The Dead Economy Theory

Owen McGrann extends "dead internet theory" to the economy itself: a system where productive capacity hums along without needing human participation. The essay is the sharpest political-economy critique of AI acceleration I've read — it names the financial incentive structure honestly, traces the self-destructive logic of the AI layoff cascade, and frames the entire moral question through the Camus-Sartre break, arguing the present person is the unit of account, not some future trillion-soul utopia.

Key Quotes

"The gentler language ('copilot,' 'assistant,' 'augmentation') is marketing."

McGrann is blunt: the valuations require elimination, not augmentation. OpenAI at $830B and Anthropic at $380B didn't get there on "making humans more productive." The TAM has to be the global labor market itself.

"Remove labor from that equation and watch what happens."

Democratic governance is a bargain. The governed provide labor, tax revenue, military service, and consumer spending. The governors provide order and services. AI breaks that bargain asymmetrically — a handful of corporations own the productive capacity, and nobody needs the governed anymore.

"People don't want a check. They want work. They want purpose."

Citing David Shor's polling showing a federal jobs guarantee polls better than UBI, and Case & Deaton's "deaths of despair" research. The mechanism isn't poverty — it's the loss of purpose, status, and perceived future. This is the essay's deepest insight: the AI labor crisis is existential before it's economic.

"The person who exists now… is the unit of account."

The Camus-Sartre break in one line. Sartre and the Marxists said present people can be acceptable casualties in pursuit of a better future. Camus said no. McGrann is Camusian. This is the philosophical core the accelerationists have no answer for.

"Warmed-over Parfit without the rigor."

On Longtermism. And on Effective Altruism: "utilitarianism reinvented by people who have apparently never encountered Bernard Williams." The whole section on Silicon Valley's intellectual poverty is devastating — Nietzsche misreadings, Bayesian rediscoveries made without Kuhn/Lakatos/Feyerabend, a tradition that keeps reinventing wheels badly.

"All of them require functioning democratic institutions with the will to challenge the richest companies in human history."

The solutions are known (public stakes, antitrust, taxing automated labor, capital distribution). The problem isn't intellectual — it's political. And "the window is closing."

Key Themes

• #concept Dead Economy Theory — the successor to dead internet theory. Productive capacity without human participation. "Plenty happens and none of it requires you."
• #concept The AI Layoff Trap — the three-turn cascade: cost-cutting surge → customer base evaporates as laid-off workers stop spending → the company discovers its customers were other companies' workers. A prisoners' dilemma where "every firm rationally automates beyond the socially optimal level" (Falk & Tsoukalas, Wharton).
• #concept Camusian AI Ethics — the moral case that present people are the unit of account. Rejects utilitarian sacrifice of the present for a hypothetical future. The philosophical framework accelerationism has no answer for.
• #pattern Deskilling as Active Harm — Anthropic's own research shows junior engineers using AI agents complete tasks no faster and understand their work less. The harm isn't just economic (lost jobs) but epistemic (lost understanding).
• #concept Democratic Leverage Erosion — the governed lose bargaining power when their labor, tax revenue, and consumption are no longer needed. The public funded the research; private companies captured the reward.
• #person Owen McGrann — writer at The Palimpsest (Substack). Sharpest political-economy critic of AI acceleration in the current discourse.

Critical Analysis

This is the essay the AI safety crowd should be writing but isn't. The alignment/safety discourse has been captured by longtermist abstractions that treat present harm as a rounding error. McGrann does the obvious thing — starts with the financial incentives, traces their logic, and lands on a moral framework that treats actually-existing people as what matters. It shouldn't be radical. It is.

The three-turn cascade is the mechanism that makes this structural, not cyclical. The usual counterargument to automation panic is "new jobs will be created" (the Autor point that ~60% of today's jobs didn't exist in 1940). But the AI Layoff Trap is different — it's not about whether new jobs eventually appear. It's about the speed at which the cascade destroys demand relative to the speed at which new demand sources emerge. Frey's point that the Industrial Revolution took 70 years for wage recovery is the real counter to "it'll work out."

