There are, at any given moment in the American economy, several million unfilled jobs and several million people who cannot find work. This is described, in the press and in the industry that surrounds hiring, as a mismatch. A skills gap. A pipeline problem. The framing assumes that the labor market is a market that has developed a fault, and that with better tooling or better training or better matching, the fault could be corrected.

I spent nearly a decade working inside that assumption. I was a recruiter, most recently at large technology companies whose names you would know. My job, stated plainly, was to reduce the gap between what firms said they needed and what the available population of workers could provide. I got progressively better at it. I was promoted for getting better at it. And the longer I did it, the more clearly I could see that the thing I was getting better at was not the thing that was actually wrong.

What's wrong is upstream of anything a recruiter can touch, and upstream of anything the firm employing the recruiter can touch, and largely upstream of anything the entire industry that has grown up around hiring can touch. The labor market is not a market that has developed a fault. It's an allocation system with no allocator, no shared measurement, and no institution positioned to notice when it stops working. What we're calling a mismatch is what that system produces when it runs.

What follows is why.

The first thing to understand is that firms used to make their own skilled labor, and they stopped.

For most of the twentieth century, entry-level work was the mechanism by which a person went from being untrained to being useful. You were hired into a role that didn't require you to already know the job, and the job taught you. Some of the teaching was formal, but most of it was structural: you sat next to someone who knew what they were doing, you were given progressively harder tasks, and after some number of years you knew what they knew. This was expensive. It was slow. It could not be justified on a quarterly basis. Over roughly four decades, firms cut it.

Peter Cappelli, who studies this at Wharton and is probably the most cited person working on it, has documented the collapse. In the late 1970s, young workers received something like two and a half weeks of formal training per year on average. That number has fallen sharply and never recovered. Firms now pay a substantial premium to hire people who already have the skills, which shows up as a cost, but a cost in a different column than training used to occupy. His summary of what happened is that employers gradually persuaded themselves, over decades, that training was someone else's responsibility. The government's. The universities'. The worker's.

Here is the part that gets less attention. Nobody picked it up.

The universities kept doing what they had always done, which was sort people and certify that the sorting had occurred. The government did some workforce development at the margins, at a scale that has never come close to what firms used to do internally. And workers were told to invest in themselves, which most of them cannot afford to do at the scale required, and which even when they can afford it does not teach the firm-specific and tacit knowledge that on-the-job training used to teach.

So supply of skilled labor became nobody's responsibility. Not in the sense that the responsibility was contested. In the sense that no institution has the job. If you asked, in America today, "whose job is it to produce a person capable of doing skilled work," there is not a straight answer, because the answer used to be firms and firms have exited the role.

A market whose supply side has no producer is not going to clear. This is not a moral observation. It's a mechanical one.

The second thing to understand is that the roles firms are trying to fill are not neutral descriptions of work that needs doing.

Over the last several decades there has been an enormous proliferation of job titles. There are more of them, they are more specific, and they have less and less to do with what a person actually does when they arrive at their desk. This proliferation is often described as a natural consequence of technical complexity, and some of it is. Most of it isn't.

A job title is a variable the firm controls. It is not a description that firms are forced to adopt by the nature of the work. It is a choice, and firms make the choice for reasons that include, but are not limited to, the actual content of the role.

Consider what a title actually does inside a company. It sets a compensation band, which lets the firm pay narrower ranges more precisely and defend those ranges to the person occupying the seat. It sets an internal status level, which participates in the endless small politics of who reports to whom and who gets to speak in which meeting. It defines a career ladder, which shapes retention. And it operates as a piece of external signaling that is deliberately illegible: a Staff Engineer at one company maps to an EngII at another, and this is not an accident. Firms have every reason to keep the mapping opaque, because opacity makes their people harder to poach and easier to underpay.

I have seen firms give someone a Principal title in what is functionally an entry-level role, because the title bought loyalty the salary could not. I have seen the same body of work labeled six different ways at six different companies. The title is not a description. It's a lever, and firms pull it for pricing, for status, for retention, for legibility control, and often for reasons that are just organizational habit.

The specialization is real, some of it. But most of the proliferation is a firm-internal accounting mechanism that the rest of the world reads as if it were an objective description of what work exists.

Now put those two facts together. Roles are specified at a granularity that is at least partly manufactured, and the population that used to be trained into those roles is no longer being produced. What you get is a labor market where demand has been engineered into unfillability and supply has been abandoned. And the resulting vacancies get counted, and reported, and cited as evidence that workers are the problem.

The millions of open jobs are not a contradiction of the millions of unemployed people. They are what the system was going to produce.

If this is roughly correct, and I think it is, it raises an obvious question: why is this not the dominant framing inside the industry that studies hiring?

The answer is that the industry that studies hiring is not primarily composed of people who study hiring. It is composed of people who sell things to people who hire. The benchmarks that shape practice come from analyst firms funded by vendors. The reports on the future of work come from platforms that sell the future of work. The maturity models come from consultants who grade companies against frameworks the consultants also sell the remediation for. The research that would matter, the labor economics literature that has actually been asking these questions for thirty years, sits in journals that nobody in the function has been trained to read, behind paywalls that nobody in the function has a budget to cross.

