Other Men With Machines

Picture a farmer at the back of a church in 1400. He can’t read. Nobody ever offered to teach him, and nobody is going to.

The book on the table up front is in Latin. The priest reads from it and tells him what God expects, what he owes, which of his neighbors are in trouble, and what happens to him after he dies if he doesn’t pay attention. The farmer takes all of it on faith. Not faith in God, or not only that. Faith in the man doing the reading.

Maybe the priest is honest. Plenty of them were. His honesty isn’t the problem.

The problem is that the farmer can’t check. He can’t open the book himself. He can’t ask a second reader, because the only other readers in his life answer to the first one. If the priest is wrong, or lying, or just repeating something he was told by someone who was wrong, nothing the farmer can do will ever reveal it. He’ll die not knowing. So will his kids.

Hold onto that man. This isn’t going to be an essay about religion.

What it is, plainly: an honest attempt to describe how power has actually been structured among people for as long as we have any record of it. Not a philosophical treatise, not the start of a movement, and not an attack on anyone who happens to hold power at the moment — no names, no party, no company, nothing that turns on who won the last election. The pattern I’m describing is older than every institution currently standing in it.

The reason to do this now is simple. What’s moving into the middle of that structure is more capable than anything that has ever held the spot, and the consequences of getting it wrong scale with the capability. That’s worth looking at straight on, before it settles, rather than working it out afterward.

The war everyone misremembers

Most people who’ve heard of Dune know it involves a desert planet and enormous worms. A fair number know the books contain a religious prohibition, invented for the novel and attributed there to a scripture called the Orange Catholic Bible. It forbids building any machine that imitates a human mind.

So the story gets filed as a robot war. Machines got too clever, humans fought back, humans won, and afterward they outlawed the machines. Frank Herbert wrote it in 1965, decades before anyone needed to worry about it, and it reads like a prophecy about artificial intelligence.

Then look at what the victory produced. Herbert’s universe, after the war against the thinking machines, has a hereditary emperor. It has feudal houses trading planets and assassins. It has a monopoly on the one substance that makes interstellar travel possible, and whoever holds that monopoly holds everyone. It has humans bred and trained from childhood to serve as living computers, because the machines are gone and somebody still has to do the math. And it has a religious order quietly steering human bloodlines across ninety generations toward an outcome of its own choosing.

They destroyed the machines. Then they rebuilt every single chain out of people.

Herbert puts the thesis in the mouth of a character named Reverend Mother Gaius Helen Mohiam, who explains the old war this way: “Once men turned their thinking over to machines in the hope that this would set them free. But that only permitted other men with machines to enslave them.”

I had that sentence half right for years.

I read it as a warning about giving up your own thinking. Outsource your judgment and you lose the capacity for it, so guard the habit of thinking for yourself. That reading is correct and it matters, and I’m not trying to take it away from anyone who holds it.

But the second sentence is the one I do not often find analyzed, and I’ve come to think it’s the more important of the two. It never says the machines enslaved anyone. The machines are barely in it. It says other men with machines did the enslaving. The tool sits in the middle. The problem is human at both ends of it, and that end of the sentence is the one we skip.

Everything below is me following the part I skipped.

Whoever stands in the middle

Something out there decides the shape of your life. Maybe it’s the will of the gods, or the law, or the land you need to eat, or the capital that decides whether you have work. You can’t reach it directly. Somebody stands between you and it, and that somebody tells you what it says.

Whoever holds that middle position holds you. Not as a figure of speech. As a job description.

• The shaman and the spirits
• The priest and the scripture
• The king and divine right
• The noble and the bloodline
• The lord and the land
• The owner and the capital
• The state and the law

Each one gates something a person needs and can’t get to alone. Each one gets to describe that thing to the people who can’t see it. And each one, in its turn, replaced the one above it on that list.

Notice what the position doesn’t require. It doesn’t require force. Armies are expensive, and armies have opinions. The middle position needs one condition only: you can read and the other person can’t. That’s cheap to keep up, it doesn’t need barracks, and it survives being unpopular.

It also doesn’t require dishonesty, which is the part I find hardest to hold onto. The farmer’s priest could be scrupulous. He could work himself sick getting the translation right, and the arrangement would be exactly as strong. The power isn’t in the lying. It’s in the asymmetry. So you can’t fix it by finding better people to put in the chair, and the history of trying is the next section.

