Owen McGrann’s essay “The Dead Economy Theory” landed on May 1 and has been ricocheting around since, surfacing again this week on Hacker News with over a thousand points. If you haven’t read it: McGrann argues that AI automating white-collar labor doesn’t just eliminate jobs — it severs the pipeline that produces expertise itself. Junior associates become senior partners by doing the grunt work. Kill the grunt work, and twenty years later there are no senior partners. His sharpest example is TSMC’s $40 billion fab in Arizona, where American engineers reportedly need a year of training in Taiwan because the institutional knowledge never crossed the ocean with the machinery.

The essay has been received, predictably, as either a Luddite’s prophecy or a case for universal basic income. Both reactions miss what’s actually interesting about the argument.

McGrann has stumbled onto something real, but he’s framed it as a tragedy. It’s not. It’s a transfer — and the institutions positioned to benefit have names.

The Firm Is a Knowledge Repository, Not a Production Function

What McGrann describes — tacit knowledge that lives in people, not documents — is something organizational theorists have understood for decades. The reason McKinsey hires Princeton grads to make PowerPoints isn’t that the slides are valuable. It’s that making the slides is how you learn to think like a consultant, and ten years later you’re the one in the room telling a CEO what to do.

But here’s what the “dead economy” framing obscures: some institutions are designed to preserve this pipeline, and others never built one in the first place. The distinction matters enormously.

Consider the difference between a large law firm and a tech startup. The law firm’s entire economic model is the apprenticeship pipeline — it bills out junior hours at a markup specifically to fund the training that produces partners. The startup doesn’t have a pipeline. It hires senior engineers from other companies and expects them to be productive in week one. When McGrann warns that AI will kill the associate role and leave no future partners, he’s describing a genuine threat to firms. But he’s also describing a world where the firms that do maintain training pipelines — deliberately, expensively, against the grain — capture an enormous premium.

The knowledge doesn’t vanish. It concentrates.

The Arizona Problem Is a Choice, Not a Law of Nature

TSMC’s Arizona fab needing a year of training in Taiwan isn’t evidence that knowledge dies without people. It’s evidence that TSMC chose not to build the pipeline stateside because it was cheaper not to. The company made a rational short-term decision: keep the tacit knowledge centralized, train Americans later if you have to. The “dead economy” reading treats this as inevitable decay. It’s actually a specific corporate strategy that worked fine until it didn’t.

This is the hole in the theory. McGrann writes as if the erosion of expertise is a force of nature — entropy. It’s not. It’s a series of decisions by specific managers at specific firms who judged, correctly in the moment, that extracting knowledge from existing experts was more profitable than cultivating new ones. The bill for those decisions is coming due, but bills go to specific addresses.

The firms that paid to maintain their pipelines will be fine. The firms that harvested expertise without replenishing it won’t. That’s not a dead economy. That’s creative destruction, operating on a delay.

Who Wins When Nobody Knows How Anything Works

There’s a version of the future where McGrann is half-right: AI does displace the junior roles that produce expertise, and a lot of institutions discover too late that they’ve eaten their seed corn. In that world, who has the leverage?

Not the AI companies — they’re selling tools, not judgment. Not the generic “capital” either. The leverage belongs to whoever still has the people who actually understand the systems. That group is small, aging, and concentrated in a handful of institutions that resisted the urge to optimize away their training programs.

One trader on a fixed-income desk, watching the debate unfold over Slack this week, put it plainly: “Everyone’s worried about the models. I’m worried about the fact that when the models break, there are about twelve people in the world who can fix the plumbing underneath, and four of them are retiring next year.”

This is the version of the story McGrann didn’t write. The economy isn’t dying. It’s becoming a seller’s market for the people who actually know things — and the institutions that kept them around, often at the expense of quarterly margins, are about to look very smart.

The Real Warning

The essay’s value isn’t its conclusion. It’s the question it forces: does your organization have a knowledge pipeline, or is it just harvesting the one built by the generation before you? If you can’t answer that, the “dead economy” isn’t a theory. It’s a balance sheet item, and it’s coming due on your watch.

Sources