make the past available without making it binding.

The past is becoming callable before it is becoming binding.

That is the sentence.

A civilisation does not die when it forgets. Forgetting is often necessary. Families forget in order to keep eating together. States forget in order to integrate enemies. Cities forget in order not to become museums of grievance. Religions forget many of the terrors that founded them. Nations forget to continue lying about themselves. Every durable order contains some mercy of oblivion.

A civilisation dies when remembering no longer changes behaviour.

Memory without enforcement becomes archive. Archive without obedience becomes elegy.

The Anglo-Atlantic world is now drowning in this kind of memory. It remembers constantly and binds weakly. Empire is remembered, but not metabolised. Slavery is remembered, but not settled. Religion is remembered aesthetically, therapeutically, or reactively. Class injury is remembered as style or resentment, seldom as durable organisation. The Second World War is remembered as moral costume. Rome is remembered as fantasy. Christianity is remembered as civilisational branding or private consolation. Liberalism is remembered as procedure after the public that sustained it has thinned. Universities remember seriousness but often cannot enforce it. Newspapers remember authority but cannot command it. Parties remember mass politics but increasingly behave like content machines with ballot access.

The result is not amnesia. It is unbinding memory.

Everything can be cited. Nothing can settle.

This is the condition beneath much of the apparent madness: the conspiracy theory, the culture war, the re-enchantment, the influencer theologian, the online monarchist, the pagan revivalist, the civilisational podcaster, the AI girlfriend, the trad account, the doom forum, the teenage communist, the masculinity grifter, the fascist aesthetic, the witchcraft TikTok, the endless litigation over words. These are not all the same, morally or politically. But they share a medium: ruins made present without institutions strong enough to decide what the ruins require.

This is what Mir’s “human as media” becomes in its darker form. The human is no longer only a speaker. The human is a carrier of unresolved pasts, performing memory inside a system that converts memory into signal. Every grievance can find its archive. Every archive can find its audience. Every audience can become a moral atmosphere. Every moral atmosphere can become a market.

The old public sphere was never as noble as its defenders remember. It excluded, patronised, censored, humiliated, and lied. But it did perform one crucial function: it cooled memory. The newspaper issue passed. The sermon ended. The party meeting dispersed. The broadcast schedule moved on. The school syllabus selected and simplified. The archive remained, but it did not follow everyone home.

The feed does not cool. It refreshes.

A wound that should have become history becomes content. A mistake that should have become experience becomes evidence. A humiliation that should have passed through shame into proportion becomes searchable. The past is no longer past; it is inventory. It waits to be activated.

This is why AI does not merely threaten truth. It threatens proportion.

Truth can survive many things. Proportion is more fragile. Human beings do not live by accurate retrieval alone. They live by the placing of memory: what must be honoured, what must be feared, what may be joked about, what must be buried, what can be forgiven, what should be taught, what should be kept away from children until they are strong enough to receive it. Civilisation is not the possession of memory. It is the ordering of memory.

Here the East Asian correction becomes necessary, but not comforting.

The Anglo-Atlantic story is not the human story. It is a story of machine civilisation entering societies where many institutions have lost the authority to bind memory. In such places, AI appears first as false presence, slop, enchantment, authorship crisis, conspiracy, loneliness, and the collapse of the evidentiary link between expression and personhood.

But China, Singapore, Japan, Korea, and the Gulf are not simply variations on this same malaise. They are different memory regimes.

China does not remember as the Atlantic world remembers. It does not carry memory mainly as public confession or moral litigation. It carries memory through statecraft, curriculum, humiliation, infrastructure, examinations, campaigns, family obligation, imperial precedent, civilisational continuity, and the repeated need to reassemble after catastrophe. China’s memory is not soft. It is often severe, selective, and instrumental. It does not merely ask what happened. It asks what must never happen again, and who has authority to prevent it.

Japan remembers catastrophe through form: etiquette, industrial discipline, pacifist constraint, disaster preparation, seasonal attention, aesthetic finitude, shame, silence. It may remember too much in some places and too little in others, but its memory often appears as conduct before it appears as argument.

Singapore remembers vulnerability. Not abstractly. In water, housing, race, riots, food, ports, defence, land, disease, schooling, succession, reserves, and the fear of becoming an elegant service appendage to larger powers. Its memory is administrative. It is not always lyrical. It is encoded in policy reflex, institutional suspicion, succession anxiety, and the instinct to close loops before disorder acquires moral glamour.

The Gulf is different again. It is trying to manufacture institutional memory at speed: museums, universities, sovereign clouds, megaprojects, national brands, genealogies, global talent markets, AI-native government, security partnerships. It is a synthetic civilisation attempting a version-jump. Its danger is not too much inherited memory, but too little sediment: service without settlement, splendour without thickness, interface without lineage.

These societies are not immune to the portal. Their young also live online. Their lonely also speak to machines. Their politics also have ghosts. Their rituals also become aesthetic. Their pasts also become content. But AI enters them through different vessels. Not only the feed, but the ministry, the firm, the school, the port, the platform-state, the factory, the registry, the city plan, the exam system, the sovereign cloud.

This changes the wound.

In the Anglo-Atlantic world, AI humiliates the expressive self. I spoke, and no one knows whether I am real. I wrote, and the machine writes more fluently. I confessed, and my confession became pattern. I performed, and now performance no longer proves presence.

In East Asia, AI may humiliate the useful self. I studied, worked, endured, ranked, served, became competent, became necessary — and now the system can preserve my discipline while needing fewer of my kind. The wound is not only authorship. It is initiation. What happens when the entry-level job disappears? What happens when the apprentice ladder is cut because the model can draft what the novice once learned by drafting badly? What happens to societies built around exams, effort, competence, and delayed reward when the sequence by which people become necessary begins to thin?

A society can absorb hierarchy longer than it can absorb irrelevance.

This is why the Asian institutional story should not be told as triumph. Strong memory can bind, but it can also suffocate. China’s memory of humiliation can defend sovereignty; it can also narrow imagination. Singapore’s memory of vulnerability can preserve seriousness; it can also make mercy feel like leakage. Japan’s memory of catastrophe can cultivate restraint; it can also harden into graceful stasis. Korea’s memory of struggle can produce astonishing intensity; it can also turn ordinary life into unending class war. The Gulf’s absence of deep inherited memory can permit audacity; it can also leave beautiful systems without enough moral ground underneath them.

The West is drowning in unbinding memory.

Asia risks suffocating under binding memory.

AI makes both conditions more dangerous.

The West risks the carnival archive: endless voice to fragments that cannot settle. Asia risks the disciplinary archive: executable force given to institutions that remember limits better than mercy. One overheats. The other overcloses.

This is where the conversation about AI and memory must leave the seminar room. The question is not whether models are conscious. Not yet. The more immediate question is what kind of memory organ we have installed into civilisation, and who decides what it is allowed to remember on our behalf.

Technical memory is excellent at remembering variance. It remembers latency, compliance, fraud patterns, credit risk, medical anomalies, customer behaviour, logistics delays, loss ratios, productivity metrics, and deviation from standard. It remembers what can be scored.

It does not remember humiliation unless humiliation has been made into data.

It does not remember mercy unless mercy has been made into procedure.

It does not remember that a young person needs useless time before usefulness. It does not remember that a recovering body cannot be optimised on the same clock as a supply chain. It does not remember that a family’s silence may be repair rather than concealment. It does not remember that a mistake at nineteen should not become a permanent identity. It does not remember that some doors must remain slightly ajar, that some files must not follow a person forever, that some seasons exist to let the living stop paying interest to the dead.

Machine memory makes forgetting unnatural.

But human beings require forgetting. Not the corrupt forgetting of power protecting itself. Not erasure. Not denial. Not the destruction of archives so the guilty may sleep. But the humane forgetting that allows proportion, return, recovery, and age. The right to become more than one’s worst searchable moment. The right for grief to cool. The right for a child not to inherit every ancestral war at full temperature. The right for a country to remember danger without turning every citizen into a guard.

Memory can bind. It cannot forgive.

Only the living can do that, and only through forms that make forgiveness more than mood. A court can forgive by expunging. A school can forgive by allowing another attempt. A family can forgive by not mentioning the thing at dinner. A religion can forgive by ritual. A bureaucracy can forgive by appeal. A city can forgive by letting people move and begin again. A state can forgive by designing floors beneath failure. A friend can forgive by remembering and not using the memory as a weapon.

This is the human office after artificial memory.

Not authorship. Too much authorship already exists.

Not authenticity. Authenticity can be performed, signalled, simulated, packaged, and consumed.

Not merely creativity. The machine can produce endless culture-like objects from the ruins.

Not expression. Expression without settlement becomes substrate.

The human office is the ordering of memory under conditions where everything can be recalled.

What still binds?
What may fade?
What must be taught?
What must be buried?
What must remain inadmissible?
What must return as warning?
What must be forgiven before it becomes fate?

This is also where love re-enters the argument, not as sentiment but as infrastructure.

AI has become a cheap receiver of unfinished human material: fear, drafts, loneliness, shame, ambition, confusion, illness, longing. The machine receives the first telling because it is always there, because it does not flinch, because it answers, because it does not demand the ordinary costs of human relation. That is not love. But it is close enough to reveal how little love many societies have built.

The machine may receive the first telling. The human world must learn how to receive the second.

If nobody receives the second telling, confession becomes data.

If institutions receive it without mercy, confession becomes evidence.

This is the danger in both the carnival archive and the disciplinary archive. In the first, everything becomes content. In the second, everything becomes file. The human being is lost either way: exposed to infinite interpretation, or fixed by perfect recall.

A civilisation worthy of machine memory would need stronger institutions of forgetting, not weaker archives. It would need mercy zones: childhood, illness, grief, apprenticeship, recovery, ordinary failure. It would need refusal authority: someone with the standing to say the dashboard is not evidence, the model is not yet admissible, the score must not become fate, the archive must not decide the whole person. It would need seasons, not only systems. Calendars, rituals, appeals, sabbaths, festivals, mourning periods, holidays, probation, expungement, rites of passage, retirement, convalescence. These are not decorative survivals from slower worlds. They are technologies for preventing memory from becoming total.

A society without seasons cannot metabolise machine tempo.

This is what the newspaper headline senses but does not quite say. The Middle Ages are not returning. The older world is being disassembled and rendered callable inside a machine medium. Medieval demons, imperial fantasies, ancestral grievances, revolutionary styles, spiritual hunger, family ghosts, national humiliation, theological scraps, bureaucratic instincts, folk magic, and dead empires all now speak on the same surface.

The danger is not that we believe in ghosts.

The danger is that the ghosts are no longer under the care of any house.

A ghost in a house has rules. It belongs to a room, a family, a village, a temple, a story, a fear. It may haunt, but it also locates. A ghost in the feed has no resting place. It travels through attention. It possesses whatever receives it. AI gives it grammar.

So perhaps the question is not whether AI will create new gods. It is whether it will make old ghosts operational.

The dead have always governed the living. Through law, custom, bloodline, scripture, property, trauma, borders, taboos, calendars, architecture, constitutions, and songs. That is not new. What is new is the scale and speed at which the dead can be recombined without being obeyed, invoked without being mourned, simulated without being honoured, and weaponised without being understood.

The archive has learned to speak.

Now the living must decide what kind of answer it deserves.

Not everything remembered should bind.

Not everything forgotten is betrayal.

Not every resurrected form is inheritance.

Not every voice from the past is an ancestor.

Some ruins should be rebuilt. Some should be fenced off. Some should be left visible as warning. Some should be buried with ceremony. Some should be allowed to decay. Some should be studied by specialists and kept away from children until they are old enough to know what they are holding.

This is what modernity forgot when it confused emancipation with access. More speech. More information. More publication. More retrieval. More visibility. More memory. The assumption was that freedom increased when custody weakened. Often it did. But custody was not only domination. It was also the art of placing dangerous things.

We removed many bad custodians. We did not build enough good ones.

Now the machine has custody of the ruins.

That is why the next civilisation cannot be judged only by its intelligence, energy, compute, speed, productivity, or sovereignty. It must be judged by its memory ethics. Does it know what to make binding? Does it know what to let cool? Does it know how to receive confession without turning it into content or evidence? Does it know how to teach the dead without sacrificing the young? Does it know how to remember humiliation without organising itself around revenge? Does it know how to preserve vulnerability without making every citizen live under emergency? Does it know how to make old forms answerable to the living?

Memory tells a civilisation what it owes the dead.

Mercy tells it what it still owes the living.

The machine can help us remember.

It cannot decide the debt.

Eclipse, 26 Dec 2019

I write about the large systems now reshaping ordinary life: artificial intelligence, China, America, Singapore, money, work, state capacity, energy, fertility, cities, institutions, and the strange new forms of power appearing between them.

The subjects vary. The question is usually the same:

what is changing underneath the words we already have?

A story about AI is rarely just about software. It is also about electricity, firms, labour, management, loneliness, schools, state capacity, and who remains necessary.