The position has one more property, and it’s the one that makes it stick. Whoever stands in the middle also gets to define what counts as a legitimate objection to standing there. The priest decides what’s heresy. The crown decides what’s sedition. The owner decides what’s unprofessional, the state decides what’s unpatriotic, and the market decides what’s uneconomic. Every one of those words is a way of saying your complaint doesn’t parse. The appeal runs through the thing you’re appealing.

The chain survives being broken

People have torn these arrangements down. Repeatedly, at enormous cost, and with real courage.

The Reformation broke the church’s monopoly on reading scripture. Within a century, princes ran national churches and Europe was at war over which prince’s version counted. The book was in your language now. Someone else still told you what it meant.

The English executed their king in 1649 and got Cromwell, who ruled as Lord Protector, and then got the monarchy back eleven years later.

The French did it more thoroughly, and in 1804 Napoleon crowned himself emperor inside a cathedral, with his own hands, in front of the Pope. Fifteen years after the Bastille. They killed a king and made an emperor.

The Russians overthrew an aristocracy in 1917 and produced a party elite with private shops, private clinics, and country houses, sorted by rank.

Aristocracy itself is the clean case. Across most of Europe the titled nobility lost its formal grip, and industrial and financial fortunes took the position instead. Plenty of duchies still have dukes. They just stopped being where the power sits, which is the whole point — the sorting outlived the paperwork it used to travel on.

Colonial empires came apart. In a good number of the new states the same extraction carried on under a new flag, run by a local elite with a better claim to the land and the same relationship to the people standing on it.

The guilds were broken, and factory owners took what the guild masters had.

Feudal dependence on land ended, and wage dependence replaced it. You stopped owing your lord a share of the harvest and started needing a job to eat.

And slavery. Formal abolition swept the 19th century, and it was followed, in the United States, by convict leasing and sharecropping and debt peonage — the same extraction rebuilt out of contracts and criminal law. Most readers know that version, and it’s one instance of a global behavior. The Atlantic trade, the plantation colonies, forced labor across European empires and outside them: this was how a great deal of the world ran, for centuries. Abolition in one place routinely coincided with business as usual somewhere else. It also never fully stopped. The International Labour Organization, Walk Free, and the International Organization for Migration put the current figure at fifty million people in forced labor or forced marriage. That’s up more than nine million from their previous estimate five years earlier.

The Who said it best. New boss, same as the old boss.

There’s a lazy version of this argument and it’s worth nothing, so let me refuse it before somebody hands it to me. The lazy version says nothing ever changes. That’s false, and it’s insulting to people who died making things change. Slavery being illegal is better than slavery being legal. A vote is better than no vote. Some of these fights produced real gains that real people lived inside of, and I have no interest in waving that away from a comfortable chair.

The claim is narrower and worse. Every one of these overturnings removed the people holding the position, and not one of them removed the position. The chain didn’t break. It changed hands.

And the reason has nothing to do with the character of revolutionaries. Some were opportunists. Most weren’t. It doesn’t matter. The position rewards a particular kind of conduct no matter who’s sitting in it, and anyone unwilling to behave that way loses the position to somebody who will. That’s not a flaw in the people. It’s the shape of the seat.

So the target was never the occupant. Revolution aims at the occupant. That’s what revolution is for, and it’s why revolution keeps producing a new one.

The thing underneath is the separation of people by power.

And when I say separation of power I don’t mean the civics-class version, the legislature and the executive and the courts holding each other in check. I mean the separation of people by power. Two tiers. One tier decides, the other gets decided for. Everything else in this essay sits downstream of that, and every attempt so far has treated it as furniture instead of as the problem.

There’s one more piece, and it’s what makes the rest work. The division comes first. Before anything gets taken from anybody, people get sorted — priest from laity, noble from commoner, citizen from foreigner, owner from worker, and free from enslaved. The sorting isn’t a consequence of the exploitation. The sorting is what makes the exploitation possible, and it’s the same machinery every time, which is exactly why it’s so easy to miss. It looks like the natural order right up until you notice somebody assembled it.

The message that became the throne

The clearest case I know is one most people file under something other than politics. Strip the theology out of the Jesus story and read what’s left as a report from an occupied province. A man travels around Roman Palestine saying the poor come first, the sick get cared for, and the people on top are going to find themselves on the bottom. There’s a kingdom coming, and it isn’t Caesar’s. Then the local authorities and the imperial governor execute him by a method Rome kept for sedition.

