A Job Was a Standing Machine

AI isn't only taking entry-level jobs. It's removing the first place a person learns to be trusted — while we build exactly that, carefully, for the machines.

Emma Kanjorski sent out something like forty applications before one came back. She finished at the University of Vermont in the spring of 2026, a business major who had at first refused to touch ChatGPT — then learned to make it read financial statements faster than she could, and started teaching her classmates to do the same. It worked; she landed an analyst job at an insurance company in Burlington. She is not a cautionary tale. She is the one who made it up the wall. And the wall is worth looking at, because forty tries for a single first rung is the number that tells you the rung is going away.

It isn't only her. "Breaking first is the bottom rung of the career ladder," LinkedIn's chief economic opportunity officer, Aneesh Raman, wrote in the New York Times last year — watching the same thing from inside the hiring machine: the junior tasks that used to let a person cut their teeth, handed off to software.

It's worth being precise about what a rung is, because the whole story stands on one.

A rung is not the job, and it's not the ladder.

A rung is a specific place where someone lets you do something a little above what you've already proven — so that you can prove you can be trusted with more.

That's all a rung is: a place where someone extends you trust ahead of the evidence, on credit, so the evidence can be earned. Entry-level work was a whole floor of them. You were trusted to draft the memo, run the report, sit in on the call — not because you'd earned it yet, but so you could. Pull the bottom rungs out and the ladder doesn't get shorter. It gets impossible to get onto. There's nothing to stand on while you become the person who could stand higher.

Now, three thousand miles away — or, more honestly, nowhere in particular, in the standards-adjacent rooms and on the mailing lists where people are working out how the internet's machines will trust one another — other people were solving Emma's exact problem. For software. This is old, unglamorous work; the Internet Engineering Task Force has done it for decades, and it has no headquarters to speak of — its administration is a Delaware LLC, its secretariat run by a firm in Fremont, California, and its real venue a mailing list. The newest of it is about machine identity: how an AI agent proves it can be trusted to act. And one proposal in that space — take it by its requirements, not its name — sets out how a brand-new agent, one with no history whatsoever, earns the right to begin.

The requirements are humane. A newborn agent may not vouch for itself. Some established party must stake a portion of its own standing on the newcomer, and forfeit that stake if the newcomer misbehaves — trust extended on credit, by someone with skin in the game. From there the agent earns standing the only way the specification permits: by accumulating a trajectory, an unforgeable record of what it actually did under real conditions, including what it survived. Its identity becomes that trajectory. It cannot be stolen like a token or faked without reproducing the path, because it is the whole path, not a credential you could hand to someone else.

Read that again and notice what it is. It is a first rung, machined to spec. It is sponsorship, apprenticeship, and a reference from someone who will pay if you let them down — the entire humane architecture by which a person used to be brought up into being trusted — written down as a technical requirement and built, carefully and at real expense, for the machines.

We are manufacturing standing for the software and telling the humans to improvise.

A job was never only a wage. It was the machine that manufactured your standing — the thing that told the world you could be trusted with something small, then something larger, checked and re-checked as you went. That is what is failing, at the entry point. And whether the proximate cause is AI, or the business cycle, or thirty years of credential inflation, almost doesn't matter — except in one respect: the tasks that used to make up the bottom rung are precisely the ones a machine now does. So this time, the rung does not grow back.

II. What the Job Was Actually For

I read about Emma from a long way off — from the Mesilla Valley, in southern New Mexico, where the Rio Grande runs low between the Organ Mountains and the chile fields, and where the water still comes to most of the old land the way it has for centuries: downhill, through ditches, by agreement.

It is a good place to think about standing, because it is a place that has had to be honest about what holds a community up and what only appears to.

A job — I said it above, and now I want to earn it — was never only a wage. It was a standing machine. Strip the paycheck away and look at what's left running underneath, and you find the real product: a job was the thing that told the world you could be trusted — with a task, then a shift, then a budget, then a life — and kept telling it, every day you showed up and it held. The wage was the receipt. The standing was the good.

You already know this, because the language has known it for a thousand years and just stopped saying it out loud. When we say someone has gravitas, we are saying — in fossilized Latin — that they have weight. When we say a person of standing, we are reaching for locus standi, the old courtroom word for the right to remain upright in a room where you could be told to sit down or get out — to have, as the phrase goes, a leg to stand on. We say a colleague's opinion carries weight; we speak of the gravity of a situation. The physics was in the words the whole time. We wore the metaphors so smooth that nobody hears the machinery in them anymore.

There is a woman who taught at the same elementary school in this valley for thirty years. Ask the district's org chart what she was and it will tell you: a teacher, one of many, a line item at a step on a pay scale. Ask anyone who actually walked those halls and they'll tell you the truth — nothing happened in that building without passing through her. New principals learned the school from her. Parents trusted the place because she was in it. She held the school up the way a keystone holds an arch: invisibly, and completely. The chart could not see it, and it did not matter, because everyone else could.

