I quit out of the interview about ninety seconds in. It was an AI. Warm voice, my name, a little prepared curiosity about my background, and a small progress bar at the bottom of the screen telling me how much of myself I had left to perform. I closed the tab. I have done this every time now, without much drama, the way you hang up on a robocall. Somewhere along the way the thing on the other end of the hiring process stopped being a person, and I stopped being willing to audition for it.

I should be honest about where I'm writing this from, because it colors everything. I am mostly checked out of the world of work. Not the doing of things, I write, I build products and sites and tools, I fill the days, but I genuinely don't know yet whether that counts as work or what is supposed to come from it. Does it make me a penniless entrepreneur waiting on a breakthrough that may not come? A once-senior technical recruiter with a cute portfolio of side projects? Or just an odd duck who thinks too much instead of holding down an actual job? I can't rule out any of the three, and I've stopped needing to. What I have checked out of is the game. The politics for a paycheck. The interview theater. The path where you do good work and stay in the same place and are eventually handed a senior title with more responsibility and less meaning, which was always somehow presented as the reward. I wanted the other trade. The organization that actually teaches you what you don't yet know, that points itself at a real problem, that grows you on purpose. I waited a long time for that trade to be on the table. I've come to think it mostly wasn't, and maybe never was.

So I watch the labor conversation now from a distance, through a screen, from a series of places that are not my home. And distance shows you things that proximity hides. The main thing it shows you is the whiplash.

Two years ago the loudest voices told us AI would replace the workers and end the hiring. Then came the pushback, the corrections, the studies showing no measurable displacement yet. And now the same voices, many of them the very same people, have pivoted to telling us the replacement talk was always overblown, that in fact there is a labor shortage, that AI is going to create more jobs and better-paying ones, and that anyone anxious about it was being gaslit by pessimists. The narrative reversed itself completely inside twenty-four months, and almost nobody who reversed has acknowledged reversing. From the inside that probably feels like staying current. From the outside it looks like a single continuous act of not knowing, performed with total confidence in both directions.

Here is what I can see from out here that the confident voices seem unable to. Among the people I actually know, the AI transition has produced two outcomes, and neither is the one we were promised. Some of my friends got laid off. The others didn't, and they are working more now than they ever have. That is the whole distribution. Nobody got the shorter week. Nobody got the leisure dividend, the time back, the augmented ease. The productivity gain that everyone is shouting about from the rooftops went somewhere, and it did not go to the people producing it. If you actually owned the productivity you gained from these tools, you would be able to work less. The fact that the opposite keeps happening tells you who owns it.

This is the quiet transfer underneath all the noise about tools and jobs. The capacity to interpret the world, to hold a model of it and act, is consolidating into a small number of hands. The rest of us are becoming the thing that gets interpreted. You know less every year about the companies you deal with. They know more every year about you. The customer service line that greets you by name and will not tell you why. The hiring process that screens you with a machine you are not allowed to question. You have become perfectly legible to systems you cannot read back, and the friction that used to keep that asymmetry partial, the slowness, the paperwork, the human on the other end, is exactly what these tools were built to remove.

And the ones consolidating this will not stop on their own, because nothing in their incentives asks them to. Centralization is cheaper and it compounds. This is not a conspiracy. It is water running downhill. If you are waiting for the people gaining this kind of power to check it out of conscience, you have not been paying attention to any century that came before this one.

Which is the part I actually want to write about, because diagnosis is cheap and I am tired of reading it. I don't want to describe the trap more beautifully. I want to know where the door is.

Power does not yield to the argument that it should. It yields when it needs something it cannot make for itself. And there is one thing this whole apparatus cannot make for itself: grounded human judgment. Taste. The correction that comes from a person who actually knows the terrain and can see that the confident output is wrong. The models are extraordinary and they are brittle in a specific, expensive way. They do not know what they do not know, and they deliver the falsehood in the same clean voice as the truth. Every serious deployment eventually discovers, usually after it costs them, that it needs human beings to ground it, to catch the failure before it becomes the lawsuit.

That need is the door. It is the thing I have been building toward, quietly, in parallel to all of this, and the reason I have not simply gone silent.

The idea is not complicated to state, though it is hard to build. If the individual can be paid for their judgment, their discernment, their hard-won sense of when the machine is wrong, then the individual stops being a consumer of this technology or a casualty of it and becomes a necessary correction to it. And a necessary thing has leverage that a subsidized thing never will. Watch the incentives line up once you see it. The individual earns, from the one asset that does not lose value as the models improve, because the better they get the more their remaining errors cost. The company lowers its risk, because a distributed layer of human judgment is cheaper than the failures it prevents. And the institutions that want compliance and accountability get a population that is legible and paid, rather than legible and idle. Three motives, one direction. The check gets built not because anyone chose virtue but because everyone at the table is getting something real.

I want to set this against the other future being offered, because the contrast is the whole point. The mainstream answer to the displacement people can feel coming is management. A basic income. A stipend for the surplus, paid to be idle and quiet and, above all, legible. It is offered as compassion, and in a narrow way it is. But look at what it makes you. In that future you are a cost to be administered. You receive, and dependence is the opposite of leverage, and the hand that pays you to do nothing gets to decide what nothing includes. The other future asks more of you and hands back more. In it you are not paid to disappear. You are paid for the thing the machines still cannot do, and that same payment is what keeps them honest. You are load-bearing. This is not idleness dressed as work. It is the old idea I keep returning to in these letters, that judgment and discernment are not the absence of value but its source. Your taste becomes the product. The quiet work of knowing something well becomes the thing that cannot be centralized, and therefore the thing that checks the center.

I won't pretend the road there is clean. The likeliest way it arrives is the way every counter-infrastructure arrives, unevenly, first to the people who have no other choice, then as a premium for the ones who can afford to own their own context while everyone else keeps renting theirs. There is a version of this where sovereignty becomes one more advantage the well-resourced hold and the rest do not. I have not solved that, and I am not going to tell you I have. But I would rather build toward the future that makes the individual necessary, watching honestly for the way it could curdle, than accept the one that makes me a well-funded line item.

So this is less a manifesto than an invitation, and mostly to the people quietly in the same place I am. The ones who have stopped auditioning. The ones working more for less meaning and wondering what the title is even for. You do not have to accept that the only two chairs are laid off or buried. There is a third position, and it starts with a small, almost unglamorous act: deciding that your judgment is a thing you own, and refusing to give it away for free to a system that has stopped putting a person on the other end. Figure out where in this you are legible, and to whom, and on whose terms. Then start making that legibility the thing they have to come to you for.

I closed the tab on that interview and I did not feel like I'd lost anything. That is either the beginning of an answer or the most comfortable way to be wrong. I'm betting on the first, and I'm building like I mean it.