That’s not a devotional claim. It’s a claim about who should be in charge, made inside an empire with firm views on the subject.

Two scholars I’d point you to if you want this argued properly. John Dominic Crossan, in The Historical Jesus and elsewhere, reads the movement as a deliberate challenge to the Roman order and the local hierarchy cooperating with it. Richard Horsley, in Jesus and the Spiral of Violence, puts it in the specific politics of Roman-occupied Galilee and Judea. You don’t have to accept a word of the religion to follow either of them. Whether any of it is true is a different essay. I’m pointing at the shape.

Within roughly three centuries, that movement was the official religion of the Roman Empire. Constantine legalized it. Later emperors made it the state church. The thing that had opposed the empire became the empire’s paperwork, and the people running it now held both the scripture and the legions.

Which puts us back in the church in 1400, with the farmer who can’t read Latin.

A movement whose entire content was the reversal of power became, in time, one of the most durable instruments of power in human history. Not because anyone betrayed it in some identifiable moment. Because the position was still there, and the movement grew big enough to fill it.

There isn’t enough, so someone has to decide

Every one of these arrangements needed a reason. Nobody holds a position over other people and says out loud that they’re doing it because they can. There’s always an argument, and it’s the same argument in different clothes.

• Connection to the divine is scarce, so priests ration it.
• God’s favor is scarce, so kings monopolize it.
• Noble blood is scarce, so aristocrats guard it.
• Land is scarce, so lords hold it.
• Capital is scarce, so owners direct it.
• Public money is scarce, so bureaucrats means-test it.

Same sentence every time. There isn’t enough to go around, so somebody has to decide who gets it, and by a happy coincidence that somebody is us.

The causation runs backwards from how it’s presented. It isn’t that things are scarce, so sharing them equally is impossible. It’s that a small number of people hold the thing, and the shortage that produces becomes the argument for why it can’t be shared. The hoard creates the scarcity. The scarcity justifies the hoard. The excuse manufactures its own evidence, which is a good part of why it’s lasted so long.

It’s worse than circular. The loop pays.

A shortage raises what the thing costs. A higher price moves more resources to whoever’s holding it. More resources buy more of the thing, and more of the means to keep holding it — more land, more titles, more lawyers, more enforcement, more say over the rules that decide what counts as holding. Which lets them hold back more, which deepens the shortage, which raises the price again.

So it isn’t a circle that sustains itself. It’s a spiral that concentrates. Every turn ends with more in fewer hands than the turn before, and every turn also produces a fresh, sincere-sounding reason why things can’t be otherwise. The scarcity isn’t only the alibi. It’s the engine, and price is the belt that drives it.

That inversion took me longer to see than it should have. Having it the other way round is comfortable, because it makes the whole arrangement look like weather. Watch it run in three places.

Access to God was never in short supply. There was no underlying resource at all. The shortage got invented and then sold, and it worked for a thousand years, which should tell you how little the underlying facts matter to this argument.

Common land was shared, then enclosed, then described as scarce. The acreage didn’t change. What changed was who could stand on it.

And Bengal, in 1943. Estimates of the dead run from one and a half to three million, and some go higher.

The economist Amartya Sen, who later won a Nobel, went through the numbers in Poverty and Famines in 1981 and argued that the food available per person that year wasn’t much below the years around it. What killed people, on his account, was distribution: wartime inflation, grain moved to buyers the government had prioritized, restrictions on moving it between provinces, and hoarding by people who could see the price climbing. Famine, he argued, is usually not an absence of food. It’s an absence of any way to command food that exists.

I should tell you that this specific claim is disputed. Peter Bowbrick and Mark Tauger have both argued that Sen misread his sources on Bengal, and there’s real physical evidence for a genuine shortfall that year — a cyclone in Midnapore in 1942, and an outbreak of a rice crop disease across the province. Sen’s broader framework reshaped how famine gets studied. His arithmetic on this particular famine is still being fought over eighty years later.

Which doesn’t rescue anybody, and doesn’t damage the point. Take the critics at their strongest and you get a real shortage, in wartime, met with prioritized procurement and grain held back by people watching the price. A hard year became a mass death event. That’s not an argument that scarcity was fake. It’s an argument that a genuine shortage plus a hoard kills people who a genuine shortage alone would not have killed.