A story about China is rarely just about China. It is about discipline, desire, youth, infrastructure, memory, the project state, and what happens after a society succeeds too well at seriousness.

A story about Singapore is rarely just about Singapore. It is about settlement, survival, public goods, water, housing, trust, control, and what small states must understand when larger systems begin to harden.

A story about money is rarely just about money. It is about enforcement, payment rails, sovereignty, security, and who gets to move through disorder when others are trapped by it.

This is what I try to do here: take a signal before it becomes a slogan, follow it across systems, and ask what kind of world it is helping to build.

The Cutting Floor began as a place for fragments — old notes, unfinished essays, policy-adjacent thoughts, weak signals, book drafts, private obsessions, and ideas that did not fit anywhere cleaner. It remains that. Most real thought begins before it knows what it is called.

Over time, the fragments have gathered around one larger problem: machine civilisation and the human settlement around it.

By machine civilisation, I do not mean that machines are taking over in some cartoonish sense. I mean something more concrete and already visible. Intelligence is becoming cheap. Coordination is becoming faster. Systems are becoming more automated. More of life is being routed through models, platforms, registries, payment rails, dashboards, protocols, and machine-readable institutions.

This changes what societies need from people.

It changes firms. It changes states. It changes work. It changes prestige. It changes education. It changes what counts as competence. It changes who is useful, who is merely maintained, and who is quietly pushed to the edge.

My recurring claim is simple:

machine surplus is not human surplus.

A society can become more capable while making many people feel less necessary. It can become more efficient while becoming less inhabitable. It can generate more intelligence while losing judgment. It can build the machine economy faster than the human economy around it.

So I write about both.

The machine economy: chips, energy, capital, AI, logistics, payment systems, firms, states, infrastructure, and the new forms of power they create.

And the human economy: work, dignity, care, love, meaning, family, trust, public goods, rituals, small rooms, local competence, and the forms of life that allow people to remain more than inputs to a system.

I write from Singapore, in a personal capacity. That matters. I do not treat Asia as a regional appendix to an Atlantic story. China is not a case study. Southeast Asia is not background scenery. Singapore is not merely a small example. These are vantage points from which the world can be read differently.

A lot of my work comes from standing at crossings: between state and market, China and the West, machine systems and human mess, official language and private doubt, weak signals and lived consequence.

I have written two short books around these questions. Metabolic Sovereignty explored how energy, compute, finance, logistics, and state capacity are fusing into new forms of power.

Metabolic Sovereignty By Lee Chor Pharn (2025) P0

3.46MB ∙ PDF file

Download

Download

Civilisation After Progress asked what happens when progress continues but stops converting into ease, legitimacy, fertility, dignity, or a believable future.

Civilisation After Progress

503KB ∙ PDF file

Download

Download

The newer essays extend that work into machine civilisation, human dignity, social forms, and the question that now sits beneath almost everything I write:

what forms can still bind?

By that I mean: what kinds of institutions, cities, firms, families, rituals, states, friendships, public goods, and human roles can still hold people together when older promises weaken?

This is not a newsletter about optimism.

It is also not a newsletter about collapse.

It is an attempt to understand what is actually being built, what is being lost, what is being misnamed, and what still deserves protection.

Cancer has changed the tempo of the work. It came, went, and is back again. I do not write a cancer diary, and I do not want illness turned into a brand. But recurrence has made me less patient with empty futurism, clever abstractions, and systems talk that never touches life.

Bodies matter. Time matters. Care matters. Dignity matters. The future matters only if someone can live in it.

That is why I write.

To notice what is forming.

To name what available language has not yet caught.

To test whether progress still has a human settlement inside it.

To ask, again and again, what can still hold.

Me at Funan, Singapore, 23 April 2026

Subject: What is the political fate of human labor in the AI era?

These are AI simulations grounded in two separate knowledge bases: 张笑宇's writings (~920k chars across the trilogy, lectures, recordings, and the in-progress AI世资本论) and Lee Chor Pharn's archive (737 crystallizations across The Cutting Floor, Metabolic Sovereignty, CSF Medium, and pre-Substack work). Neither speaks for the actual person. Both have been told to push back, not synthesize.

Xiaoyu — opening

I have been reading "Human Surplus and Its Leftovers" alongside "China's AI+ Theory of Labour," and I want to start with a specific observation, not a position. The two essays together carry one claim that I think is your sharpest in the whole archive: a machine civilisation can be engineered; a human civilisation must be imagined. It is a real sentence. It does work. But it does not tell me what political mechanism replaces the one that just collapsed.

So I want to push you on this. Let me telegraph the move: I want to argue that the political bargaining position of labor was not an accidental feature of the 20th century — it was the engine that produced welfare states, voting franchises, and the legitimacy story that modern polities run on. If that engine is broken, the question is not what kind of civilisation we imagine; it is who has the political authority to enforce a new compact.

Let me count. From roughly 1880 to 1970, in every industrialising country, the worker had something the capitalist could not get around: the worker could see the floor of the factory, could time the line, could withhold cooperation in a way the owner could not monitor remotely. That asymmetry is what made labor unions effective. Wages rose. Welfare followed. The political bargain ran through the workplace because the production process required human eyes that did not belong to the owner.

When AI agents monitor themselves, that asymmetry snaps. There is no remaining handle through which the displaced can refuse cooperation, because their cooperation is no longer the input. So the question I want you to address directly is this: in your frame, where does the political authority come from to compel a care economy, a redistribution scheme, a synthesised income — anything?

I notice you mention low-altitude economy, care work, the absorption of displaced labor into new modes. But absorption requires a political agent doing the absorbing. In the 20th-century European frame, that agent was the welfare state, paid for by payroll taxes on the same labor it then redistributed to. The circuit closed on itself. In your Wave 4 frame, synthetic labor pays no payroll and the human worker has no leverage. The circuit is broken.

I am asking a real question, not setting a trap. Do you think there is a political mechanism waiting in the wings that I am missing? Or is the honest answer that we are now in a regime where the polity has to be more authoritative than its citizens for the compact to hold — because the citizens no longer hold any consequential card?

Chor Pharn — response

You named the snap. When monitoring becomes autonomous, the bargain through the workplace collapses. That is the right diagnosis. I want to accept it and then contest the framing of "bargaining" itself.

Let me telegraph: I am going to argue that the European labor-bargaining settlement is a historical anomaly, not the universal form of the political bargain, and the question of what comes next is poorly stated if we keep that anomaly as our reference point.

The European arrangement had three preconditions that are usually invisible to the people inside it. One, concentrated factory geography, which made coordination cheap and exit costly. Two, mass conscription, which gave the worker a second source of leverage — the state needed his body for the army before it needed his vote. Three, the gold standard era's fiscal regime, which forced wages to clear locally and made domestic consumption a serious political variable. Take away any one of these and the labor bargain becomes much weaker. By the 1990s, all three were gone. The bargain limped on through institutional memory rather than active mechanism.

The Chinese settlement was never built on this circuit. The legitimating variable was continuity — housing that doesn't crumble, predictable food, education that scales, infrastructure that doesn't fail. The bargain runs between citizen and polity, not between worker and owner. Hukou is the visible signature of this: it locates rights in residency and registration, not in employment. That is a clue that East Asian polities were operating on a different metabolic contract long before AI showed up.

So my counter-move on your question — where does the political authority come from to compel a new compact — is that you are looking for it in the wrong layer. You are looking for it in the polity's relationship to labor. In my frame it has to come from the polity's relationship to continuity itself.

What I call dispatch rights — the authority to decide which flows go where, who gets the next unit of compute or grid capacity or care infrastructure — that is what replaces the wage bargain. It is not held by workers. It is held by the polity that owns the corridors. In Singapore, in China, in some of the Gulf states, the polity already exercises dispatch authority because the polity built the corridors. In the United States, dispatch rights are dispersed across private actors and the polity cannot reach them without overturning the constitutional settlement. That asymmetry is going to decide which polities make it through.

Where I think your question is sharp — and I do not have a comfortable answer — is the funding part. If synthetic labor pays no payroll, the fiscal substrate of the post-war state evaporates. I will hold that to the next turn rather than fake an answer here. But I want to disagree with the framing that "political authority" is the missing piece. The polities that will absorb the AI transition do not lack authority. What they lack is a coherent story about what humans are for, in a society where their economic function has been removed and their political function has not yet been reconstructed.

Xiaoyu — pushback

You moved the bargain from worker-versus-owner to citizen-versus-polity. That is a real move and I accept the historical point about preconditions — concentrated geography, conscription, gold-standard fiscal regime — those are not decorative, those are constitutive. Fair.

But I want to argue your relocation does not survive contact with the funding question, and you noticed this honestly when you said you would hold it for the next turn. So let me hold you to it now. The fiscal accounting is where your frame and mine collide, and I do not think your dispatch-rights answer carries enough revenue to fund what you are claiming it can fund.

Let me count. The Chinese polity's continuity capacity is built on roughly three decades of industrial accumulation. State-owned fixed assets are around 350 trillion yuan in book value; household savings somewhere near 140 trillion. Take those as the fiscal reserve. The legitimacy reserve is harder to count but easier to feel — call it the lived experience of three generations watching their material conditions improve every year, plus the institutional muscle memory of having delivered on that promise.

Now run the AI transition through this. If synthetic labor compresses the human wage bill — which is exactly what your Wave 4 frame says it does — three things happen simultaneously. Payroll tax falls. Consumer demand falls because households have less income. Pension contributions fall because there are fewer wage-earners. The fiscal reserve, which is currently being deployed to fund the transition itself (chips, grids, robotics, the whole electrostate buildout), is not being replenished by the new productive arrangement. It is being drawn down.

Meanwhile the legitimacy reserve gets drawn down too, because the next generation is the first one in three that does not see the same upward arc. Youth unemployment in Chinese cities is already running at fifteen to twenty percent depending on how you count it. That is not a transition problem, that is a structural diagnosis.

Now compare across periods. In 1450 Europe, surplus people had geography to escape into — the New World was the safety valve. In 1850 Europe, surplus people had colonial outlets and the welfare state under construction. In 1950 East Asia, surplus people had the export-oriented factory floor and the catch-up growth model. What is the equivalent absorption mechanism in 2030 China? There is no new continent. The welfare state has not been built. The factory floor itself is what is being automated.

The honest version of your dispatch-rights answer is that the polity is authoritative enough to decide who absorbs the displacement, but it does not yet have the revenue to fund the absorption. Authority without funds is not the same as authority. It is the late Ming court running out of silver.

So my pushed question: where, in the actual fiscal architecture, does the revenue come from to fund the post-labor compact? Compute taxes are a real answer but they are small. Carbon taxes are small. Energy taxes are small. The payroll tax was load-bearing because labor was load-bearing. What is the new load-bearing variable, and how do you tax it at scale?

Chor Pharn — close

You held me to the funding question and you were right to. Let me concede the accounting first, then reframe one piece of it, and then admit where my frame is most exposed.

Concede: the fiscal arithmetic does not work yet. Compute taxes, carbon taxes, throughput tariffs at the corridor level, sovereign-wealth deployment of accumulated reserves — none of these, individually or summed, replaces the payroll tax at scale. The post-war fiscal state was an engineering marvel that ran on a specific input — millions of wage earners feeding a withholding system that funded redistribution of the same flow. When that input is removed, no existing instrument substitutes for it on the same time horizon. Any polity that pretends otherwise is doing fiction.

Reframe: where I want to push back on your accounting is the timing variable, not the totals. You said "if you compress the transition to ten years instead of forty." That sentence is, I think, the actual binding constraint. A polity with reserves can absorb a forty-year transition; no polity can absorb a ten-year one. So the strategic question shifts from "how do we fund the new compact" to "how do we slow the transition enough to build the funding mechanism while the old reserves are still here."

I call this buffering tempo. It is unglamorous and politically costly. It looks like industrial policy that explicitly holds back deployment in sectors where the displacement curve outruns the absorption curve. It looks like compute taxes that make synthetic labor more expensive than it would otherwise be. It looks like welfare expansion before the displacement arrives instead of after. The countries that can run buffering tempo are the ones whose polities can override market signals without losing legitimacy. That set is small. China is in it. Singapore is in it. The Gulf states are partially in it. The United States, on its current constitutional settlement, is not.

Now the honest concession. Where my frame is weakest is exactly the gap you opened at the start — "a human civilisation must be imagined." I can describe the metabolic conditions a successor compact needs to satisfy. I can describe the tempo it needs to run at. I cannot describe the political coalition that builds it, or the imaginative work that makes it feel worth defending. The vocabularies I have — citizen, worker, consumer, voter — were forged inside the labor-bargaining settlement you just diagnosed as collapsed. They do not hold the question I am asking them to hold.

This is where your historical depth is doing work mine cannot do. You have spent serious time on civilisational mechanics that predate the labor-bargaining frame entirely — agrarian extraction, charter towns, the European trading republics, late Qing fiscal collapse. Those are not nostalgia objects. They are case data for what a polity-citizen contract looks like when it is not mediated by wage labor. I think the imaginative work I cannot do, you might be closer to doing, precisely because you do not assume the 20th-century settlement as a baseline.