Cappelli himself has written about this. His observation, and it is more damning than most of what he writes, is that decades of research on hiring and workforce planning sit largely unread by the practitioners who would benefit most, while a well-funded vendor economy shapes what those practitioners actually think about their own work.

Structural explanations don't circulate because no vendor can sell against them. There is no product for "the way we organize labor is broken." There is a product for every possible band-aid. So the band-aids circulate. So the conversation stays at the level of tactics. Practitioners are not incurious. The information environment they operate inside has been shaped, quite thoroughly, by parties whose interests do not include arriving at the structural answer.

All of this describes a system that was already broken. What is arriving next is going to be worse, and it is going to be worse in a specific way.

A great deal of what specialization was, historically, was a response to coordination cost. It was cheaper to hire three people who each did one thing than to develop one person who could do all three, given the tools available. That calculation is changing. A single person, right now, with current technology, can build software, market it, run the numbers, handle the operations, and produce the media that supports it, at a level that would have required a small team five years ago. This is not a prediction. This is happening.

The consequence, if you extrapolate, is title consolidation. Not for ideological reasons and not because anyone is planning it. Because the economic reason for the specialist structure is being removed. The fifteen roles collapse into one because the coordination overhead that justified fifteen roles is no longer there.

A distinction worth drawing. There is a version of this argument that becomes utopian. The consolidation is not going to be aimed. It is not going to reorganize labor around what society actually needs, because there is no mechanism in the system that represents what society actually needs. To aim it, you would need something that does not exist: comprehensive labor market data that isn't owned by vendors, an institution with the authority to act on it, and a policy apparatus that could reach fifteen years upstream into the education system on signals that change faster than that. You can describe such a system. You cannot build one in the current political moment, and the version of it that could be built would be something most readers would not want.

So consolidation arrives without a driver. The specialist structure collapses, the training pipeline that was already gone stays gone, and what remains is a smaller number of very broad roles filled by generalists, with no institution positioned to notice who got left out or to do anything about it if it did notice.

I spent nearly a decade being handed a problem and getting better at solving the problem I was handed. What I could not shake, and what eventually made the work impossible to keep doing, was that the problem I was handed was not the problem that existed. Every fix I got better at was a band-aid. The wound was that we have built a labor market with no producer of workers, no honest description of roles, no institution charged with the health of the system, and no data anyone can trust that isn't owned by someone selling something. And now we are about to run a large and rapid reorganization of that system through it, driven by a technology that removes the last economic reason the existing structure made sense, with no hand on the wheel and no wheel to put a hand on.

That is what I mean when I say the labor market is not a market. Markets have prices, and prices carry information, and the information is used by allocators who face consequences for getting it wrong. What we have is a set of firms optimizing locally, a set of vendors selling into that optimization, a set of workers who are told the outcomes reflect their own inadequacy, and a public conversation dominated by parties whose business it is that the conversation not reach its conclusion.

If you are reading this and you have been laid off, or you are waiting for the layoff you can feel coming, or you have been applying for months into a void that returns nothing, I want to say this plainly: it is largely not your fault.

You are inside a systemic breakdown. The pipeline that used to produce people like you was dismantled before you entered it. The roles you are applying for were specified by people optimizing for their firm's internal accounting, not for anything you could reasonably prepare for. The advice you are being given comes from a discourse that has been captured by parties who profit from your continued search. None of that means you are powerless. It means the thing you are struggling against is much bigger than you were told, and the fact that you have not solved it alone is not evidence of anything about you.

I do not want to end this with the usual advice, because the usual advice is part of what I have been describing. Build a personal brand. Diversify your income streams. Upskill relentlessly. That is the vendor economy speaking through your career coach, and most of it assumes conditions the people who most need help do not have. So instead, three things that follow from what I have argued above.

Stop treating your title as a description of what you do. It is a firm-internal accounting variable that happens to be written in English. The skill you build is not the title you hold. Track your actual capabilities separately from your resume, in whatever form makes sense to you, because when you leave, the title stays with the company and only what you can actually do comes with you.

The training your employer is not providing is your responsibility to construct, and it does not look like a course. A course cannot tell you that you are wrong. It can only tell you that you completed it. What you need is something with an honest feedback loop, which means something real: a project that either works or does not, a piece of writing that either lands or does not, a small business that either has customers or does not. The certificate is the band-aid version of what you actually need. The capability is the thing itself.

Assume no institution is coming. Not your employer, not your government, not your industry, not your professional association, not anyone. This sounds bleak and it is clarifying. Every decision about your working life over the next decade should be made with the assumption that the old contract is not going to reassert itself and that there is no safety net designed for what is arriving. The people who navigate this well will be the ones who priced it in early.

And if you want a place to actually start, this week, here it is. Pick one capability you want to genuinely have, not one credential you want to be able to claim, and begin the thing that will tell you when you are wrong about it. Not a certification. Not a curriculum. A project, a repo, a client, a piece of writing, a small operation, a domain you go deep in on your own time where reality gets to grade your work. It does not have to be ambitious. It has to be real. By the end of the month, you should know something you did not know at the start, and you should know it because the world told you, not because a platform did.

Nobody is driving. That is not a slogan. It is the situation. And it is also, if you look at it the right way, the beginning of a different kind of freedom, because a system with no hand on the wheel is also a system with no permission required.