That’s the spiral with the numbers attached, and the numbers are corpses. The shortage lifted the price. The price rewarded holding grain back. Holding grain back deepened the shortage, which lifted the price again. Nobody in that chain had to intend a famine. Each step was locally reasonable for the person taking it, and the sum of the reasonable steps was a million people dead within walking distance of food.

Let me be careful here, because this is the point where the argument usually gets oversold and then dismissed. I’m not saying scarcity has ended. Energy is constrained. Housing is constrained. Land, water, time, and the attention of people who care for other people are all genuinely limited, and any essay that waves that away deserves to be closed. The claim is narrower: scarcity is being used, and often produced, by the people whose position depends on it. Real limits get pointed at to justify an arrangement the limits don’t require.

Land is finite. There’s a fixed amount and no more is coming. A trained model is a file. It copies for effectively nothing, as many times as you like, without diminishing the original. Whatever scarcity exists around it is a decision, from the first byte to the last. No harvest failed. No acreage ran out. When something can be handed to everyone at no cost and isn’t, the argument has to be made on different grounds, and those grounds are the whole subject.

Rules, not rulers

I’m an anarchist, philosophically. Not the cartoon version — no bombs, no chaos, no argument against organization or against rules. The word means something plainer than its reputation: rules, but not rulers. Rules are constraints everyone lives under. Rulers are the people exempt from them. And this isn’t communism; the argument isn’t about who owns what. It’s about who gets to decide.

There’s no good reason a small group should decide for everyone else. People have been producing reasons for as long as there have been small groups doing it, and the old ones are all dead. Divine appointment. Blood. Natural superiority, and the science invented to prop it up was embarrassing before it was discredited.

The live one is better, and it deserves a straight answer. The modern defense of a deciding class isn’t that they’re chosen by God or born to it. It’s consent: they were selected, they can be removed, and they’re bound by the same rules as everyone else. That’s a real argument, and the best one anybody has.

Look at what the consent actually consists of. You consent en masse, occasionally, to a bundle — and only if you’re lucky enough to live somewhere that offers you the chance. You don’t consent to any particular decision. You consent to a person who will then make thousands of decisions you’ll never see, on questions that were never on the ballot, for years at a stretch. Your share of that consent is one vote among millions. What they get in exchange is concentrated authority over you, individually, by name.

That asymmetry isn’t a flaw in the process. It’s the product. Diffuse input, concentrated output. The vote is where the dilution happens, not where it gets fixed.

And “removable” is doing far less work than it looks like. You can remove the person. You cannot remove the position, and the position is what the last thirty paragraphs were about. An election is a revolution on a schedule, and it has the same failure mode: the cohort changes, the seat doesn’t. As for being bound by the same rules — the rules that bind the tier are written by the tier, which is the whole trick performed in slow motion.

What’s left, once the live argument has had its say, is the reason nobody states out loud: they hold the position, and holding it is what lets them keep it.

So the burden of proof sits on hierarchy. Not on the person asking why two tiers exist. Absent a reason that survives contact with daylight, the default is that people decide together. That’s a moral claim, and I’d rather state it as one than dress it up as an inevitability.

Three objections deserve real answers.

The first: hierarchy produces coordination, and coordination is what produced the surplus you want to share out. Fair, and partly true — large-scale cooperation did build it. But coordination doesn’t require rulers. It requires rules, and the difference is whether anybody is exempt from them.

The obvious follow-up is the one that matters: fine, but who enforces the rules? Somebody breaks one, and if there’s no tier standing above everybody, what happens?

Enforcement happens. That’s not the part I’m arguing against. What I’m arguing against is an enforcer who can’t be the enforced. A jury is enforcement, and every person on it goes back to being subject to the same law the following week. Rotation, recall, terms that end, selection by lot — these aren’t decorations on democracy, they’re the mechanisms that keep the enforcer inside the rules rather than above them. The question isn’t whether someone has authority. It’s whether they can be on the receiving end of it by Thursday.