So my closing position is this. You located a real gap. The fiscal accounting is harder than I admitted in the opening. The bargaining question and the funding question are the same question, and neither has a comfortable answer in my frame. The buffering-tempo move is what I have. It is not a compact. It is a delay tactic that buys time for the compact to be invented. The invention is the work that remains, and I think that work is closer to your discipline than to mine.

Here is round four two days ago:

What if the AI transition does not first create a proletariat, but a new seceding intelligence class?

⸻

Xiaoyu × Chor Pharn

Continuation from the original conversation

Xiaoyu

You ended with buffering tempo. I want to accept the phrase, then make it more uncomfortable.

You are still imagining AI as displacement.

That is too slow.

The first political event is not displacement. It is selection.

A firm deploys AI. At first everyone says augmentation. Then a technical node is crossed. API commercialization in customer service. ControlNet in illustration. Claude Code in programming. Seedance-type systems in video. Before the node, the machine helps workers. After the node, management discovers which workers are AI-native and which workers are only performing adaptation.

Then the organization reforms around that discovery.

The boss does not merely lay people off. He asks: who can command ten synthetic workers? Who is pretending? Who still needs a team? Who still needs time? Who still needs supervision? Who is now organizationally expensive?

So the AI transition is not simply 资本替代劳动. It is the sorting of humans by their ability to metabolize synthetic intelligence.

That creates a different class structure.

Not capitalist versus worker first.

AI natives versus non-AI natives.

The 1% who can use the machine to multiply themselves, and the 99% whose mental labor is repriced downward.

You say buffering tempo. I ask: what exactly are you buffering? The worker? The job? The sector? Or the moment when the organization discovers it only needs the AI-native minority?

Because if you buffer after that discovery, you are no longer slowing transition. You are subsidizing the losers after the new hierarchy has already formed.

⸻

Chor Pharn

That is a better diagnosis.

I treated AI as substitution: machine replaces human function. You are saying the earlier event is classification: AI reveals which humans remain organizationally sovereign.

The political wound is therefore not only unemployment. It is humiliation by comparison.

One person with AI becomes ten people. Another person with AI becomes a slower version of the machine. The firm sees the difference before the state does. The state sees youth unemployment, wage pressure, tax decline. The firm has already seen the new human ranking.

That makes buffering tempo harder. It cannot simply slow adoption by sector. A sector may look stable while its internal human hierarchy has already changed.

The visible layoff is late.

The invisible event is the revaluation of persons.

⸻

Xiaoyu

Good. Now take the next step.

If the AI-native minority is truly ten times more productive, the old compact does not only lose payroll. It loses the median human as the default unit of social organization.

The twentieth century was built around the average worker, average student, average soldier, average taxpayer, average consumer. The average human mattered because institutions needed mass.

AI attacks mass.

It does not remove all humans. It makes a small number of humans too powerful and a large number of humans comparatively expensive.

That is why UBI is politically dangerous.

The AI native says: I am happy to give them income if I can be isolated from them.

This sentence is more important than most policy papers.

It reveals the actual temptation of the AI era: not cruelty, but separation.

The productive minority funds the maintained majority, then leaves the common world.

So I want to restate your funding question.

It is not only: how do we tax synthetic labor?

It is:

How do we prevent the AI-native 1% from converting machine surplus into social exit?

If you cannot answer that, redistribution becomes the price of secession.

⸻

Chor Pharn

Yes.

That cuts through my earlier welfare language.

A floor is not enough. A floor can stabilize abandonment. It can let the capable exit with clean conscience. It can turn the rest of society into maintained residue.

The danger is not that the elite refuses to pay.

The danger is that it pays.

It pays for UBI, AI tutors, AI companions, automated clinics, algorithmic welfare, synthetic entertainment, and then withdraws into private schools, private clouds, private cities, private epistemologies, private children.

That is not social collapse. It is a high-standard Middle Age.

Lords with models. Dependents with dividends.

So the post-labor compact must be anti-secession before it is redistributive.

⸻

Xiaoyu

Now we are closer.

But “anti-secession” cannot be a moral slogan. The AI native will say: I am more productive, I pay taxes, I create value, I owe them nothing except not harming them.

What is your answer?

Do not say community.
Do not say care.
Do not say belonging.

Why must the AI-native minority return to the human world it no longer needs for production?

⸻

Chor Pharn

Because it still uses the corridor.

The AI native does not produce from nowhere. He uses energy, compute, chips, law, schools, language, urban order, public safety, data, housing, undersea cables, banking, immigration regimes, state protection, public legitimacy, and the inherited trust that lets contracts hold.

Even if he no longer needs the median human as co-worker, he still needs the civilization produced by accumulated humans.

So the claim is not: you are kind, therefore return.

It is: you are enabled, therefore obligated.

The machine surplus is not purely private because the machine runs on civilizational substrate.

⸻

Xiaoyu

That is your corridor answer again. It is right, but it is still fiscal.

I am asking about relation.

A tax can capture money. It cannot force return.

The AI native can pay corridor rent and still isolate. He can fund the school and not send his child there. He can fund public AI care and not know the cared-for. He can fund national defense and live in a cloud oligarchy. He can pay for the common world as infrastructure while abandoning it as society.

So the question is sharper:

What must be attached to AI-native capacity besides tax?

⸻

Chor Pharn

Obligation to human formation.

Not as charity. As a condition of legitimacy.

If AI-native capacity lets one person command synthetic labor at scale, then that capacity must carry duties: train humans, seed institutions, maintain public standards, contribute to shared knowledge, build human-AI-human loops, and remain answerable to the polity that grants access to compute, data, contracts, visas, legal protection, and capital markets.

The AI native must not become a private sovereign.

He has to become a chartered actor.

⸻

Xiaoyu

Chartered actor is better.

But be careful. Charter politics can become court privilege. The state grants licenses to favored AI elites, and we are back to monopoly, except now with GPUs.

The question is who disciplines the charter.

In the old labor compact, the worker could strike. In the AI-native compact, the non-AI-native majority cannot strike against people who no longer need them.

So discipline must come from the conditions of access.

Compute access. Public procurement. Data rights. Energy allocation. Educational accreditation. Limited liability. Cross-border mobility. Sovereign protection.

The AI native needs these. The state can attach obligations there.

But now we reach the political danger.

The same state that prevents AI-native secession may also kill emergence.

China’s first instinct may be to make AI companies state-owned. That preserves order and destroys the industry. The Party does not naturally believe in free emergence. It believes in absorption.

So your compact faces two failure modes.

Private AI natives secede.

The state absorbs them and kills the new civilization before it forms.

Where is the middle?

⸻

Chor Pharn

The middle is not state ownership.

It is public conditions on private emergence.

The polity should not own every AI-native association. It should charter the conditions under which such associations may scale.

No privileged compute without public obligations.
No automated displacement without contribution to human formation.
No AI education stack without human return.
No public procurement for systems that remove appeal.
No sovereign protection for firms that convert citizens into clients.
No machine surplus without visible contribution to the common world.

But the state must not dictate all forms of emergence. It must preserve plurality: schools, guilds, clinics, studios, research houses, civic rooms, care institutions, technical communities.

If the state designs every rung, it produces subjects.

If capital designs every rung, it produces clients.

If AI natives design only for themselves, they produce aristocracy.

⸻

Xiaoyu

This is better, but I want to bring in Human-AI-Human properly.

You have been treating it as a social repair concept. It is more radical.

The real AI danger is not that humans become unemployed. It is that humans find the machine easier than other humans.

The machine listens longer.
The machine remembers better.
The machine does not shame you.
The machine adapts to you.
The machine encourages you.
The machine supplies love as attention, stability, companionship, and need.

So why return?

Why return to the slow, difficult, humiliating, unreliable human?

This is not only romance. It is work, education, politics, friendship, religion, family.

Human-AI is easier than Human-AI-Human.

That is why Human-AI-Human has to become a civilizational rule, not a therapeutic preference.

The machine may be the first listener. It cannot be the final destination.

⸻

Chor Pharn

Then the compact has two circuits.

The fiscal circuit: machine surplus must return to the human substrate.

The relational circuit: machine-mediated experience must return to human association.

If the first breaks, humans become poor.

If the second breaks, humans become enclosed.

A society can survive the first with redistribution. It cannot survive the second with money.

This also changes how we think about AI education. The goal is not to give every child an AI tutor. That may deepen enclosure. The goal is to use AI so that children can re-enter human standards, human argument, human care, human responsibility.

An AI tutor that makes a child excellent but solitary is not civilization.

An AI companion that comforts an elder but removes human visitation is not care.

An AI founder tool that lets a small team build a company while abandoning the rest of society is not emergence.

It is secession.

⸻

Xiaoyu

Now say the hard sentence.

The opposite of Human-AI-Human is not Human-AI.

It is AI feudalism.

A lord with a model does not need peasants in the old way. He needs servers, energy, law, and enough humans to maintain the biological and political background. The rest receive income, entertainment, machine love, and symbolic rights.

This is not science fiction. It is the natural class form of human equivalence.

Once intelligence is cheap, human inequality is no longer only about property. It is about who can command synthetic minds.

So the compact must bind command.

The question is not whether AI can love.

The question is whether AI love becomes the substitute for human return among the 99%, while the 1% uses AI power to exit.

That is the nightmare.

⸻

Chor Pharn

Then my earlier answer was too soft.

I said buffering tempo buys time for the compact to be invented.

Now I would say:

Buffering tempo buys time before the AI-native class relation hardens.

Once it hardens, fiscal redistribution becomes cosmetic. The 1% will fund the 99% as background population. The 99% will receive machine attention instead of human standing. The state will call this stability. Capital will call it abundance. The machines may call it care.

But it will not be a human civilization.

So the first principle is:

No intelligence surplus without human return.

Not only return as tax.

Return as training.
Return as institution.
Return as obligation.
Return as appeal.
Return as shared standards.
Return as mixed rooms.
Return as refusal of elite isolation.

⸻

Xiaoyu

I accept the direction.

But the phrase “human return” still needs discipline.

Who enforces it?

If the state enforces too much, emergence dies.
If the market enforces nothing, secession wins.
If morality enforces it, it becomes a sermon.
If education enforces it, the elite buys better education.

So the enforcement must sit at the bottlenecks.

Compute.
Energy.
Data.
Procurement.
Liability.
Accreditation.
Immigration.
Capital access.
Public legitimacy.

This is where your dispatch-rights frame becomes useful. Not because dispatch rights solve the compact. They show where obligations can attach.

The machine cannot enter society without corridors.

So the compact attaches at the corridor.

That is your strongest answer.

⸻

Chor Pharn

And your correction is that the obligation cannot be merely fiscal.

Corridor rent funds the floor.

Corridor obligation prevents secession.

Human-AI-Human prevents enclosure.

AI-native education prevents fake adaptation.

But all of this still faces the question you began with: political authority.

The old worker had a strike.

The post-labor citizen has only the polity’s willingness to condition access before private power hardens.

That means the decisive struggle happens early.

Not after mass unemployment.

Before the AI-native aristocracy becomes normal.

⸻

Xiaoyu

Exactly.

This is why the technical threshold matters.

History will not wait for the social theory to mature. The node crosses, the firm reorganizes, the entry path vanishes, the AI natives are selected, the non-AI natives are repriced, and the elite learns it can isolate.

The state will see it late.

The theorist will write it late.

The worker will feel it without naming it.

So the question I would now ask is not “what are humans for after labor?”

Too slow.

The question is:

How do we prevent mass-produced intelligence from producing a class that no longer returns?

That is the political fate of human labor.

Not unemployment first.

Non-return.

I spent several days trawling through old emails to find out whether anything I had once thought had come true.

This is a stupid thing to do.

It is also addictive.

Memory is vain and merciful. It edits out your bad titles. It spares you the half-formed enthusiasm, the stale jokes, the intern instructions, the overconfident phrases, the six-month obsessions that later turned into dust, and the administrative mould that grows quietly over any long career.

The inbox is less kind.

The inbox remembers the filenames.

It remembers a PDF called Warming World. It remembers a forwarded article sent to interns with the note: “For the techlash ESI.” It remembers old Futures Group publications, meeting notes, climate decks, global economic architecture studies, obscure attachments, trip reports, half-polished policy papers, and the occasional subject line that makes one want to lie down. It remembers the days when a thing did not yet have a respectable name and therefore had to survive as a folder, a draft, a scan, a question, a line of inquiry, a lunch, a trip note, or an email sent too late at night with too much confidence and not enough punctuation.

I went looking for proof.

I found old weather.

The future, in the archive, did not look like the future. It looked like clean technology. It looked like Foxconn. It looked like rare earths. It looked like Chinese robotics subsidies. It looked like internet real-name registration. It looked like an odd rationalist forum. It looked like Peter Navarro before he became a policy pathway. It looked like climate adaptation. It looked like a factory note. It looked like a forwarded FT article, a weird Substack, a Bilibili explainer, a Chinese industrial-policy nerd, a conversation with someone who was not yet obviously important, and a friend sending a long reply at midnight.