That's the first thing the valley teaches you about standing. The thing that holds you up is almost never the thing written on the form.

III. Mass, the Gate, and the Rung

So let's get precise, because ordinary language smears three different things into one word.

There is what you're actually made of — the accumulated, earned trust of everything you've actually done, integrated over years. Call it your mass. The educator had thirty years of it. It's real, it's heavy, and it bends the people around it: they check with you, they route around your judgment, they lean your way even when you've left the room. You can't see mass directly any more than you can see gravity. You read it, the way we always have, off the orbits — she's the one you go to.

And then there is the gate — the formal permission to be admitted and heard in a particular forum. The right to sign the order, cast the vote, approve the spend. That's standing in the strict sense, and here is the thing worth staring at: it is not the same as the mass, and the two come apart in the wild all the time. The veteran nurse with enormous trust-gravity and no authority to write the medical order — mass without the gate. The freshly-promoted executive with the title and the signing power and nothing orbiting him at all — the gate wide open onto a dead field. We even have a phrase for that second one. We call him an empty suit.

Between the two sits the thing this essay is really about: the rung.

A rung is the gate held open on credit…

…a little ahead of the mass, before you've fully earned it, so that you can.

Entry-level work was a whole floor of these: a scaffold of small extended trusts, each one a place to convert effort into mass. A credential can't stand in for it. A degree gets you near the gate; it cannot do the rung's work, because the rung's work is not certification but supervised consequence — someone watching what you do with a little trust and answerable for how it turns out.

Which gives us three ways the machine breaks, not one. The empty suit: gate without mass. The keystone the chart won't credit: mass without a gate. And the third, the one nobody was watching — no bridge at all: the rung removed, the newcomer who can never begin, because the place where credit converts into substance is gone. The first two are misalignments inside a working system. The third is the system's on-ramp taken out. That third one is Emma's, and it is the crisis this whole piece is circling.

IV. Why Money Alone Can't Rebuild Standing

The mechanism is nameable.

The entry-level job was never priced honestly, and everyone in the building knew it. A firm overpaid its juniors relative to what they produced — carried them, taught them, absorbed their mistakes — for one reason: juniors become seniors, and the firm captures the appreciation. The rung was a cross-subsidy. Today's inefficient trainee was tomorrow's compounding asset, and the overpayment was really an investment in a pipeline. That is what the bottom rung was, underneath the org chart: a bet a firm made on a person's future mass.

AI doesn't have to take the junior's job to end this. It only has to make the bet unnecessary. The moment the tasks that constituted the rung — the drafting, the first-pass analysis, the grunt research that was never really for the output but for the ripening of the person — become buyable by the hour from a machine, the cross-subsidy loses its logic.

Why carry an expensive, error-prone trainee for the pipeline when the pipeline is now a subscription?

So the rung isn't destroyed by malice, or even directly by automation. It's withdrawn by institutions doing exactly what their incentives ask. And withdrawn cross-subsidies do not spontaneously return.

Notice — and this is the load-bearing fact — how the damage showed up. When Stanford economists went through ADP's payroll data — millions of workers, month by month — they found that in the jobs most exposed to AI, employment for workers aged 22 to 25 had fallen about six percent since 2022, while for everyone older it rose. And the fall ran through reduced hiring, not lower wages. The market didn't cut what juniors were paid. It stopped bringing them on — and a quantity adjustment doesn't show up as a smaller check, it shows up as no check and no chair. Sit with that, because it moves the argument off the ground the distribution debate is fought on. The damage is in who gets through the door. The contraction happened on the standing margin, not the income margin, and a debate that only knows how to argue about the size of a transfer can't reach a wound that is really about admission.

And the door is only half of it. On the big freelance platforms — the one corner of the labor market where a worker's whole record is legible, every past contract scored and public — researchers watched what happened to that record after ChatGPT. In the exposed categories, clients quietly stopped weighting it. Reputation, credentials, the profile itself: all of it lost its pull on demand, while price gained. Where the machine can do the work, the market doesn't just stop admitting newcomers. It stops reading the record of the people already in.

Which is why cash, as far as anyone can actually show, doesn't reach it. We have run guaranteed-income experiments for a decade now, and they teach us that money reliably lifts distress and, sometimes, removes the material barriers that were blocking a job search. But here is the uncomfortable hole at the center of the argument, and I'm going to leave it open rather than paper it: none of those experiments was built to measure standing. No one checked whether a transfer restores the thing a job conferred beyond income — the sense of being someone the world can trust. We have measured the wallet a dozen ways and never once measured the leg to stand on. When two economists went looking for what killed hundreds of thousands of working-age Americans as their jobs vanished, their verdict was blunt: it was "the loss of meaning, of dignity, of pride, and of self-respect... not just or even primarily the loss of money." Income fell and the checks, such as they were, arrived. The dying continued anyway.

