My first essay showed a capability that has arrived ahead of the public's picture of it. My second argued that the danger it represents is not, at root, a problem of bad people: that catastrophic outcomes are produced by structures that reward the wrong things, and that swapping out the individuals inside those structures changes nothing while the structures stand. This essay is about the best mechanism I can find for changing the structure itself.
I'll say once what I'll only say once: the specific ideas below are not a condition of joining the work I think is necessary. You can reject every prescription here and still belong to it. The work requires only that we accept the shape of the problem: that it is structural, and that structural problems take structural solutions. Everything after that is open.
The hard part is this. Great structural change almost always arrives as the bill for great structural cost. The pattern is consistent across history: the structure rewards short-term gain over long-term survival, the long-term cost accumulates unpaid, and only when the cost finally comes due, in collapse or war or catastrophe, does the window for real change open. We reform after the disaster, not before it. We have done it that way nearly every time.
We may not be able to afford to do it that way this time. When the people who study these systems most closely are telling us the threat may be existential, the after-the-disaster model stops being available, because the disaster is the kind you do not get to learn from. We are being asked, then, to make the rare kind of change, the kind that comes before the cost instead of after it, against every habit history has trained into us.
It can be done. Our founders did not invent the Constitution on the spot. Behind that document lay centuries of accumulated thought: English common law, the long argument of the Enlightenment, the hard example of republics that had failed, and the immediate lesson of the Articles of Confederation proving inadequate. The intellectual foundations were laid long before the moment they were needed; the framers were drawing on work that had been done in advance. That is the posture I am arguing for: do the thinking now, while we still have room to think, about what we would want to build, so that if the moment comes, we are not improvising the design in the rubble.
It's worth being concrete about what we are trying to prevent, because vagueness invites people to file the whole thing under alarmism and move on. There are two threats, and they are not the same.
The first is the one the movies trained us to picture: an intelligence we build and then lose control of, a process started that cannot be stopped, the sorcerer's apprentice at civilizational scale. It is the more catastrophic threat and the less likely one. If it arrives, it takes everything, and we do not get a second attempt.
The second is quieter, more likely, and already underway. It is the use of these systems as instruments of manipulation: machines that learn to move us faster than we can notice we are being moved, deployed by companies whose incentives reward exactly that. This is not science fiction; it is the present business model of much of the internet, about to become far more capable. It is also the clearest illustration of the argument from my last essay: the companies building these tools are not staffed by villains. They are responding to a structure that rewards capturing attention and shaping behavior for short-term profit, with no incentive to weigh the long-term cost to the people being shaped. The manipulation is not a moral failing of the engineers. It is the structure, doing what it rewards.
One of these threats steals our lives. The other steals every chance we have to think a thought that is actually our own. Both are existential, in different registers.
We do not get past either by reaching for old-school politics. We must reach for politics anyway, because structural change has to begin where the rules are made. What we cannot do is treat this as one more battle for one party to win over the other. The conviction that the answer is your side finally defeating the other side is not the path to the solution. It is the thing guaranteeing there won't be one.
Consider two politicians most people would place at opposite ends of everything: Alexandria Ocasio-Cortez and Josh Hawley. On almost any partisan question they are adversaries, and their supporters experience them that way. Ask a narrower question, though: should members of Congress be allowed to trade individual stocks while writing the laws that move those stocks' value? They end up in the same place. Both have backed a ban. They arrive from different directions, one naming it the corruption of democracy by capital, the other naming it a Washington class enriching itself at the public's expense, and they land on the same structural reform. They are not alone. On a whole set of questions that are really about whether incumbent power gets to insulate itself, the populist edges of both parties quietly agree, and the comfortable, well-funded middle of each party quietly does not.
That tells you where the real axis runs. It is not left against right. It runs between those who benefit from the current structure and those who want to change it, and that line crosses both parties at a right angle to the one we are trained to see. The people most ready for structural change sit at both ends. The people most invested in the structure as it stands sit in the well-funded middle of each party, where the donors and the committee chairs and the safe seats are.
Why doesn't the agreement ever become a coalition? Because the structure defends itself with the one weapon that reliably works: it makes each side believe the other is the existential threat. As long as the progressive is certain the populist conservative is the danger, and the populist conservative is certain the progressive is the danger, the two of them spend every available ounce of energy fighting each other, and the structure that constrains them both is never touched. Partisan hatred is not a side effect of the system. It is the system's immune response: the mechanism that turns the people who might reform it into each other's enemies, so dependably that most of them never notice the thing they actually had a quarrel with was the structure, not the person across the aisle.
The first move, then, is not a policy. It is a refusal. It is the decision to stop treating the other side as the existential threat, not because the disagreements aren't real, but because the real threat is somewhere else, and we cannot see it while we are staring at each other. AOC does not have to stop being AOC. Hawley does not have to stop being Hawley. They keep every disagreement they hold. What they give up is the belief that the other's existence is the emergency. That is the entire ask, and it is harder than any platform, because the conviction that the other side is the emergency is the most carefully maintained belief in American public life.
Before I say what I'm asking of most of us, I want to speak to the few who hold real power in the current structure: the executives, the major founders, the people the present rules are built to protect. You will read a movement like this as a threat to your interests. I think you should look again, because I don't think the thing you're defending is the thing you actually value.