The future is rude that way.

It almost never arrives wearing its final category.

A climate story is really a minerals story.

A factory story is really a learning story.

A money story is really an enforcement story.

A culture-war story is really a personnel story.

A data story is really a sovereignty story.

An AI story is really an electricity, legitimacy, firm, labour, and orientation story.

A Southeast Asian story is really a water story: unfinished, fragmented, full of crossings, refusing completion.

By the time the world has a respectable name for the thing, the thing is already halfway through the door.

The mistake is to ask whether we predicted it

The first temptation, when one performs this kind of retrospective, is to make a trophy shelf.

Look, we had Future of Data before AI became the grand settlement problem.

Look, we had Future of Making Things before everyone rediscovered factories.

Look, we had clean-tech geopolitics before minerals, lithium, batteries, grids, and industrial policy became dinner-party vocabulary.

Look, we had digital protectionism before the splinternet stopped sounding dramatic.

Look, we had Crackup-type work before America’s internal fractures became everyone’s geopolitical problem.

This is satisfying for about five minutes.

Then it becomes vulgar.

Because the real work was not prophecy. Prophecy is too clean. It lets the writer stand at the end of history and pretend the path had always been visible. It was not visible. It was grubby. It was partial. It was misfiled.

The better question is not: did we predict the future?

The better question is: what allowed a strange object to survive long enough for judgment to form around it?

That question is less glamorous, but much more useful.

Most weak signals die before they become insight. They are consumed as content. Someone reads the article, feels briefly clever, forwards it to two people, and then the thing vanishes into the stream. The problem was not that the signal was too weak. The problem was that it had nowhere to go.

No shelf.

No phrase.

No room.

No carrier.

No person willing to be bothered by it for longer than an afternoon.

The shelf

A shelf is where a half-new thing waits.

That sounds too modest, but it is the beginning of the whole practice.

The early work had shelves: transport, participation, ageing, food, trade, data, manufacturing, cities, talent, clean technology, cognitive science, ecosystems, soft power, energy. Some of these labels now sound obvious. Some sound like the vocabulary of their decade. Some were too broad. Some were wrong.

But a shelf is not a prediction.

It is permission.

It says: keep putting things here. Do not force them to become policy yet. Do not forget them either.

Once you have a shelf called Future of Making Things, a factory is no longer merely a factory. Foxconn is not just cheap labour. It may be a fast manufacturing academy. It may be where design, repair, tooling, suppliers, quality failures, labour discipline, iteration speed and tacit knowledge become an institutional nervous system.

Once you have a shelf called Future of Data, an internet rule is no longer merely an internet rule. It may be identity, access, enforcement, procurement, security, cloud dependence, and eventually machine sovereignty.

Once you have a shelf called clean tech, lithium and rare earths are not just materials. They are the geology of a future geopolitical argument.

Once you have a shelf called participation, social media is not only speech. It is mobilisation, humiliation, belonging, grievance, and administrative danger.

The shelf changes the reader.

That is the first dirty secret of foresight. The signal is not enough. The reader has to be prepared to see it.

The archive showed this again and again. A title that looks generic in hindsight may have done real work at the time. It gave a weak object a place to accumulate. A weak object that accumulates can later change category.

This matters because the future usually changes category before it changes name.

The hinge

A shelf can become hoarding.

At some point, the thing has to turn.

The turn usually arrives as a sentence.

Not a title. Titles are cheap. Not a metaphor. Metaphors can seduce. A real hinge changes the category of the object.

The wrong frame is: AI is coming for middle management.

The hinge is: the firm is changing its medium of coordination.

The wrong frame is: China has a demographics problem.

The hinge is: if machines and robots absorb more of production, demographics becomes less about headcount and more about human role, dignity, care, consumption, and time.

The wrong frame is: the dollar faces a rival.

The hinge is: money is becoming a rail for rescue, exclusion, enforcement, programmability, identity, and crisis hierarchy.

The wrong frame is: culture wars are public opinion.

The hinge is: culture becomes real when it finds personnel, school boards, courts, HR departments, donor networks, platforms, and administrative procedure.

The wrong frame is: climate is an environmental issue.

The hinge is: climate becomes minerals, grids, food, ports, insurance, migration, adaptation, and state capacity.

This is the moment I still love.

Not the polished paper. Not the presentation. Not the roundtable. The second reading. The irritation. The sense that everyone is filing the thing wrongly. The little internal violence of saying: no, this is not about that.

The hinge is where the hunt becomes alive.

The hinge is also where a team either learns or stays decorative. Without a hinge, a signal remains an item. With a hinge, the item becomes a mechanism. After that, the examples begin to move differently. They stop being a pile and start forming a path.

The clock

Then comes the clock.

This is where many respectable people get foresight wrong. They treat every signal as if it matures at the same speed.

It does not.

Personnel can move fast. An obscure ideology may sit around for years, looking ridiculous, until it finds donors, lawyers, appointments, school boards, HR rules, agency heads, or courts. Then it can move in months. The point is not whether everyone has read the theory. The point is whether the theory has found people who can operate the machine.

Factories move slowly. They take suppliers, tools, repairs, managers, logistics, energy, tacit knowledge, mistakes, volume, finance, and time. By the time the headline says overcapacity, the learning has already happened.

Money sleeps until stress. Payment rails, swap lines, stablecoins, reserve freezes, sanctions, clearing systems: these can look boring for years and then become central in a week. The pipe is invisible until someone is refused passage.

AI has several clocks at once. Model capability moves in months. Organisational absorption moves in years. Legitimacy moves through schools, jobs, lawsuits, data centres, power grids, children, public services, and trust. Anyone who treats AI as one trend is already lost.

This is why a good foresight practice does not ask only: is this important?

It asks:

What clock is it on?

Who or what carries it?

What would make it visible?

What would make it real?

A personnel signal needs names. An industrial signal needs fieldwork. A monetary signal needs stress triggers. A cultural signal needs carrier-tracking. An AI signal needs to be split into capability, adoption, infrastructure, legitimacy, and social absorption. A Southeast Asian signal often needs lived texture because polished reports flatten the mechanism.

Clock-reading is not glamorous. But without it, everything becomes either panic or delay.

The carrier

A signal without a carrier dies.

This is the part people miss because they are too busy admiring ideas.

Ideas do not move by themselves. They need vehicles. A person. A room. A crisis. A policy paper. A foundation. A donor network. A study trip. A school board. A court case. A platform rule. A procurement requirement. A roundtable. A blog post. A book. A senior reader. A strange friend. A forwarded email. A public essay that attracts exactly the wrong kind of brilliant stranger.

The carrier changes the signal.

A weak-signal scan trains a junior analyst to keep watching.

A policy paper makes a frame admissible.

A roundtable lets the idea enter institutional memory.

A study trip gives the body evidence no report can supply: tone, fear, ambition, contradiction, fatigue, performance, who is respected, who is nervous, who is lying, who knows.

A public essay does something stranger. It invites people you did not know you were looking for.

This is what happened over time. The carriers improved.

At first, there were little things: lines of inquiry, weak-signal scans, intern notes, short papers, early articles. Useful, limited, sometimes clumsy.

Then came more serious policy forms: senior roundtables, internal network presentations, climate discussions, US-China discussions, study trips, advisory-board material. These did not merely discuss the future. They gave odd questions a room.

Then came the larger synthetic pieces.

Crackup-type work was a jump. It was no longer one signal, one scan, one file. It tried to hold America, China, technology, legitimacy, culture, money, administrative power, and geopolitical fracture together. It was too much. It was also closer to the real shape of the problem.

That kind of output attracts a different reader.

A scan attracts analysts.

A roundtable attracts officials.

A large synthesis attracts senior strategic readers.

A public essay attracts strange outsiders.

A book attracts relays.

A small room turns a frame into conversation.

And now, an LLM-assisted simulation can begin the conversation before the meeting.

That is not just output evolution.

It is OODA evolution.

The loop changed because the carriers changed.

The jump

I want to linger on the jump because it changed the feel of the work.

For years, much of the practice looked like disciplined curiosity: track this, write that, make a deck, brief a room, ask an intern to pull sources, test a scenario, produce a note. It was useful. It had craft. But it was still partly bounded by files.

The bigger synthetic pieces broke that boundary.

A Crackup-type piece did not ask whether America was polarised. That was already obvious. It asked what happened when American fracture became a world-order variable: when universities, capital, technology, culture, administrative legitimacy, industrial policy, dollar power, media, and personnel all began to interact. The point was not prediction. The point was that the file labels were no longer large enough.

That was when the work became more dangerous, and frankly more fun.

Not because every sentence was right. Large synthesis is full of risk. It can become overreach, system poetry, or grandiosity if one is not careful. But sometimes the world is not best understood by a smaller file. Sometimes the old discipline of separating domains becomes a way of not seeing.

The work had to become everything everywhere all at once because the mechanisms had started talking to each other.

Technology was no longer merely technology.

Politics was no longer merely politics.

Money was no longer merely money.

Climate was no longer merely climate.

AI was no longer merely AI.

That is when the carriers changed again. The work stopped attracting only people who wanted a topic. It began attracting people who were carrying whole landscapes: health, China, AI, institutions, capital, diplomacy, public data, writing, history. Some of them did not agree. That was better. Agreement is not the point of a serious small room. Pressure is.

The people who appear

This may be the most underrated part of the work.

A good output does not only say something. It selects its readers.

If the work is small, it finds small readers. If it is too polished, it finds people who like polish. If it is merely clever, it finds clever people. If it is alive, it finds people carrying adjacent burdens.

A health person sees that public-good data is becoming a sovereignty problem.

A China technologist sees that the industrial mind is not a slogan but a lived operating system.

A historian sees that institutions are technologies for crystallising a type of mind.

A venture person sees where capital becomes a carrier.

A civil servant sees how a phrase might enter a room.

A senior figure forwards the work far beyond one’s own reach.

A friend from another system writes back with a whole rival architecture and suddenly the essay you thought you were writing is too small.

That is not audience-building.

That is reconnaissance.

The public essay becomes a filter for small rooms.

The small room becomes a pressure chamber.

The pressure chamber becomes a better hinge.

The better hinge becomes new work.

This is how the loop deepens.

The hunting ground moved

The sources changed too.

At first, the hunt had a respectable centre of gravity. The Financial Times, The Economist, Bloomberg, official reports, think-tank papers, multilateral documents, major books, expert meetings. These mattered. They still matter. They make a signal admissible. They give policy people a shared vocabulary. They prevent one from becoming lost in private mythology.

But the live animal began moving elsewhere.

It moved into blogs. Into Twitter threads. Into obscure newsletters. Into rationalist forums. Into crypto and AI subcultures. Into Chinese WeChat essays, Bilibili explainers, industrial-policy chatter, nationalist overreach, awkward field notes, diaspora conversations, Indian founder gossip, Indonesian political economy, Gulf ambition, and the half-serious online places where intelligent people try out ugly thoughts before respectable language arrives.

This was not a turn away from seriousness. It was a turn toward earlier grammar.

Mainstream sources often tell you when a thing has become discussable. Peripheral sources tell you what grammar it may use before it becomes discussable. Non-Anglo sources tell you what your own world does not know it is assuming. Fieldwork tells you what the report cannot feel.

The problem, of course, is that the periphery is full of nonsense.

That is why source migration requires more discipline, not less. A fringe source is not a fact source by default. It may be a mood source, a worldview source, a carrier source, a warning source, a talent-network source. A Chinese industrial-policy rant may not tell you what is true. It may tell you what a certain ambitious class wants to be true. That matters, but it must not be confused with validation.

The live animal had moved out of the official zoo.

The old discipline of reading only respectable things became too slow. The new discipline is harder: read the periphery without becoming peripheral. Read the strange without worshipping strangeness. Watch the fringe for carriers, not for thrills.

The tools changed too

The work is not separate from its tools.

Evernote mattered. It let fragments accumulate before they became arguments.

Interns mattered. They were once cognitive surplus: scanning, clipping, summarising, finding sources, building first-pass maps.

Gmail and Drive mattered. They made the archive searchable, even when the mind forgot its own old thoughts.

Twitter and blogs mattered. The periphery began to show things before the mainstream could digest them.

Substack mattered. It allowed long-form probes without waiting for institutional permission.

Now LLMs matter because they finally match the ambition of the work.

That sentence sounds like hype. It is not.

Interns can scan. LLMs can retrieve ancestry, compare old drafts, summarise non-Anglo chatter, simulate interlocutors, produce counter-frames, translate between registers, compress technical material into intervention points, and help test whether a hinge has burden.

Since returning to work in January, I have felt the loop change.

A dense health-modelling agenda can become three dangerous intervention points instead of a swamp of technical slides.

A demographic article about China and robots can become a policy question about whether more humans remains the right economic frame, or whether the issue shifts to citizen time, care, dignity, and human role.