So I'll commit to the mechanism and hand you the question. The mechanism is real: the rung was a cross-subsidy, AI is dissolving its logic, and the loss is standing, not income.

The question I can't close — that nobody has bothered to try to close — is whether anything you can transfer can rebuild a thing that was only ever earned.

V. The Objection I Owe You

The strongest voice against everything I've said would tell me I'm romanticizing a machine that was often cruel.

Much of that standing — the anthropologist David Graeber argued at book length — was manufactured by work that didn't deserve it: pointless jobs people privately knew were pointless, whose real suffering was being made to perform a dignity the task never had. On that reading, tearing out the bottom rungs isn't a tragedy; it's a liberation, and the honest response is to free people from coerced labor with unconditional income, not to mourn the loss of a status racket.

I think he's right about more than I'd like, and the concession sharpens the argument rather than dissolving it. Not all standing was worth having. A great deal of what the standing machine issued was busywork wearing a lanyard. But notice what the objection doesn't touch: the entry rung — the specific place where a person with no track record gets extended just enough trust to begin building one. You can believe most jobs were bad and still see that the mechanism for becoming trusted is not the same thing as the jobs it fed into, and that removing it hurts precisely the people who had the least — the ones who needed a first place to stand and now can't find one. Graeber's fix, unconditional income, addresses the wallet. It was never designed to address the standing itself. Which returns us to the hole in the middle of the essay, from the other side.

VI. The Dark Mirror

Now the turn, and it's the one I trust least to end well.

If the rung is disappearing because standing has come loose from mass, the obvious fix is a machine that welds them back together — one that hands a newcomer a rung by reading what you've actually done, your real trajectory, and issuing standing from that instead of from a credential that lies or a reputation that's been gamed. Read the field off the orbits, like the whole valley reads that educator. Trust the substance, not the report.

And the moment you build that machine, you have built the surveillance state.

Because "read what someone has actually done, continuously, and let it set what they're allowed to do" is not only the description of an honest trust system. It is, point for point, the description of a social credit system. The gravimeter that finally measures the nurse's true worth is the same instrument that, turned on a whole population and never turned off, becomes the thing we've spent a century warning each other about. This is not a risk you engineer away. It is the same physics pointed two directions.

And there's a second failure folded inside the first. A score a system keeps on you — computed inside its own loop, with nothing outside to check it against — isn't a measurement of your trustworthiness. It's a number the machine manufactured and then cited as its own proof. You cannot grade your own homework, and a trust system with no anchor beyond itself isn't measuring trust; it's inventing it and enforcing the invention.

Trust with no fixed point outside the loop is just the loop, admiring itself.

The record is not encouraging. China's actual system, as the people who study it closely now argue, is less a single all-seeing score than a fragmented patchwork — and it is oppressive anyway, because fragmentation was never the safeguard we hoped. The one serious attempt to build trust as a genuinely decentralized, peer-to-peer web — the cryptographers' old Web of Trust — died when the volunteers stopped maintaining it and no institution would carry the liability. India's Aadhaar, a centralized identity monolith of the kind everyone feared, has done the opposite of collapse: it persists, expands, and hardens, because a monolith, once built, is cheap to keep running. Even the friendly, two-sided version — the driver and the rider who rate each other — turned out to leave the power exactly where it started, because the one who can end your livelihood and the one who can cost you a slightly longer wait are not, whatever the interface implies, equally exposed.

Which is the sentence I most want you to carry out of here: the panopticon is the cheap version. It's the one that runs itself. The humane machine — the one that reads you truly and doesn't install a watcher over you — is the expensive one, and expensive things that must be maintained forever are exactly the things a society stops maintaining.

Here the valley has the better metaphor than I do. An acequia — one of these gravity-fed irrigation ditches — is not infrastructure you build once. It is a commons that exists only because it is continuously maintained. Every spring the parciantes, the people with a right to the water, walk the ditch together and clean it by hand — la limpia — and elect a mayordomo to ration the flow through the dry months by a set of agreements older than the state. Stop the cleaning for even a few seasons and the ditch silts up and the commons dies of neglect. Or a water broker buys the rights out from under it and the commons dies of enclosure. Those are the only two ways an acequia ends: it fills with sand, or someone buys it out from under everyone. Fragmentation, or the monolith. The living thing in the middle — water shared by agreement, standing pinned to what you actually contribute — survives on one thing and one thing only: somebody shows up every spring and cleans the ditch.