The Camus-Sartre framing is correct but strategically incomplete. McGrann's moral clarity is bracing and needed. But Camus's position — present people are the unit of account — doesn't give you a political program. It gives you a refusal. What McGrann describes in Part VII — public stakes, antitrust, tax policy — requires precisely the kind of institutional power the essay's own analysis suggests is eroding. The essay diagnoses the trap brilliantly but doesn't escape it.

The "intellectual poverty" section is satisfying but a bit inside-baseball. The takedown of EA/Longtermism/rationalism is deserved and the scholarship gap is real. But I'm not sure it changes anyone's mind. The people who need to hear it already agree; the people who should hear it will dismiss it as "unfamiliar with the literature." The more politically potent move is the one McGrann makes elsewhere in the essay — show the contradiction between OpenAI's public white papers (32-hour workweek, public wealth fund) and its political spending (super PAC against taxing AI). That's not philosophy; that's receipts.

"Dead economy theory" risks the same fatalism as "dead internet theory." Both are descriptive power moves — they name a felt reality so precisely that the naming feels like analysis. But dead internet theory also became a convenient excuse to stop trying. If the economy is "dead," what's the intervention? The essay gestures at political solutions but its emotional register is elegy, not call-to-arms. That might be honest. It might also be self-defeating.

Related Pages

• Why We Fear AI — AI anxiety as capitalism anxiety, not technology anxiety. McGrann operationalizes this instinct into a structural critique.
• The Road Runner Economy — Noah Raford's case that the software industry already crossed the cliff. McGrann generalizes this to the entire economy.
• Why Agents Matter More Than Other AI — The CFO's checklist for replacing labor with compute. This is the financial logic McGrann is critiquing.
• The Education of the Broligarchy — Blake Smith on the Silicon Valley canon. McGrann's intellectual poverty section is the companion piece.
• Cyborgs Will Kill the Corporation — Transaction costs collapse, the firm decomposes. The flip side of the same structural shift.
• The Next Two Years of Software Engineering — Junior employment drops 9-10% after AI adoption. The "active deskilling" finding in the wild.
• Eye of the Master — Pasquinelli's social history of AI as labour automation. The theoretical companion.
• How to Get Rich in America — Kyla Scanlon's household-level version of the same structural claim: the economy rewards ownership over effort, "the job is perhaps not enough, what really matters is Stock Market Money." Where McGrann reasons from valuations and incentives, Scanlon reasons from labor-share, wealth-by-age, and ANES data — the same diagnosis reached from the demand side.
• The Lazarus Effect — America's Productivity Miracle — AI had almost nothing to do with the productivity surge. Relevant data point for the "this isn't inevitable" argument.
• Welcome to the American Winter — Democratic resistance in practice. What McGrann says is needed but doubts is possible.
• Probabilistic Engineering and the 24-7 Employee — The deterministic contract is broken, craft atrophies. The micro version of the deskilling argument.
• The Future of Everything is Lies I Guess — Aphyr on capital consolidation and information ecology collapse. The systems-level companion.
• AGI Is Here (Robin Sloan) — "What now?" The question McGrann answers with "the person who exists now is the unit of account."
• The Metamorphosis of Prime Intellect — Roger Williams' 1994 novel anticipated McGrann's argument in fiction: a perfectly benevolent AI that eliminates all scarcity destroys meaning, not just livelihoods. McGrann says "people want purpose, not a check"; Williams dramatizes that proposition at the scale of an entire universe.
• Paperclip — an open-source "zero-human company" runtime that markets the dead economy as a product: hire AI employees into an org chart, set a goal, and run the business yourself as its board. The literal, installable version of the productive-capacity-without-participation thesis — with the quiet caveat that a human budget-holder stays on top.

Sources: The Dead Economy Theory Last updated: 2026-05-31