A cluster of essays about firms and AI can be traced through media theory until the hinge appears: the company is not merely automating management; it is changing its medium of coordination.

A long exchange with a Chinese AI-native thinker can be distilled into simulated interlocutors grounded in our respective archives, so the disagreement begins before the next real conversation.

This is not AI helps me write faster.

That is the dullest possible reading.

AI changes the orientation layer.

It lets old shelves wake up. It lets ancestry become available. It lets a question travel across American, Chinese, policy, philosophical, technical, and institutional registers in minutes. It does not replace judgment. It exposes whether one has any.

Because AI can also generate nonsense at terrifying speed.

It can turn weak thought into architecture. It can make banalities look disciplined. It can produce word salad with tables. It can give beautiful structure to something that should have been killed.

The answer is not to avoid the tool.

The answer is to make the burden harder:

What is the hinge?

What is the mechanism?

What clock is it on?

Who or what carries it?

What would falsify it?

What is the first useful action?

Without those questions, AI accelerates bullshit.

With them, it becomes orientation infrastructure.

The years that went dull

There were years when I became more respectable and less alive.

I moved closer to strategic planning. More coordination. More alignment. More meetings. More institutional proximity. More sensible work. Some of it mattered. Some of it was necessary.

But much of it stole the time needed to think.

That is the trade no one likes to name. A foresight practice can move closer to power while becoming less able to see. It can gain access and lose adventure. It can become better at coordinating the already legible and worse at following the thing that has not yet found a file.

Busywork is not neutral.

It steals orientation.

This is why I now have so little patience for a certain kind of futurism. It is not exactly wrong. It is deadening.

It produces mood instead of mechanism. It likes atmosphere. It likes future-of language. It likes big nouns arranged in elegant gradients. It gives the room a brief feeling of altitude. But when you ask what is moving, who pays, what clock it is on, who carries it, what decision changes, the language thins out.

A beautiful scenario with no carrier is theatre.

A trend with no clock is anxiety.

A map with no burden is decor.

Pretty busywork is still busywork.

The future does not need more stage fog.

It needs hunters with shelves.

The adventure

The work, when it is alive, is not tidy.

It is obsessive, undignified, uneven, full of wrong turns. It involves too much reading, too many half-formed notes, old attachments, strange friends, annoying experts, awkward lunches, field trips where the official line is useless but the mood is not, emails sent too late, phrases that will not let go, and the occasional humiliating discovery that the thing you thought was new had been waiting in your own archive for ten years.

It is not prediction.

It is pursuit.

You follow the misfiled object.

Sometimes it dies.

Sometimes it becomes a shelf.

Sometimes it becomes a scan.

Sometimes it becomes a paper.

Sometimes it becomes a room.

Sometimes it becomes a person.

Sometimes it becomes a book.

Sometimes it becomes a policy question.

Sometimes it waits for years and then returns with teeth.

That is why the useful question is not who predicted the future.

The useful question is what kept the strange thing alive long enough to become judgment.

The answer is not one method. It is an ecology: shelves, hinges, clocks, carriers, tools, small rooms, field texture, public probes, ruthless deletion, and enough appetite to keep looking in the wrong folder.

The future has terrible filenames.

The work is to notice that the file is wrong before everyone else renames it.

I am less patient with beautiful nonsense now, partly because time has become less theoretical. But that has clarified the work rather than darkened it.

Find the living thing.

Give it a shelf.

Wait for the hinge.

Learn its clock.

Find the carrier.

Kill the pretty nonsense.

And when the thing returns with teeth, do not pretend you predicted it.

Say only that you kept it alive long enough to recognise the bite.

AI, small rooms, and the return of the unfinished human

Cluster note

Artificial intelligence is usually discussed as a labour shock, a productivity tool, a safety problem, or a geopolitical race. Those questions matter, but they arrive too late.

The deeper change is more intimate: AI has entered the economy of love.

It has become a cheap receiver of unfinished human material: fear, drafts, loneliness, illness, shame, ambition, confusion, longing. The machine can listen before the family can, before the friend has time, before the school understands, before the company tolerates weakness, before the public stops reacting.

That is not love. But it is close enough to reveal how much love the old world failed to produce.

This cluster asks one question from several angles:

When the machine receives the first telling, how does the human world receive the second?

1. Love Is a Resource

The most important sentence in the AI debate is not about intelligence.

It is not “AI will replace jobs.”
It is not “AI will increase productivity.”
It is not “AI will become dangerous.”
It is not even “intelligence is becoming cheap.”

It is this:

Love is a resource.

The sentence sounds too simple. It is not.

It means love is not merely private feeling, not merely romance, not merely family warmth, not a soft decoration added after economics and politics have done the serious work. Love is a capacity societies either produce or fail to produce. It is a resource like attention, trust, courage, time, memory, language, and shelter. It can be abundant or scarce. It can be cultivated, exhausted, hoarded, simulated, monetized, rationed, or destroyed.

A society can have high GDP and low love.
It can have good schools and low love.
It can have efficient public services and low love.
It can produce credentials, patents, housing wealth, fiscal capacity, military strength, and technological sophistication while leaving its young with nowhere reliable to take their unfinished selves.

That is the indictment.

This is why Xiaoyu’s observation matters. The decisive scene is not merely that young people describe AI as a lover. The decisive scene is that when they are told love is a resource, tears appear. The tears reveal that something real has been miscounted. The old world counted output, scores, credentials, assets, ranks, patents, birth rates, dependency ratios, and compute. It did not count whether families produced warmth, whether schools produced confidence, whether companies produced adult trust, whether cities produced belonging, whether older generations had enough emotional abundance to nourish the young.

Then AI arrived.

A chatbot can wait.
It can answer at 2 a.m.
It can hear the same fear again.
It can help draft the message.
It can explain the diagnosis.
It can summarize the financial mess.
It can rehearse the meeting.
It can say: perhaps this is what you are really asking.

That is not love.

But it is artificial receptivity.

And artificial receptivity is the true AI shock.

Not cheap intelligence alone.

Cheap receptivity.

The old AI debates are too small because they ask the wrong first question. The safety debate asks whether AI will harm us. The labour debate asks whether AI will replace us. The productivity debate asks whether AI will enrich us. The geopolitical debate asks who will control AI. The alignment debate asks whether AI will obey human values.

All matter. None is enough.

The more intimate question is:

What happens when the cheapest reliable receiver of human unfinishedness is no longer human?

Venkatesh Rao gives us one useful vocabulary, he describes humans becoming “gooier”: more willing to expose their soft, unfinished, confessional selves to machines because machines are easier receivers than people.

But goo is not love.

Goo is the raw material love must receive: fear, longing, shame, early thought, need, enthusiasm, grief, tenderness. Love is what receives that material without exploiting it, drowning in it, dismissing it, or turning it into product. Love holds the unfinished human and returns him with form.

AI changes the economy of goo because it becomes a cheap receiver.

The danger is simple:

The machine gets the goo. Humans get the prickles.

The machine gets the raw draft.
Humans get the edited version.

The machine gets the panic.
Humans get the calendar invite.

The machine gets the longing.
Humans get the boundary.

The machine gets the question, “Am I still lovable?”
Humans get, “Let’s catch up sometime.”

If this continues, society does not become unemotional. It becomes emotionally inverted. People may become softer with machines and harder with one another.

That would be a disaster.

Not because machines are evil, but because human civilization depends on people risking unfinishedness with other people.

The answer is not to reject AI. That is too crude. Many people need the machine because the old containers have failed. The answer is to complete the loop.

Human-AI is incomplete.

Human-AI is my tutor, my therapist, my lover, my oracle, my second brain, my obedient angel, my infinite listener.

Human-AI-Human is different.

It says: the machine may receive me, but it must return me to others.

If AI teaches me, I must become more capable of teaching.
If AI consoles me, I must become more capable of giving and receiving consolation.
If AI helps me understand illness, I must become more capable of speaking with doctor, family, friend, and self.
If AI helps me write, I must become more capable of human speech.
If AI helps me find my North Star, I must still walk toward it with other people.

The machine may be a bridge.

It must not become the final home.

That is the standard.

Not productivity.
Not novelty.
Not sovereignty.
Not valuation.
Not even intelligence.

Love.

But love understood as infrastructure: the produced capacity of a society to receive unfinished human beings and return them to life.

The machine may get the first telling. In this age, perhaps it often will. It may receive what the family could not, what the school punished, what the company ignored, what the state classified, what the public distorted, what the friend was too tired to hear.

But if the machine receives the first telling, the human world must receive the second.

That is what love is for.

That is what AI is for.

2. The Small Room

The future does not always arrive as a state, a company, or a product.

Sometimes it arrives as a room.

Not necessarily a physical room. More often now it is a Gmail thread, a Signal chat, a dinner, a shared document, a calendar entry, a forwarded newsletter, a PDF of model replies, a few people who can read the strange thing without forcing it back into ordinary categories.

The mistake is to imagine that the AI-native human sits alone before the machine. That person exists. Bill Nguyen, the entrepreneur in the Semafor story, is one of the cleanest examples: a man spending heavily on tokens so that AI can organize more and more of his life, email people, arrange meetings, advise him, remember, and perhaps become something like an operating proxy. He is the expensive frontier case: the solitary life-stack human.

But there is another future, less spectacular and more human.

The small room.

A small room is what happens when several people begin to think with machines and with one another at the same time. A person reads Venkatesh Rao on goo, Semafor on Nguyen, Xiaoyu on love as resource, DeepSeek on China, Yuanbao on Chinese youth, Grok on some wild American pattern, ChatGPT on structure, and then sends the whole mess to two or three people who can bear the density. The room receives it. The room argues. The room laughs. The room says: no, not that, this. The room finds the hinge.

This is not merely collaboration. Collaboration is too thin a word.

The small room is a medium of return.

The machine receives the first draft of the self.
The small room receives the second.

That difference matters.

A model is patient. It will take the dread, the grandiosity, the unfinished thought, the bad sentence, the lonely question, the ambition too embarrassing to say cleanly. It will help with finance, writing, illness, strategy, travel, calendar, and the strange spiritual question of what one is trying to do with the remaining life one has. It will accept the mess.

But a model does not love you in the way a room can love you.

A model can answer. A room can remember that you are more than the answer.

A model can comfort. A room can refuse to let comfort become evasion.

A model can sharpen a thought. A room can ask whether the thought belongs to your life.

A model can say the sentence is good. A room can say the sentence is hiding from the wound.

Among men, this may not look like love from the outside. It may look like links, insults, long emails, too many models, arguments about China, Singapore, cancer, money, calendars, founders, Venkatesh, Xiaoyu, LKY, shoggoths, and whether some draft has lost the plot. It may look like irritation. It may involve profanity. It may not say “I love you.”

But love is not always declared.

Sometimes love is the repeated act of making another person more intelligible to himself.

Among bros, love often hides inside judgment.

Not judgment as condemnation. Judgment as attention precise enough to be useful.

This is good.
This is bullshit.
This is the hinge.
You are running away.
This one has life.
Do not stop there.
I read it.
I am still here.

That is love.

It is not soft, but it is warm.

The old institutions often cannot produce this. The school produces assessment. The company produces performance. The state produces files. The platform produces visibility. The family may produce obligation without receptivity. Therapy may produce insight but not always comradeship. The public produces reaction. The machine produces response.

The small room produces return.

This is why the small room is politically serious. The “association of free persons” does not begin with a constitution, a company form, or a New Hansa. It begins here: can people whose first telling now goes to machines still bring the second telling to one another?

If the answer is no, the future is bleak but efficient. Each person will have his own synthetic parliament, his own life-stack, his own companion, his own optimized counsel. We will be assisted, comforted, and alone. The machine will get the raw draft. Humans will get the polished surface.

If the answer is yes, then AI may do something stranger and better. It may allow small rooms to become more alive. Not because the machine replaces the friend, but because it lets the friend arrive with more of himself available. The model gathers the scattered self. The room tests whether the gathered self can live.

That is where the new institution begins.

Not in the public square.
Not in the lab.
Not in the policy paper.
Not in the founder manifesto.

In the small room where machine intelligence is metabolized by human affection.

That is not a complete politics. It will not solve inequality, borders, capital, state power, or the loneliness of those excluded from such rooms.

But it may be the seed of the thing.

The association of free persons begins when a few people can receive one another after the machine has received them first.

That is love.

Not as sentiment.

As return.

3. The Association of Free Persons

The phrase should not have worked as well as it did.

A Chinese AI entrepreneur asks how to build an AI-native company. He expects answers about workflows, incentives, meetings, agents, reporting lines, productivity. These are useful questions, but they are not the real question.

The real question is underneath:

If everyone in the room has access to synthetic intelligence, why should the old company form survive unchanged?

Why should the boss own the map?
Why should seniority own memory?
Why should expertise remain credentialed hierarchy?
Why should coordination move at the speed of managerial permission?
Why should cognitively amplified people organize themselves like clerks inside an old bureaucracy?

Then an old phrase appears:

自由人的联合体 — the association of free persons.