I'm not going to pretend you're comfortable. You can't slow down, because you know someone else won't. You can't step off, because stepping off feels like unilateral disarmament. You have looked for the exit and not found one that doesn't cost you your position. That isn't villainy; it's a prisoner's dilemma with the species as the stakes, and you didn't build it. The structure trains one fear into everyone it favors, that any change to the rules is a change that takes something from you.
Sit inside that fear instead of around it. Imagine the version you're most afraid of: you wake up tomorrow and the leverage is gone. Not your competence, not what you know how to build, not the regard of the people who have worked with you. Only the leverage, the part of your advantage that bends outcomes other people can't bend and absorbs mistakes that would end someone else. Picture the world where that specific thing is smaller, and then ask the honest question. In that world, are you worthless, or are you just less able to force the result?
If you built something real, you already know the answer. Your value does not live in the leverage. It lives in what you can make, and what you can make does not evaporate when the rules get more level. A more level structure reduces what you can extract beyond your contribution. It does not touch the contribution. The engineer is still the engineer. The founder who saw the thing three years early can still see the next one. What shrinks is the premium the structure paid you on top of that, the part that was never earned on the merits, the part you would lose the least sleep over once you're honest about which part it was.
I'll mark the limit of this plainly. It does not hold for everyone. Some people's entire sense of themselves is the leverage, the moving of others, and for them a level world is a genuine loss. Nothing here will reach them, and I'm not writing to them. I'm writing to the larger number of you who know the difference between what you have built and what you have been allowed to extract, and who suspect the second has grown out of proportion to the first.
There is a second thing the structure keeps you from feeling, because it's diffuse. You have seen what primitive AI did to the attention economy, recommendation systems optimizing for engagement with no model of the person on the other side, and the polarization and the wreck of adolescent mental health that came with them. Now picture systems that model one specific person's vulnerabilities in real time, generate the persuasion to fit, and test a thousand variants to find the one that moves them, not because anyone is evil but because the incentive is still engagement, still ad spend, still influence, and the precision just rose by orders of magnitude. That isn't Skynet. It's a billion small rooms, each tuned to one person. You are not standing outside it, and neither are your children. Your leverage does not exempt you from being moved. It only means whoever aims it at you can afford the best version.
There is the larger risk too, the one where the insulation that has always protected people like you simply fails, because the failure is the kind that doesn't sort by net worth. Pompeii didn't check anyone's accounts before the ash came down. I won't lean on that, because you can rate its probability low and I can't prove you wrong. I'll only say it belongs on the scale, and that it's the one item on the scale your protections were never designed to cover.
The offer, then, is not "stop." Stopping was never yours to choose; that's the trap. The offer is to change the game instead of the move: same competitors, a more level track, the leverage premium pared back for everyone at once so that restraint stops being the thing that loses. You don't have to design that from inside the structure, and the rest of this essay is about a channel that can build it without depending on you. I am not asking you to be the safeguard. I learned in the last essay not to build the floor out of anyone's better nature, yours included. I am asking you to make the floor cheaper to install, by not spending your weight to stop it. What I'm asking is narrower and harder than trust. Stop defending the premium, once you've admitted to yourself it was never the part of you that mattered.
To the rest of us, which is to say almost all of us, here is what I am actually asking, and it is smaller and harder than it sounds.
Keep voting for your party, and for the people who best represent your values. I am not asking you to change your politics. I am asking you to do one thing at your primary: among the candidates who share your views, find the one willing to work with the other side on the structural reforms, and reward that willingness with your vote. The primary is where this is decided, and it is also where your single vote carries the most weight, because almost no one shows up. If you want to know who those candidates are, don't listen to who talks about reaching across the aisle. Look at who actually did it when it cost them something with their own base: who voted to end congressional stock trading, who moved on revolving-door restrictions, who backed the transparency measures that expose self-dealing. The talk is cheap and universal. The votes are rare and legible.
Then, after you have helped put those people in office, do the harder thing. Remember that they cannot accomplish this with speeches. It will take working with people their base has been taught to treat as the enemy, and every time they do, someone will call them a traitor. You are the only one who can make that survivable. The reason our representatives cannot cooperate is that we punish them for it. We can also be the reason they can. Give them the room.
This mechanism is slow, fragile, and messy. It will involve temporary coalitions full of tension, and the appeal to the powerful is more hope than guarantee. Still, it is the best wedge available, because it routes around the incumbents at the one point they cannot fully control: who shows up to a low-turnout primary, and what those voters decide to reward.
We do not need to agree on the ideal tax rate or the precise shape of AI regulation. We need to agree that the current incentive structure makes honest compromise on those questions nearly impossible, and then change the rules of the game first, so that a new middle, less invested in protecting the status quo, becomes possible. That is the test I'd ask you to bring to any proposal, from any direction: does it change what the structure rewards, or does it just punish the people the structure produced? If it only does the second, it is theater, however satisfying. If it does the first, it deserves your attention regardless of which party is offering it.
We have almost always made this kind of change too late, after the cost was already paid. This is one of the rare times we are being asked to make it before the cost comes due instead of after. I don't know whether we will. The future will not be decided by who among us is virtuous. It will be decided by what we decide to reward.