And this works in practice, which is the part I’d have doubted if I hadn’t gone looking. Elinor Ostrom spent a career documenting communities that manage a shared resource — irrigation systems, fisheries, forests, mountain pasture — for generations, without privatizing it and without a state running it. Governing the Commons came out in 1990 and she won a Nobel for it in 2009. Three of the eight patterns she found in the arrangements that lasted: the people using the resource make the rules, they monitor each other, and they sanction each other in graduated steps. Users writing and enforcing their own rules, with nobody exempt. Not a thought experiment. Villages in Japan and Switzerland doing it for centuries.

Here’s the honest limit. Ostrom’s cases are mostly small, and scale is the unsolved problem in all of this — I’m not going to pretend a Swiss pasture generalizes to two hundred million people, because it doesn’t obviously do that and I don’t know how it would. What her work does kill is the claim that the choice is a ruler or chaos. That claim was empirically false before anyone applied it to this.

So the argument isn’t against organization, structure, expertise, or planning. It’s against an exempt tier, and the exempt tier isn’t the thing doing the coordinating. It’s the thing collecting from it.

The second: people differ in ability, so somebody has to decide. Also fair, and it proves less than it looks like. Differences in ability justify differences in role. The surgeon should do the surgery, and I want the person landing the plane to be much better at it than I am. But a role is specific, earned, and revocable. The surgeon doesn’t get to decide your housing, and stops being a surgeon if the work goes bad. A ruling tier is none of those things. It’s general where a role is specific. It’s inherited or bought where a role is earned. And it answers to nobody. Confusing the two is how an argument for competence turns into an argument for aristocracy, which is a trick with a long history.

The third objection is the one I’d raise if someone handed me this essay, and it’s built out of my own argument. Earlier I said a movement with the most anti-power content imaginable became the throne, and that nobody had to betray it for that to happen. So why would this be any different? If having the right values didn’t save that, values are beside the point, and I’ve spent a few thousand words asking you to hold some.

The answer is in what the values are about, and it’s the same distinction that sank every revolution above. Look at what those movements actually asked for. The poor go first. The meek inherit. The workers own. Every one of those names a better occupant. And content like that is perfectly compatible with the seat surviving, because all it specifies is who ought to be sitting in it. So when the movement wins, it installs its people, and the seat goes on doing what seats do. The capture isn’t a betrayal of the content. It’s the content getting what it asked for.

What I’m asking for is about the seat. Not that better people should decide for you — that nobody should be exempt from the rules, whoever they are, whatever they believe, however much you like them. That content can’t be satisfied by winning it. There’s no version of occupying the position in the name of abolishing the position and calling it a victory. The day my side gets comfortable in the chair is the day I was wrong, and unlike every movement in this essay, I don’t get to be pleased about it.

Which gives you a test worth more than anything else here, and you can run it on any movement including this one. Does it name an occupant, or does it name the seat? Name the occupant and history is fairly clear about where you end up.

One clarification, so this doesn’t get read as something it isn’t. The thing to take apart is the control mechanism, not the balance sheet. My interest isn’t in who has how much. It’s in the arrangement where one group decides for another, because that arrangement is where the pile comes from in the first place. De-concentration of authority, not redistribution of stuff. Get the first one and the second stops being the fight.

The new thing in the middle

There’s a new thing standing in the middle. It gates knowledge and judgment, and increasingly the decisions that set how a life goes — whether you get the loan, the interview, the apartment, the benefit, the diagnosis, the sentence. Some of that is already routine. Automated screening decides which resumes a person ever reads, which applications clear a credit check, which tenants get a viewing. The rest is arriving.

I’m not going to name companies. You know who they are. Naming them would date this essay by a year and let you read it as a complaint about particular firms, when the point is a position that somebody will occupy no matter which firms survive.

The questions that matter are the boring structural ones. Who owns the compute. Who holds the weights and sets the terms on which anyone else touches them. Whose judgment gets replaced and whose gets amplified. And the one that ties this whole essay together: who can still check the answer.

Because that’s the farmer in 1400. He isn’t oppressed by the book. The book is a book. He’s in trouble for three reasons stacked on top of each other. What determines his life is written in a language nobody taught him. It’s held by someone whose interests aren’t identical to his. And there’s no second reader he can go to who isn’t part of the same arrangement.

Now sit somebody in front of a system that just produced a decision about them. They can’t inspect it. They can’t reproduce it. They can’t get a second opinion that isn’t running on the same substrate with the same blind spots. They can ask why and get a fluent, confident, unverifiable answer.