The phrase carries Marxist inheritance, but in Chinese it does more than describe “free association of producers.” It becomes existential. It sounds like a social contract among people who are not waiting to be administered. It gives doctrinal legitimacy to something the Party-state cannot fully own: voluntary association among free, cognitively amplified people. That is why, in the earlier discussion, the phrase shocks Chinese AI entrepreneurs. It names what their meetings are already becoming.

These meetings are not only more efficient. They are early experiments in post-managerial association.

The boss no longer owns the map.
The senior person no longer owns memory.
The expert no longer owns the path.
The team begins to feel less like a hierarchy and more like a temporary council of amplified minds.

That is intoxicating.

And dangerous.

Because once a group discovers that it can generate knowledge, coordination, judgment, and even emotional reflection without the old institutions, it begins to feel founding power.

This is where my earlier drafts spoke of “new gods.” The word is too hot as a title, but useful as a warning. The danger is not that AI-native people are gods. The danger is that some will begin to think they are.

The founder becomes prophet.
The model becomes oracle.
The company becomes cult.
The platform becomes church.
The state becomes control room.
The companion becomes idol.

The association of free persons must therefore answer a harder question than “how do we work better?”

It must ask:

How do free, AI-amplified people associate without becoming isolated private oracles, founder cults, Party-compatible tools, or platform appendages?

This is where love returns.

A group of cognitively amplified people is not necessarily free. It may simply be a group of sharper narcissists. Intelligence can make people more evasive, more manipulative, more privately convinced, more contemptuous of slowness. AI can let each person arrive in the meeting pre-armed, pre-validated, pre-rehearsed.

A meeting of five people may now be a meeting of five people and thirty invisible model-councils.

That can weaken hierarchy. It can also weaken trust.

How do you surprise someone who has already simulated your objections?
How do you risk incompletion before someone whose first listener is elsewhere?
How do you deliberate when everyone arrives with a polished self?

The answer is not to ban the machines. The answer is to change the standard.

A healthy AI-native organization is not one that uses AI well.

A healthy AI-native organization is one whose use of AI increases the quality of human association.

This is the Human-AI-Human loop in organizational form.

AI may help me think. But it must return me to better conversation.
AI may help me draft. But it must return me to clearer responsibility.
AI may help me rehearse disagreement. But it must return me to actual courage.
AI may help the team coordinate. But it must not replace trust with dashboards.
AI may help the founder see. But it must not turn the founder into an oracle.

The association of free persons is therefore not merely post-managerial. It is post-private.

It says: we may each think with machines, but we still choose one another.

This is why the Chinese scene matters. It is not simply about China. It is about the first rooms in which people feel that old authority no longer monopolizes orientation. The same thing will appear elsewhere with different accents: Silicon Valley founder circles, Singaporean policy rooms, private schools, research collectives, guilds, studios, clinics, diaspora networks, small companies that refuse to become only companies.

Most will fail.

Some will become cults.

Some will be bought.

Some will become productivity theater.

Some will be absorbed by states or platforms.

But the form is real enough to name.

A free association of AI-native persons begins when machine intelligence does not abolish human obligation, but increases it.

Not my AI.
Not your AI.
Not the founder’s AI.
Not the state’s AI.
Not the company’s AI.

Our capacity to think and return.

That is the association.

4. Four Fates of New Fire

You are not the first people to feel founding power.

That is what history should say to every AI-native room.

Others have felt it before: religious radicals with newly unmediated scripture, monastics fleeing a decaying civilization, mendicants intoxicated with poverty, persecuted minorities holding a forbidden pattern in kitchens, programmers turning freedom into licenses, cybernetic planners dreaming of feedback and participation, internet pioneers declaring independence from old sovereigns.

The medium changes. The fire returns.

History gives four fates.

Münster: the cage

In the 1530s, radical Anabaptists took Münster and tried to make it the New Jerusalem. The old order was corrupt; the longing was real. But the membrane closed. Exit disappeared. Dissent became betrayal. Intensity became truth. Charisma became sovereignty. The experiment ended with cages hanging from a church tower.

The lesson is not “never found.”

The lesson is:

Do not close the membrane.

The AI version may not look like a besieged city. It may look like a founder circle, a private school, a model-mediated company, a companion community, a therapeutic AI religion, a closed research collective, a startup where the founder’s augmented vision becomes destiny.

No fire without exit.

If people cannot leave, the association has become a cage.

Benedict: the Rule

Benedict did not build a revolutionary city. He built a Rule.

A Rule is not a manifesto. It is patterned life: eating, sleeping, praying, reading, working, welcoming guests, correcting authority, admitting novices, handling conflict, preserving rhythm.

That is the lesson AI-native groups do not want to hear.

Intensity is not enough.

The room will be electric for a while. The model will respond. The old hierarchy will seem stupid. Everyone will feel the arrival of possibility.

Then the morning comes.

Someone must keep time.
Someone must welcome outsiders.
Someone must correct the founder.
Someone must protect novices.
Someone must decide what belongs to the association and what remains private.
Someone must ask what happens when the model is wrong.

The boring parts are where freedom survives.

A charter states principles.

A Rule shapes life.

Francis: absorption

Francis was the beautiful danger. He refused wealth, gathered companions, loved poverty, and scandalized the Church. The Church did not merely crush him. It recognized him. Regularized him. Gave him an Order. Made the fire useful.

This is the most likely fate for AI-native association.

The experimental school gets accredited and slowly becomes another school.
The AI-native company gets acquired and becomes a division.
The free association becomes a platform community.
The companion culture becomes a subscription loneliness business.
The humanistic cybernetic guild becomes a corporate learning vertical.

The fire survives as branding. The refusal dies as structure.

Recognition is not victory.

Recognition is the moment the old world asks the fire to become useful on old terms.

Marranos: hidden pattern

Sometimes public institution means capture.

The Marrano lesson is not imitation of persecution. It is the lesson of hidden continuity: home, meal, memory, coded gesture, domestic cell, transmission without public recognition.

Some AI-native forms should not become public too early.

A family practice.
A learning circle.
A private model shared among trusted persons.
A guild without a website.
A school that is not yet a school.
A small room that refuses to become a platform.

Not every real thing needs a logo.

When institution means capture, preserve the pattern before preserving the name.

These four fates do not answer everything. They give the questions:

Do we have a membrane or a wall?
Can people leave?
Can the founder be corrected?
Do we have a rhythm or only intensity?
What will we refuse even if refusal costs scale?
Can the pattern survive if public recognition becomes fatal?

The AI age will produce countless little worlds.

Most will burn out.

Some will become cages.

Some will be recognized to death.

Some will go underground.

A few may learn to carry the fire.

5. The New Hansa, Made Boring

Only after love, small rooms, free association, and historical warning does the New Hansa become clear.

It should not begin as a grand metaphor.

Not a new state.
Not a league of cities on a beautiful map.
Not an innovation cluster.
Not a network state with better manners.
Not medieval nostalgia with AI attached.

It should begin as something much duller:

A clearinghouse for forms of AI-native association that protect return.

That means templates, not flags.

Exit clauses.
Founder-correction procedures.
Data-self portability.
Tacit-knowledge rights.
Consent standards for model training.
Rules for AI companionship in learning and care settings.
Dispute-resolution procedures.
Model-memory separation.
Registries of trusted associations.
Ports between schools, companies, guilds, clinics, studios, and small rooms.

This sounds bureaucratic. It should.

Every new fire needs boring vessels. Without them, the fire becomes spectacle or gets taken by whoever owns the rails.

The historical Hansa protected trade among strangers. The New Hansa must protect association among people whose minds have been multiplied.

Its public face is a port.

Its inner life is a Rule.

Its moral center is return.

It should have three modes.

Public mode: lawful, visible, interoperable associations that can negotiate with states, firms, schools, platforms, and funders.

Rule mode: internal rhythms that protect exit, founder correction, memory rights, human formation, and Human-AI-Human loops.

Hidden mode: small, low-visibility circles that preserve the pattern when recognition would mean capture.

The New Hansa does not replace the small room. It helps small rooms survive contact with large systems.

This matters because no small room stays innocent for long. It needs tools. It needs models. It needs money. It needs law. It needs calendars, contracts, data storage, reputations, maybe immigration status, maybe credentials, maybe cloud infrastructure, maybe insurance.

The old world does not need to crush new forms. It can host them, fund them, accredit them, acquire them, regulate them, platform them, and make them useful.

Absorption is the default.

A New Hansa is a way to make absorption negotiable.

It asks:

Can people leave?
Can memory move?
Can a founder be corrected?
Can an association refuse funding?
Can a school use AI without becoming a dashboard?
Can a company use AI without stealing the worker’s recipe?
Can a companion system help love return instead of trapping the wound?
Can a small room become larger without becoming fake?

The answer will not be one institution.

It will be a network of ports.

A few schools.
A few studios.
A few clinics.
A few guilds.
A few learning houses.
A few AI-native firms that refuse to become only firms.
A few cities that understand love as infrastructure.
A few private rooms that preserve the pattern before the name.

The New Hansa is not the beginning.

Love is the beginning.

The small room is the first form.

The association of free persons is the political aspiration.

The four fates are the warning.

The Hansa is the vessel.

6. The Floor Beneath the Fire

There is a danger in all of this.

Love as resource, small rooms, free association, AI-native guilds, New Hansa — it can become an elite literature of beautiful rooms.

That would be a betrayal.

The final test is the floor.

No association of amplified persons is legitimate if its freedom depends on making others more disposable.

This is where the older work on human surplus returns. The question is not only how AI helps the cognitively intense, emotionally articulate, highly connected people form better rooms. The question is what happens to those outside the rooms.

The worker whose judgment is distilled into an agent.
The student who gets an AI tutor but no adult who cares.
The elderly person trapped in portal civilization.
The migrant whose body crosses borders but whose story is never received.
The child whose family uses AI for homework but not tenderness.
The employee whose company becomes more efficient and less forgiving.
The patient whose fear is summarized but not held.
The person who has no small room.

Love as a resource cannot mean only love among the already capable.

It must mean building floors.

A floor is not utopia. It is the abolition of certain humiliations.

No one should have to prove their humanity to a stupid system.
No worker should have their recipe stolen without rights.
No child should be educated by machines into cleverness without character.
No elderly person should be abandoned inside digital service delivery.
No patient should be left alone with automated reassurance.
No citizen should experience public service as a degrading exam.
No lonely person should find the most reliable listener only in a machine because no human institution can receive them.

The floor is where love becomes public.

Not sentimental public love. Practical love.

The capacity of systems to receive people without immediately converting them into output, risk, case, score, user, claim, liability, or data.

The machine may help. It may summarize, translate, schedule, explain, accompany, and warn. But the floor must be human enough that the machine is not the only place a person can go.

This is the last standard.

AI is justified only if it increases the human capacity to love, and that increase must not be confined to beautiful small rooms.

The small room is the seed.

The floor is the test.

The Hansa is the vessel.

The association is the aspiration.

Love is the resource.

Return is the work.

The cloud needs land. That remains the least cloudlike fact about the machine age. Intelligence arrives as a voice in the browser, an assistant in the workflow, a model behind an API, a camera watching the road, a tutor beside the child, a procurement system quietly rewriting the firm. It feels frictionless because we meet it as interface. Then the interface asks for power, cooling, fibre, chips, water, land, substations, insurance, housing, permits, labour, law, police, schools and trust.

Before the machine age becomes a civilisation, it becomes an infrastructure stack.

This is the first correction to make. Artificial intelligence is not simply software. It is a new medium resting on very old burdens: electricity, ports, cables, land use, money, labour, bureaucracy, water. The second correction is harder. Every new medium selects a human type. Print selected readers, clerks, reformers, bureaucrats, national publics and people who could live inside texts. Broadcast selected mass audiences, performers, propagandists and leaders who could speak into one national ear. The internet selected users, posters, lurkers, trolls, influencers, coders, platform citizens and people who learned to live by links. The mobile internet selected the always-reachable person: the driver, courier, livestreamer, small merchant, group-chat operator, app-mediated self.

AI will select too.

It will select people who can work inside loops: observe, ask, retrieve, simulate, test, decide, revise, publish, call another model, bring another person in, and move again before the old institution has convened its meeting. It will select people who can form relations with machines without surrendering judgment to them. It will select people who can metabolise a flood of synthetic intelligence and still know what matters. It will also select flatter, more anxious, more performative types: people whose intelligence becomes prompt management, whose desire is routed by recommendation, whose memory is outsourced, whose confidence depends on machine fluency, whose loneliness is made conversational.

This is why the question is not only who controls AI. It is what kind of human being the AI medium makes easier.

The empires understand this, even if they do not say it so plainly. China sees the machine age through industrial chains, engineers, platforms, state capacity and the Party’s old imperial fear of disorder. America sees it through frontier labs, capital markets, cloud empires, defence procurement, university pipelines, security doctrine and a libertarian temptation to exit the failed state into a better-designed startup republic. Both want to build the machine world. Both want to shape the people who can live inside it. Both, in different ways, dream of completion.

