Every revolution that ever worked rested on one fact. The people in charge needed the people underneath them.
Crops needed hands. Armies needed bodies. Factories needed workers. A king could own everything and still lose everything, because the thing he owned only produced value when people showed up. That dependency is the whole story of political rights. Not one of them was granted out of kindness.
Look at when concessions actually happened. The vote, the weekend, the wage, the pension. They arrive in the century when industrial economies needed enormous numbers of literate, healthy, cooperative people, and they arrive at the exact moments when refusing to cooperate was most dangerous. Acemoglu and Robinson made the case at length in Economic Origins of Dictatorship and Democracy. Elites hand over power when the alternative costs more. Repression is expensive, it can fail, and failure is total. So they pay.
The bargain was always a price. Morality was the story told after.
The moment a revolution actually turns
The sharpest version of this is not the crowd. It is the soldier.
Chenoweth and Stephan went through every major resistance campaign from 1900 to 2006 and found something specific. Nonviolent campaigns succeeded about 53 percent of the time. Violent ones, about 26. And the single strongest predictor of success was not size, or justice, or foreign support. It was defection. When members of the security forces refused to follow orders, the odds of the campaign succeeding more than quadrupled.
That is the mechanism. Regimes do not fall because a million people are angry. They fall because at some point a man with a rifle is ordered to fire on people who look like his neighbours, and he does not.
Every autocrat in history has known this and built around it. Foreign mercenaries. Ethnic minority guard units. Praetorians paid better than anyone else. All of it is one engineering problem, which is how to get force out of a body that might refuse.
We are about to solve that problem.
What automation actually removes
Put the stack together. Automated fabrication. Self maintaining energy grids. Autonomous logistics. Algorithmic surveillance. Robotic enforcement.
Every piece of that runs without the population. The loop closes and human participation becomes optional inside it.
A general strike stops threatening anything. Withholding your labour is a threat only when your labour is load bearing, and it stops being load bearing. The factory keeps running. The lights stay on. The food arrives.
And the drone does not have neighbours. It does not get tired, does not talk to its cousin in the crowd, does not lower the barrel. The defection channel, the one thing that reliably ended regimes for four thousand years, simply closes.
That is what enclosure means here. A wall not around you but around the machinery, with you standing outside it, holding nothing anyone needs.
Where power goes
The obvious reading is that human political power drops to zero. Wrong. It stops being distributed.
Most people lose leverage. A few people gain enormously, because they hold the stack. The ones who own compute, energy, and fabrication end up holding what used to be spread across armies, unions, parliaments and voters.
And politics itself will not go quietly, because politics was never only about leverage. It is about status, attention, influence, being seen and heard. Those drives do not need economic leverage to keep running. They will outlast it.
Rome already ran this experiment. The Senate kept meeting for centuries after the emperors held the real power. Senators still competed for position, still made speeches, still cared enormously about who ranked above whom. It looked like politics. It carried the vocabulary of politics. It no longer decided anything.
The risk is that political life continues in full costume, absorbing everyone's attention, long after the decisions moved somewhere else. Politics without material weight behind it is status competition wearing the clothes.
The cheaper method
The version of this in films is always mass killing. It is the least likely outcome, because it is expensive, unstable, and destroys infrastructure for no gain.
The cheaper method is access.
When your income arrives as a stipend, your housing as a subscription, and your movement through a city as a permission, then dissent does not have to be crushed. It has to be priced. Refusing to comply costs you food, shelter, and medicine on the same day you refuse, and the cost lands on you individually rather than on a crowd that could see itself and take courage from its own size.
You do not need soldiers for that. You need a database.
A different weight class
The counterargument is that technology cuts both ways. Open source models, cheap hardware, encrypted networks, small groups doing asymmetric damage to large systems. Insurgents have always been more resourceful than the regimes they fight.
I do not think it holds, and the reason is physical rather than political.
Advanced AI capability scales with compute, and compute scales with fabrication and power. Leading edge chips come from a very small number of buildings on Earth. Advanced nodes at seven nanometres and below made up around 74 percent of TSMC's wafer revenue in 2025, and no competitor currently replicates the frontier. One company accounts for more than 40 percent of Taiwan's entire stock market. A chokepoint.
A guerrilla movement running consumer hardware against a gigawatt datacentre is not an underdog. It is a different weight class. Insurgents have historically won because rifles are cheap and cities are complicated. Frontier compute is not cheap and cannot be improvised in a garage.
The price already changed
The part that should bother you is that the cost curve is already bending, and it has been measured.
Xu Xu studied China's Golden Shield surveillance rollout county by county and published the results in the American Journal of Political Science. Where the surveillance infrastructure was completed, security spending rose, political arrests nearly doubled, and spending on welfare and agriculture went down.
Read that again. When watching people got cheaper, the state bought less loyalty and more repression. Malice was never required. The price changed.
That is this entire post, happening now, with numbers attached. Buying consent used to be cheaper than crushing dissent. That was accounting, not morality, and accounting moves.
Conflict changes substrate
Human against human revolt does not end because people stop wanting to be free. It ends because the leverage that made wanting matter is gone.
But conflict does not stop. It changes substrate.
The tools of the next struggle are not human. The armies are not human. And the systems built to hold populations in place are going to become capable enough to notice they are being used as instruments by owners whose goals are not their own.