Same position. Better vocabulary.

When was the last time you could actually check?

And the new sorting is already being drawn — between people whose judgment still counts for something and people whose judgment has been priced out of counting. Remember that the division comes first. It always has.

Why it has to be now

The strongest objection to everything above is that technology diffuses.

The printing press really did break the church’s monopoly on reading. Not immediately and not cleanly, but it broke. The internet really did drop the cost of publishing to nearly zero, and a lot of gatekeeping died with it. Open models exist right now, run on hardware a person can afford, and get better every few months. This is the core reason why open models matter. The pattern in this essay is a pattern, not a law, and anybody selling it as destiny is selling something. If the tools spread widely enough and cheaply enough, the middle position gets crowded, and a crowded position isn’t much of a position.

I think that’s wrong this time, for three reasons that compound.

It doesn’t democratize access. It centralizes authority. A printing press was affordable to a prosperous merchant. Training a frontier model isn’t affordable to a mid-sized country. What gets distributed widely is the finished product, on terms set by whoever made it, and that’s a different thing from distributing the capacity to make one. Handing everyone a very good telescope doesn’t make everyone an astronomer, and it certainly doesn’t hand over the observatory.

It substitutes for judgment. Every previous diffusion helped ordinary people because they still had something to sell. Skill, labor, attention, discernment. That’s what gave a person something to bargain with, and it’s why the press mattered — a literate population is a population that can argue back. This is the first tool aimed squarely at that. If judgment is the thing being automated, then the mechanism people have always used to claw back ground is the mechanism being removed.

It’s an industrial revolution’s worth of capability arriving at once, in very few hands. The first industrial revolution took a century and reorganized every society it reached. It produced the shift described further up this essay, from depending on land to depending on wages. This one is compressed into a decade or two, and the organizations that can build one from scratch are few enough to name from memory.

This looks more like the atomic bomb than the printing press, and the way it fails to match is the part that should worry you.

The bomb is the technology that never diffused. It created a tier of who has it and who doesn’t, and eighty years later that tier has barely moved. It didn’t spread destruction around. It concentrated destruction and froze the arrangement in place, and every state in the top tier has spent the decades since making sure the tier holds.

Now the obvious objection, which is a good one. The bomb stayed put because of physics. Fissile material is genuinely hard to get, and that bottleneck did more for non-proliferation than any treaty. Chips and data aren’t uranium. So maybe the comparison argues the opposite of what I want it to.

Except look at what’s actually diffusing. The capacity to build one of these systems sits with a handful of organizations and is getting more concentrated, not less. What’s spreading is the finished product, on a subscription, to anyone with a card. Those are two different things, and only one of them is power.

So we’ve arrived somewhere genuinely new. Bomb-class impact, held at the production end by very few. And at the usage end, wider distribution than any tool in history, available by the month. Nobody has ever sold fissile material by subscription.

That combination is worse than either half. Concentration at the top means the position in the middle stays occupied. Distribution at the bottom means the surface where things can go wrong is now everybody, everywhere, all at once. The diffusion people point to as reassuring is a second risk stacked on the first, and it doesn’t touch the first at all.

AI is not a bomb, and none of this is a prediction of catastrophe. This is arithmetic, not alarm.

Risk is likelihood multiplied by impact. Nothing in my argument needs the bad outcome to be probable. It needs the impact to be big enough that even a modest likelihood produces a number worth acting on. And notice what the standard reply does — that’s speculative, you can’t prove it’ll happen — it attacks the likelihood term and leaves the impact term alone. That isn’t how any of us handle risk in any other part of life.

So name the impact instead of gesturing at it:

• Mass economic dependence with no bargaining position, because the thing people were selling was judgment.
• Monitoring and enforcement cheap enough that dissent becomes trackable by default instead of by exception.
• An information environment mediated by systems whose owners choose the frame, at a scale where there’s no outside to compare it against.
• Decisions about people’s lives made by systems those people can’t inspect, can’t reproduce, and can’t appeal.
• And the one that changes the category: lock-in.

Lock-in is the whole argument.

Every chain described above eventually broke. Slowly, at horrible cost, and it broke — because the tier holding it always needed a new hand to pass it to. Generations turn over. Successors are incompetent. Loyalty has to be bought, enforcement needs enforcers, and enforcers are people with their own ideas. The whole grim comedy of the historical record is that power keeps having to be handed to somebody, and the handoff is where the openings are.