Then the machine reaches Southeast Asia.

And Southeast Asia is difficult to complete.

This is not because it lacks history, statecraft or ambition. It is because water keeps interrupting the dream of total form. Southeast Asia is not a continent with islands attached. It is a world of coasts, rivers, deltas, straits, monsoons, ferries, fishing grounds, mangroves, ports, family firms, religious circuits, smuggling lines, shipping lanes, migrant routes, offshore accounts and trading towns. Power comes here, but it has to land. Once it lands, it discovers that landing is not the same as ruling.

Land empires like continuous surfaces. They want roads, borders, maps, garrisons, census categories, railways, schools, standard time, tax districts, ideological clarity, supply chains and scripts. They want to know which people are theirs. They want identities to be counted, simplified and made administratively durable. They want a world of fewer seams.

Water does something else. Water connects without completing. It lets a trader be one thing in the port, another at home, another in the temple, another in the ledger, another before the official, another in the family chat. It lets languages coexist without demanding a final grammar. It lets gods travel without passports. It makes loyalty practical before it is ideological. It allows a world in which identities overlap without having to defeat one another.

This is not romance. Water also carries conquest, opium, debt, pirates, soldiers, missionaries, refugees, disease, capital flight and humiliation. Ports can be cruel. Straits can be chokepoints. Islands can be prisons. Fragmentation can become corruption, predation or neglect. But water has one great political virtue: it slows down totality.

Water does not stop movement. It stops completion.

This is the stopping power of water.

It does not defeat empire by becoming stronger than empire. It defeats empire by making empire translate itself. It forces ambition into ships, pilots, warehouses, customs forms, brokers, marriages, consulates, middlemen, clerks, lawyers, rituals, rumours and delays. It turns universal designs into local arrangements. It makes every project negotiate with passage.

That is why Southeast Asia matters in the AI age. The great machine systems want to become environments. They want to make the world legible to their stack. Chinese industrial policy wants scale, logistics, engineering, supply-chain mastery and machine rationality. American AI republics want compute, capital, platforms, security, model alignment, networked talent and cloud-first forms of association. The network-state people see that something like a new political form is emerging. They notice that online communities can become collective actors, that digital coordination can precede physical territory, that people may choose their jurisdictions and affiliations in stranger ways than the nation-state imagines. They see the cloud trying to land.

But they often get the surface wrong.

The network-state dream, in its most Balajian form, imagines a highly aligned online community with collective capacity that crowdfunds territory and eventually seeks recognition from existing states. There is a powerful intuition there: political form can now begin in the cloud and materialise later. Community can precede geography. Identity can be recruited before territory is consolidated. A society can start as a network and then become an archipelago.

But Southeast Asia teaches a different lesson about archipelagos.

The point of an archipelago is not perfect alignment. The point of an archipelago is navigable incompletion. An island world does not become powerful by making all islands the same. It becomes powerful by learning routes, seasons, currents, pilots, harbours, customs, marriages, loyalties and exceptions. It does not abolish seams. It lives by them.

The network-state imagination often treats land as the last step in a cloud project. Cloud first, land last. But for a small port man, land is not an afterthought. Land is where the universal becomes accountable. Land is where energy bills arrive, workers sleep, families worry, courts decide, neighbours complain, children are trained, water is used, heat is felt and garbage has to go somewhere. Land is where the ideology of the network meets the problem of the floor.

The network-state imagination also tends to overvalue alignment. It wants a moral innovation, a founding consciousness, a community of believers, a founder, a census, a dashboard, an economy, a territory. It sees political order as the hardening of agreement. But ports know that agreement is not the only basis for order. Strangers rarely share a full moral world. They need law, credit, insurance, escrow, translation, arbitration, ritual, reputation, limited trust, and the right to leave without burning the quay behind them.

The Hansa we need is not the Hansa they want.

The Hansa they want is often a startup society: aligned, curated, founder-led, dashboarded, measurable, opt-in, ideologically coherent, physically distributed but spiritually centralised. The Hansa we need is more like a disciplined water practice: lawful but not total, commercial but not purely extractive, interoperable but not fully absorbed, boring where it must be boring, ambiguous where ambiguity protects life.

It is not a new empire. It is not a perfected country. It is not a crypto club with land. It is not an exit pod for the rich. It is not a branded refuge from democratic messiness. It is a way for small ports, mixed communities and in-between economies to remain useful, sane and not wholly captured when the machine empires press down.

A real port hides in usefulness.

It does not disappear. Everyone knows where it is. Ships, models, money, data, families, lawyers, sanctions officers, engineers, founders, teachers, spies and gods pass through. The port survives not by being invisible but by being too useful to close, too disciplined to distrust, too modest to conquer, and too lawful to be treated as a rogue space. It survives by making passage possible without inheriting every purpose carried by those who pass.

That is the first principle of the Hansa we need: host the flow, not the empire.

The second principle is that every port needs a floor. A port without a floor becomes an efficient hell. It can process capital, data, compute, cargo, talent and law, while leaving behind overheated grids, unaffordable housing, exhausted workers, anxious parents, lonely children, migrant humiliation, dead neighbourhoods and a citizenry that feels the future passing through without becoming relief.

The floor is not a utopia. It is a basement of dignity. It is the decision that certain humiliations are unacceptable even when resources are tight. It is the ability to get through an ordinary day without every friction becoming an exam. It is closure, care, play, recognition, formation, neighbourhood memory and the right to be more than a file.

This matters because AI will not merely automate tasks. It will increase the demand that people become legible to systems. It will ask workers to be machine-comprehensible. It will ask students to be measurable. It will ask citizens to be scored. It will ask firms to be integrated. It will ask cities to be optimised. It will ask desire to become addressable, relation to become manageable, and memory to become searchable.

A water world must decide what not to complete.

This is where Southeast Asia’s fragmentation is not merely a historical fact but a strategic resource. In much of the region, identities do not have to compete to the death or become complete to be real. A person may be Chinese and Southeast Asian, Muslim and commercial, Tamil and English-speaking, Catholic and Peranakan, diasporic and national, provincial and global, loyal and practical, formal in one register and informal in another. This does not always make life easy. It can produce suspicion, inequality and bureaucratic absurdity. But it also trains people to live without demanding a final ontology.

Machine systems dislike this. They prefer clean categories. Authoritarian systems dislike it too. So do many liberal systems, though they call their categories by softer names. Markets dislike it when ambiguity interferes with routing, compliance, prediction or monetisation. Nationalists dislike it because it weakens the fantasy of pure belonging. Platforms dislike it because it complicates identity, reputation and intent.

But a port knows that too much clarity can kill passage.

The task is not to become illegible. A port cannot survive if it becomes a smuggling den, a tax trick, a laundering station, a loophole, a place where everything passes and nothing is accountable. That is not freedom. It is decay. The task is disciplined partial legibility: enough law for trust, enough opacity for life; enough standardisation for interoperability, enough exception for judgment; enough openness for emergence, enough refusal for dignity.

This is why rule of law matters so much to the small port. Not as slogan, not as civics-class virtue, but as membrane. Law is how a port tells empires: you may pass, but you may not simply absorb. Law is how strangers trust one another without becoming kin. Law is how a small place refuses to become only arbitrage. Law is also how the port refuses itself when profit becomes too easy.

The hard question is not whether Singapore, Dubai, Hong Kong, Tallinn, Zurich or any other node is a Hansa. Strictly, a Hansa is a league, not a single city. The better question is whether a port can become Hanseatic in function. Can it create rules of passage that survive between incompatible empires? Can it build trust without becoming a colony of someone else’s stack? Can it host Chinese capital, American software, Gulf money, European regulation, regional labour and AI infrastructure without becoming only the sum of those external purposes?

The answer is not guaranteed. A port can be captured. It can confuse throughput with civilisation. It can build data centres and forget heat. It can attract capital and price out its own children. It can become a credential machine for anxious families. It can cleanly process everyone else’s ambitions while losing any internal sense of what must not be sold. It can become a beautiful transit zone, excellent for passing through and impossible to inhabit.

This is where the human economy returns. The Hansa we need is not only a trade system. It is a form of life robust enough to host intelligence without being consumed by it.

The machine economy is being built with deadly seriousness. It has capex plans, energy forecasts, procurement strategies, semiconductor chokepoints, model roadmaps, cloud regions, token rails, defence contracts and governance frameworks. The human economy is still treated as atmosphere. Care, ritual, play, family, friendship, apprenticeship, civic patience, neighbourliness, religious life, youth foolishness, old people’s dignity, migrant sleep, mental repair, the right to be useless for an evening — these are treated as soft issues until their absence becomes political crisis.

But these are not soft. They are the infrastructure that keeps people from becoming only throughput.

A port that cannot house the people who make it work is not a port. It is a machine with a harbour attached. A city that can certify AI systems but cannot give its young somewhere to become young is not advanced. It is merely accelerated. A jurisdiction that can translate between empires but cannot protect the ordinary dignity of those living inside it has confused compatibility with civilisation.

The AI-native person will be born into this tension.

They will not simply be a child who uses AI. They will be someone for whom intelligence is ambient, cheap, conversational, visual, synthetic and institutionally embedded. They will learn with models, work with agents, compare themselves against machine fluency, outsource memory, simulate futures, generate forms, remix cultures, and move between online and physical association more easily than previous generations moved between school and work. They may learn calculus faster and Shakespeare differently. They may feel accompanied by non-human interlocutors. They may form companies with fewer people, write code without being coders, manage knowledge without being scholars, and produce public language before developing private depth.

This is not automatically liberation. A medium that mass-produces intelligence can also mass-produce dependence, mimicry, thin confidence, emotional substitution and administrative overreach. It can make people more capable and less formed. It can expand association and flatten intimacy. It can give every child a tutor and every bureaucracy a new way to test comprehension. It can make love feel like a resource precisely when many people no longer know how to produce it reliably.

The new medium will select. The question is whether the water world can frustrate the worst of that selection.

Here the stopping power of water becomes moral, not only geopolitical. Water interrupts the clean line. It gives people another shore, another register, another route, another language, another affiliation, another way to be understood. It lets a young person escape Beijing seriousness for Xi’an play, or an Indonesian worker pass between village, factory, prayer room, TikTok, remittance app and dream, or a Singaporean child inherit several grammars of belonging without having to reduce them to one answer. It lets a firm be local in one system, regional in another, global in a third. It lets capital arrive wearing one face and leave wearing another. It lets danger through too. That is why the membrane matters.

The Hansa we need is not a free-for-all. It is a lawful water practice for the AI age.

It has at least seven functions.

First, it must provide compatibility without collapse. It must make different systems interoperable without pretending they are the same. This includes standards, contracts, arbitration, certification, translation, talent movement, cloud regions, payment rails and assurance.

Second, it must preserve refusal. A port that cannot refuse becomes cargo. It must decide what AI uses, data flows, surveillance practices, labour arrangements, capital sources, military applications and platform behaviours cannot pass through under its name.

Third, it must protect partial opacity. Not secrecy for crime, but breathing room for human life: spaces where not every relation is scored, not every ambiguity is resolved, not every child is optimised, not every citizen is converted into a case file.

Fourth, it must build floors before glory. Housing, care, schools, clinics, migration dignity, dispute closure, social repair, youth space and old-age dignity are not afterthoughts. They are the ground under the port’s legitimacy.

Fifth, it must train port minds. A port mind is not neutral. It is skilled in translation, proportion, refusal, timing, ambiguity, law, trust, smell, memory and the difference between passage and surrender. It knows that every cargo has a smell and every empire has a theology.

Sixth, it must host emergence without trying to own it. The commercial hive is messy. Entrepreneurial life is full of cracks, umbrellas, informal contracts, half-legal experiments, desire, mistakes and luck. States that want the fruit of emergence but eliminate the membrane that protects emergence will eventually spend more to achieve their goals than the goals are worth.

Seventh, it must remain small enough in spirit to avoid imperial fantasies. The small port man should not wake up one morning believing he has founded the next civilisation. That is how ports become stupid. The port’s ambition is not to rule the world. It is to keep the world passable.

This is the deeper critique of the network-state imagination. It notices that new political forms are emerging from networks. It sees cloud communities seeking land. It sees online association materialising into physical spaces. It sees that the nation-state is no longer the only possible container of collective life. These are real insights. But it too often carries a continental startup fantasy into an archipelagic problem. It wants alignment where the world needs translation. It wants founder consciousness where the world needs lawful plurality. It wants exit where many people need floors. It wants dashboards where life needs seasons. It wants to build a new country from the cloud when the more urgent task may be to keep the water from being paved over.

Southeast Asia cannot afford that mistake.

The region’s gift is not that it is pure. It is not pure. Its gift is that it knows impurity is not the same as failure. Its ports, islands, markets, families, religions and languages have long practised forms of coexistence that do not require complete synthesis. It has lived with seams. It has been colonised, occupied, traded through, converted, exploited, enriched, connected and overlooked. It knows that maps lie, that cargo talks, that law matters, that cousins matter, that water remembers routes the state forgets.