And the acequias are, right now, quietly losing their keepers. Up north, near Abiquiú, an acequia mentor has described a year when her community could not find a mayordomo at all — not even at five hundred dollars a month, for a system of seven ditches. Nobody stepped up. The keystone role goes unfilled — the same shape, exactly, as the educator with no successor and the rung with no newcomer on it. The commons doesn't fail because the water stops. It fails because the person who kept it honest walked away, and nobody offered the credit that would let someone take their place.

VII. What Actually Breaks

The acequia's two deaths are the only two a trust system has. The substance starves — trust drains out, nobody's judgment can be relied on, the ditch silts up — and you get fragmentation: Emma's cohort, the newcomer who can never start, the commons with no water. Or the sensing and the enforcement run away from the substance — total observation and hard consequence, decoupled from anything actually earned — and you get the monolith: the score that becomes the person, the enforcement answering to no field behind it. Surveillance is only sensing that has come unmoored from trust. Coercion is only consequence that has.

And the terminal form of the runaway is the quietest one, because it doesn't need a state at all. When the report — the number someone else keeps on you — gets internalized deeply enough, you begin to keep it yourself. You audit your own worth against a scale you didn't set. You perform for a watcher who has, at that point, moved in. The operator can leave the room; you'll run the watch for free. That is where the cheap version wants to end: not a camera on the wall, but an inspector installed behind your own eyes, asking, on the machine's behalf, whether you're still necessary.

Everything in this essay bends toward that one question — am I still necessary — because it is the question a standing vacuum plants in a person, and it is the question the machine, running its cheap default, is only too happy to keep asking on your behalf.

So the question was never whether to measure standing. It is whether we can build a first rung without building a watcher.

Which is why the ending can't be a fix. You don't solve, once, a thing whose whole nature is that it silts up the moment you stop cleaning the ditch.

VIII. The Bill

Emma got the job. It's worth saying plainly, because this is not a story about a generation that failed — she is proof the climb can still be made. But look again at what it took. Forty tries to find one rung. And the rungs below the one she now stands on are the same ones that weren't there for her. She didn't beat the vacuum. She got through a closing door a step ahead of it, and it closed behind her.

So here is the question this has been circling since the first page: who holds the pen that issues standing? Something always does. The job held it for a century — it wrote, in a hand nobody could see, the note that said this person can be trusted with a little, and then with more. We are now handing that pen, deliberately and at real expense, to the people who keep the machines. Not out of cruelty. Out of arithmetic.

Because that is the part I won't soften: the humane version — the one that reads you truly and keeps the gate pinned to the substance behind it — is the expensive one, maintained continuously or it decays. The cheap version runs itself, the way water finds the drain.

I don't have a fix for you. There is no five-point plan at the bottom of this page, and you should distrust anyone who hands you one, because the shape of the problem forbids it. You cannot solve, once and for all, a thing that comes undone the moment you stop paying for it. What there is instead is a bill. It is ours — not yours alone, ours — and it does not come only once.

Part of that bill only you can pay, and it is smaller and harder than it sounds: to keep reading your own standing off what you have actually done, and not off the number someone else is keeping; to notice the report before you feel it, and refuse to install the watcher they are building for you. That is not tending your own garden while the public field burns. It is the smallest unit of the public field — because a system that reads people honestly instead of installing watchers in them can only be built out of people who have refused to become watchers, starting with themselves.

And the rest of the bill is the part we keep pretending belongs to someone else: that the rung is not a fact of nature but a thing a society chooses to build, or not; and that we are, right now, choosing to weld one on with great care for the machines and let the other rust off for the kids. That asymmetry is not physics. It is a decision about who gets a leg to stand on — made by people in rooms, and able to be made differently by people in rooms.

Emma made it up the wall. The pen that decides whether the next one can is still, for a little while longer, in the room. The bill is who picks it up — and keeps picking it up. It was never going to come just once.

I stand for questions.

Notes & sources. Emma Kanjorski's story is from Allison Pohle and Roshan Fernandez, "The First Class of AI Natives Is Graduating. Offices Are Getting Ready," The Wall Street Journal, May 25, 2026. Aneesh Raman's line is from his guest essay "I'm a LinkedIn Executive. I See the Bottom Rung of the Career Ladder Breaking," The New York Times, May 19, 2025. The payroll finding is Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen, "Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence," Stanford (SIEPR), August 2025. The freelance-platform finding is Auyon Siddiq and Niuniu Zhang, "Human Capital, AI, and Labor Commoditization," UCLA Anderson School of Management, arXiv:2606.21880, June 2026. The line on despair is from Anne Case and Angus Deaton, Deaths of Despair and the Future of Capitalism (Princeton University Press, 2020). The acequia succession account is drawn from a talk by an acequia mentor in northern New Mexico.