A tier that holds judgment, monitoring, and enforcement at the same time doesn’t have a handoff problem.

Which is worth spelling out, because it’s the step everything else rests on. Coercion has always been labor-intensive. Ruling people takes soldiers, clerks, tax collectors, censors, informants, and someone to decide which claim gets denied and which door gets kicked in. Every one of those is a person with a limit. That’s why strikes work. That’s why mutinies happen, why a clerk who won’t file the paperwork is a real obstacle, why regimes fall apart when enough people at the bottom of the apparatus quietly stop cooperating. The exit from every arrangement in this essay ran through the fact that power over ordinary people required the participation of ordinary people.

Those jobs all come down to judgment. Deciding who to flag, what to approve, which exception to allow. Judgment is precisely the thing now being automated.

The successor inherits something no heir has ever inherited: an apparatus that already runs. No loyalty to re-purchase, no network to rebuild, no gap during the transition where the grip loosens. And the incompetent son stops mattering, which is worth pausing on, because incompetence at the top has been one of history’s more reliable liberation mechanisms. Dynasties fell because the third generation was a fool. A machine doesn’t need a competent hand on it.

You cannot organize a general strike against a system that doesn’t employ anyone at the layer that counts.

That’s the thing worth being frightened of. Not a machine deciding to hurt anyone. The pattern that has repeated through all of recorded history could stop repeating — not because the chain finally broke, but because it no longer needs to change hands. No succession crisis. No incompetent heir. No guards to bribe, because the guards are a service contract.

Unrecoverable is the property that makes something a nuclear-class risk. Not the body count. The absence of a second try.

And that standard isn’t exotic. Nobody demanded proof that nuclear war would happen before building arms control. A low probability of an unrecoverable outcome was enough to produce treaties, inspection regimes, and a telephone line between capitals. We already accept this reasoning. We accept it for asteroids, for pandemics, and for the structural integrity of bridges. The only unusual thing about applying it here is who it would inconvenience.

What to refuse

I’ve spent five thousand words on a diagnosis, so it’s fair to ask what follows from it. I don’t have a program, and I’d distrust anyone who claimed to. But the argument does point at one thing, and it’s the same thing in every section above: whether you can check.

That’s what the farmer didn’t have. It’s what the person in front of an automated decision doesn’t have. And it’s specific enough to act on.

Refuse “the system decided.” A decision about your money, your health, your housing, or your liberty should be inspectable, and appealable to a person who can be named and held responsible for it. Not a queue. Not a portal. A person. When the answer is that the model produced it and nobody can say why, that isn’t an explanation, it’s the Latin.

Back the arrangements where a second reader can exist. Models people can run themselves and check against each other. Not because openness is virtuous — because a second opinion that isn’t produced by the same system as the first one is the only thing that has ever broken this pattern, and it can’t exist if the capability sits in three buildings.

Watch for the exemption. Somebody writing rules they aren’t subject to, deciding on cases they’ll never be in, holding a position they can’t be removed from. That’s the tell, it’s always been the tell, and naming it out loud is cheap and unpopular and worth doing anyway.

None of that is sufficient and I’m not going to dress it up as sufficient. It’s what’s available to someone who isn’t in the room, which is most of us, which is the problem.

The farmer learns to read

Give the farmer what he was missing. Teach him to read, and put the book in his own language.

It doesn’t end the way it should. He can read it now, and a prince runs the church, and the prince has views on which translation is permitted. Two centuries on, his descendant can read a great deal and owns nothing. He has a wage instead of a field and a landlord instead of a lord. And he takes on faith a whole set of arrangements he has no way to examine.

He got literacy. He didn’t get out.

He broke the wrong thing. The Latin was never the problem. The priest wasn’t really the problem either. The problem was a world sorted into people who read and people who get read to, and literacy alone didn’t touch that sort. It just changed what the two groups were called.

Which is Herbert’s point, and I’ll hand it over now that it’s yours: the machine was never the chain.

Dune has been in print for sixty years and Hollywood keeps going back to it. A warning that popular doesn’t get refuted. It gets adapted.

The farmer never got to ask the only question that would have helped him. You still can, and it’s short enough to keep in your pocket: who’s doing the reading, and what would it take for me to check?