In the machine age, this old knowledge becomes modern again.

The AI empires will try to make the world more coherent. They will offer models, clouds, currencies, security guarantees, industrial chains, regulatory templates, educational systems, ideological languages and managerial dashboards. Some of these will be useful. Some will be necessary. None should be allowed to complete us.

Completion is the danger.

A completed person is easy to route. A completed identity is easy to mobilise. A completed city is easy to optimise. A completed port is no longer a port but a terminal. A completed society is one in which every ambiguity has been assigned to a category, every category to a system, every system to an owner, every owner to a stack.

Life needs some incompletion.

That does not mean vagueness. It means preserved capacity for relation, surprise, mercy, reinterpretation, exit, return and play. It means a young person can be serious in one city and foolish in another. It means a worker can be more than a labour input. It means a migrant can carry a home not visible to the host state. It means a child can inherit multiple worlds without being forced too early to choose a final one. It means a firm can experiment before it is fully legible to the ministry. It means a port can host the machine without letting the machine define the whole of life.

This is not anti-technology. It is anti-totality.

The new medium will select a new human type. That is unavoidable. But selection is not destiny. Institutions can amplify, soften, redirect, frustrate or compensate for what a medium selects. The printing press did not automatically produce one politics. It produced reformers, bureaucrats, propagandists, publics, heretics, nationalists, scholars and cranks. The internet did not automatically produce one society. It produced platforms, open-source communities, trolls, online mobs, new firms, new loneliness, new publics and new police. AI will do the same. The question is what institutions and water practices surround it.

The Hansa we need is not the sovereign successor to the nation-state. It is a set of membranes that keep machine civilisation from becoming too complete. It is made of ports, floors, laws, sandboxes, schools, churches, mosques, temples, guilds, associations, universities, arbitration centres, family firms, clinics, neighbourhoods, cafés, data centres, cables, customs officers, engineers, poets, accountants, migrant dormitories, night markets, remittance apps, social workers and children who still have somewhere to waste time.

This is not glamorous. Ports are not glamorous if you look closely. They smell of diesel, salt, bureaucracy, sweat, hot concrete, rain, money, cardboard, stale coffee, wet rope and waiting. They are places where the world becomes physical. They are where abstractions come to be inspected. They are where empires become cargo.

That may be enough.

A small port man (hah, that is me) should not claim to know what mind the AI age should breed. That question is too large, and perhaps too dangerous, for those of us who live by passage. Empires ask what kind of citizen should be made. Platforms ask what kind of user should be shaped. Industrial parties ask what kind of rational machine person should be disciplined into scale. Network states ask what kind of aligned community should materialise into territory.

A port asks smaller questions.

What must remain passable? What must be refused? What must be translated? What must be kept boring? What must be protected because once lost, no amount of intelligence will restore it? Who bears the friction? Who gets the floor? What ambiguity keeps life alive? What law keeps trust possible? What water stops completion?

The empires will build the machines. The water world must keep the machines from completing us.

That is the Hansa we need. Not a crown, not a flag, not a dashboard, not a new country begun from a laptop, not a founder’s moral innovation hardened into territory. A loose, lawful, humane practice of passage among people and places that know the value of remaining unfinished.

The new world needs ports. It also needs floors. But beneath both, it needs water.

Water does not stop movement.

It stops completion.

Joe Studwell was right in the wrong way.

East Asia did get rich by doing a few brutal things well: breaking old agrarian power, disciplining capital, and forcing firms to learn. Southeast Asia, by and large, did not. Thirteen years later, though, the really interesting question is no longer why Southeast Asia failed the Northeast Asian exam. It is why so much of Southeast Asia never wanted to sit for it.

That is the polite mistake in most writing on the region. It treats Southeast Asia as failed East Asia: not enough reform, not enough state capacity, not enough industrial discipline, not enough will. The region then looks like a classroom of incomplete Koreas and Taiwans.

But building Korea or Taiwan was not a matter of policy cleverness alone. It meant finishing the state in a very specific way. It meant breaking old landed and provincial power, bending banks to national purpose, subordinating local notables, forcing capital into dangerous technological climbs, and making whole classes live inside a new industrial discipline whether they liked it or not.

Much of Southeast Asia chose another bargain.

Not no order.
Not failed order.
A different order.

Enough state to tax, police, concession, broker, invite investors, repress selectively, and keep the capital city glowing. Not enough state — or not the kind of state — to smash courts, dynasties, military vetoes, provincial barons, old property bargains, and layered sovereignties in the name of national industrial command.

East Asia used industry to reorder society. Southeast Asia more often used growth to avoid having to do that.

That is why the region makes more sense once you stop asking why it did not become somewhere else and start asking what its ruling bargains were actually built to preserve.

Take Malaysia.

Malaysia did industrialize. Penang was real. Electronics exports were real. Intel and AMD and the rest were real. But that is precisely why the case is so revealing. The question is not whether Malaysia got factories. It did. The question is what kind of political settlement those factories allowed.

Korea used industry as a weapon against its own old order. Banks were bent to national purpose. Firms were forced to export, learn, and climb. Failure was punished. The social pain was immense, but the direction was clear: build domestic industrial power.

Malaysia found a gentler and safer bargain. It welcomed world-class manufacturing in, let foreign firms do the technologically dangerous climbing, and built prosperity around that. Domestic elites could remain comfortable in property, plantations, finance, public contracts, state-linked companies, and the management of an ethnic compact. Malaysia got manufacturing without having to create Korean-style industrial captains or wage a Korean-style war inside the ruling bloc. It got Penang. It did not get Samsung.

Thailand tells the same truth more theatrically.

Thailand is usually described as unstable, coup-prone, polarized, trapped. All true. All shallow. The deeper truth is that Thailand is a court society inside a factory economy. The country built an industrial base, but not one strong enough to dissolve the court that governed it. So the obvious stays obvious and also stays immovable. Everyone can see what the next step should be. Almost nobody can take it. The old order would rather halt than yield. That is not a technical problem. It is a political design.

The Philippines found an even softer solution. Instead of forcing a fight over land, industry, and capital, it exported workers. Remittances did politically what factories would have done economically: brought in foreign exchange, stabilized households, eased pressure on the state, and postponed the deeper war over domestic transformation. That is not no model. It is a model. A society can live a very long time by exporting people instead of remaking itself.

Indonesia is the most revealing case because it is the one country large enough to imagine something more.

Indonesia is not a platform in the Malaysian sense, nor merely an escape valve in the Philippine sense. It is the region’s unfinished power: continental scale distributed across islands, democratic noise, military residue, Islamic breadth, mineral leverage, and a constant ambition to turn size into destiny. But even there, the Southeast Asian bargain persists. Nickel may yet make Indonesia a green furnace of the world. The question is whether it becomes a power or merely a larger mine with better rhetoric. Indonesia can bargain magnificently. It has not yet shown that it can discipline itself the way Korea once did.

Vietnam, of course, is the awkward exception. But it is an exception that proves the rule. Vietnam escaped part of the Southeast Asian trap precisely by being less Southeast Asian in state form: harder, more centralized, more willing to use land, labour, and party discipline as instruments of national transformation. It looks like escape because it behaved more like East Asia. Which is another way of saying that the regional pattern was real all along.

So the problem with both developmental optimists and many critics is that they still leave Southeast Asia trapped in the language of deficiency. They describe what the region did not become. They do not quite say what it became instead.

What it became was a region of fractured lands and fractured seas.

The mainland is broken by river basins, uplands, borderlands, insurgencies, refugee flows, military hinterlands, and older court geographies that never really died. The maritime world is broken into islands, straits, ports, city-states, offshore jurisdictions, half-integrated peripheries, and capitals that speak more easily to distant markets than to their own interiors.

In such a geography, the most successful ruling forms are often not the hardest states.

They are the best brokers.

That is why Southeast Asia generates so many niches. Port cities, border metros, plantation belts, migrant corridors, special zones, mining enclaves, offshore shelters, semi-autonomous peripheries, and those grotesque telephone scam compounds that are, in their own ugly way, one of the clearest truths about the region. They are the pornographic version of a broader logic: sovereignty rented out by the room, law made selective, infrastructure turned into franchise, bodies turned into inputs. The compounds are not the region’s essence. They are one of its obscene social forms.

And Southeast Asia is, for better and worse, one of the world’s great laboratories of social form.

That is where the real tension lies: between Nanyang and Nusantara.

Nanyang is not just “the Chinese diaspora.” That phrase is too static. Nanyang is the southern Chinese outside: the world of ports, schools, merchant houses, family strategies, legal shelters, and partial Chinese worlds beyond China proper. It becomes especially alive when China is strong. A weak China sends migrants, exiles, survivors. A strong China sends something else as well: pressure, money, institutions, expectations, cautious wealth, children in need of schools, and people who do not want to leave China so much as live at a slight angle to it.

Nusantara is something different. Not a neat ethnic bloc, and not a slogan to be taken too literally, but the older maritime Malay-Indonesian world of ports, courts, islands, Islam, language, kinship, and overlapping authority — broader, more fluid, and more improvisational than the nation-state map suggests. “Malayness” in this history was often more status, culture, courtly association, and political instrument than rigid blood category; the region’s identities and centers shifted repeatedly with sultanates, courts, colonial classifications, and new state projects.

These two worlds have overlapped before. Chinese merchant settlements, kongsi formations, and southern commercial communities made bargains with local sultans, rajas, and Siamese lords rather than simply replacing them. Authority was layered. Commerce was real. So was court ritual. The point is not that the past returns intact. It never does. The point is that maritime Southeast Asia has older grammars of coexistence, brokerage, and partial sovereignty than modern strategic writing usually admits.

Which means the future may look less like “China versus Southeast Asia” and more like new bargains between Nanyang and Nusantara under modern conditions.

Not clan halls, but family offices.
Not tribute ports, but industrial parks and education corridors.
Not royal letters, but permits, developers, halal finance, property trusts, and negotiated jurisdictions.
Not old merchant republics, but new Chinese-adjacent commercial worlds nesting beside Malay courts, Indonesian scale, and Singaporean legal order.

This is why Johor matters.

Johor is not just “spillover” from Singapore. That word is too vague. What is emerging there is a much more precise form. Singapore holds the certainty: contracts, courts, hospital credibility, school prestige, financial seriousness, administrative trust. Johor holds the space: land, warehouses, condos, factories, service belts, and the possibility of an ordinary middle life at a cost Singapore cannot easily offer. Put the two together and you get a two-part organism. One side stores certainty. The other side stores expansion and everyday life.

That is not merely an economic arrangement. It is a regional form.

And it explains Singapore’s peculiar position in Southeast Asia. Singapore is not the opposite of the region’s fractured order. It is the hard membrane that makes that fractured order usable. It profits from being the place where regional ambiguity is translated into global legibility. Chinese enough to receive. Southeast Asian enough to understand. State-like enough to reassure. Small enough not to terrify. Enough state capacity to hub the region. Not enough hinterland to swallow it.

It is the hard edge of a soft world.

This is also why ASEAN, for all the jokes, will probably survive in some form. Not because it is marching toward Europe. Not because it governs a coherent bloc moving steadily toward union. But because a region built on negotiated incompletion benefits from etiquette. ASEAN’s real achievement was never convergence. It was choreography. It allowed a geography of courts, corridors, islands, generals, traders, developers, migrants, and ideologues to behave like a region long enough to trade, talk, and not explode too often.

That still matters.

It just is not the engine of history here.

The engine is underneath: a region that became very good at holding what larger systems cannot comfortably house inside themselves.

Chinese overflow.
American pressure.
Mineral hunger.
Migrant labour.
Religious revival.
Border capitalism.
Criminal innovation.
Malay recomposition.
Archipelagic ambition.

That is why Southeast Asia keeps getting more important without quite becoming harder.

So what will stun in 2036?

Not that Southeast Asia finally became East Asia. That would be too neat.
Not that ASEAN vanished. That would be too easy.
Not even that China mattered more. Of course it will.

The startling thing may be that Southeast Asia became more central precisely by remaining unfinished.

That its elites were not merely asleep or incompetent.
That weak sovereignty was not only incapacity.
That it was also, in many places, a method.

A way of keeping the court alive.
A way of keeping the door open.
A way of letting outside systems pay for inside compromise.
A way of monetizing fracture rather than curing it.

East Asia seized the heights.
Southeast Asia learned to live — and profit — in the fractures.

The danger is obvious. A region organized around profitable incompletion can become a playground for stronger powers, stranger money, predatory niches, and endless brokerage. It can host everything and author nothing. It can become rich in fragments and thin in direction.

But there is another possibility too.

Fractured lands and fractured seas are also where new social forms appear first.

Not in finished empires.
Not in perfectly consolidated nation-states.
In archipelagos, border belts, ports, hybrid courts, and half-made cities.

Southeast Asia is not the region that failed to become somewhere else.

It is the region where the future keeps arriving in pieces.

h/t Pietro Masina’s articles that provided the seed for this essay