Tyler Cowen - The Great Reset

The interview opens as a test of a forecast rather than a discussion of current events. Dwarkesh quotes a 2017 passage from The Complacent Class in which Cowen warned that the country would eventually face an immediate crisis without the resources, or more fundamentally the flexibility, to handle it, and asks whether the pandemic is the great reset he had predicted. The framing matters: the question is not whether Cowen was right in general but whether a specific, dated claim about American institutional capacity survived contact with an actual shock.

Cowen's answer is more concessive than a vindicated prophet's would be. He says he needs to reread the book, and that he had imagined the reset as something five to ten years away rather than something arriving in 2020. He does not claim the timing was right; he claims the substance was. The distinction he draws is between fiscal room and behavioural room. The United States could still borrow at low interest rates, so the constraint was not money. The constraint was the ability to adopt mask-wearing, or to stand up a test-and-trace regime serious enough to matter.

I think I need to reread that book.

He then offers an institutional scorecard that is unusually specific for a public intellectual speaking in the middle of a crisis. The Fed and the Supreme Court come out well, not because he endorses every decision but because, as institutions, they kept operating. Everything else, in his phrasing, is rotting. The regulatory state, the executive branch and Congress are named as failures. This is a claim about throughput rather than ideology: the question is whether an institution can still convert decisions into action, and on Cowen's account only the two least democratic bodies in the federal government could.

The scorecard is memorable but under-evidenced within the conversation. No comparison is offered between, say, testing capacity in March and testing capacity in June, or between the speed of congressional relief legislation and the speed of central bank facilities. The reader is asked to accept the ranking on Cowen's authority. That is a real weakness, and it is worth flagging because the rest of the interview repeatedly returns to the same methodological move: a broad institutional judgement delivered with confidence and then qualified into unfalsifiability.

Asked whether the post-reset world will be dynamic again, Cowen immediately downgrades the prediction to a conditional one. This is the most disciplined moment in the first half of the interview. Rather than asserting a general return of American dynamism, he partitions the economy and makes a strong claim about one part and a weak claim about the rest.

The strong claim concerns biomedicine. Cowen argues that the volume of research activity directed at coronaviruses, virology, immunology, vaccines and pandemic response is large enough that the field will look substantially different in five to ten years. He is careful to separate that from the question of whether the pandemic was worth it. The gains are a by-product, not a justification, and he says so explicitly. He also attaches a number to his confidence, putting the probability well above fifty percent, which is rare in this kind of commentary and makes the claim at least nominally testable.

I think this will go down in history as a phenomenal blast of innovation in a very important area.

The weak claim concerns everything else. The tech sector is described as still innovative but has been for some time, which is a way of saying that the pandemic added little to a trend already in place. Other sectors are left as an open question. The one piece of market evidence he offers is Tesla's valuation, which he says heartens him as a signal of dynamism while conceding that he cannot see why the company should be worth more than General Motors, Ford and several other automakers combined. He resolves the tension by deferring to the market rather than by defending the valuation.

That deferral is the weakest link in the section. If the market's judgement is taken as evidence of dynamism when it is encouraging and bracketed when it is puzzling, the signal cannot do any work. Cowen's own hedge, that the chance of a burst of dynamism has never looked better but that the country is still capable of blowing it, closes the argument on a note that cannot be falsified by anything that happens next.

Dwarkesh raises what looks like a genuine inconsistency between two of Cowen's books. Stubborn Attachments argues for prioritising economic growth because of the enormous long-run returns to sustained growth. The Complacent Class, by contrast, works with a cyclical view in which ongoing progress is not guaranteed. If history cycles, the case for growth should be weaker, because the compounding never gets to run.

Cowen's reconciliation is a matter of time horizon rather than of substance. He accepts that there will be slower and faster periods, and dates the most recent slow period to roughly 1973. But he argues that if institutions remain reasonably liberal, growth can advance steadily across centuries even though any given decade may be stagnant. His self-description is precise: cyclical in the short run, but a believer in substantial persistence in culture and institutions. The Stubborn Attachments recipe is then claimed to hold over the longer horizon.

Maybe I’m cyclical in the short run, but I believe in a fair amount of persistence in culture and institutions

The reconciliation works logically but costs something the interview does not price. If a down-cycle can run for fifty years, as the 1973 date implies, then the argument from stupendous long-run returns is being applied across a period longer than any policymaker's or any individual's horizon. Cowen never specifies what ends a cycle or what makes institutions revert to the liberal configuration that the argument depends on. Persistence is asserted as a cultural fact rather than derived from anything, which means the two books are compatible only because the cyclical claim has been weakened enough to fit inside the growth claim.

The Stanford lecture on time horizons is introduced with a Greta Thunberg quotation dismissing eternal economic growth as a perverse fairytale, and Cowen's first move is to disclaim memory of the lecture altogether. He describes his episodic memory as extremely weak: good for facts, poor for narrative. He is confident the ideas were his and that he has not repudiated them, but he cannot reconstruct the argument. This is an odd but useful moment, because it forces the reconstruction to be done live rather than recited.

What he reconstructs is that the case for stubborn attachment requires a horizon that is neither too short nor too long. If the horizon for successful growth stretches out too far, the rational posture becomes total risk aversion, because the only thing that matters is that the world not end. Growth stops being interesting and safety becomes everything. He recalls estimating, half-jokingly, that civilisation had another seven or eight hundred good years in it, and notes that he knows this only because he saw someone quote it back to him.

I thought we had another good 700 or 800 years left in us

Dwarkesh then presses the obvious objection: perhaps growth is itself the risk-reducing strategy, since some problems are guaranteed to arrive and can only be solved with more capacity. Cowen agrees immediately and says he probably made the point in the talk. The concession is important because it reframes the whole lecture. The horizon is not a parameter to be tuned for its own sake; it is a constraint that determines which argument applies. A very long horizon does not refute growth, it changes what growth is for.

The qualifier Cowen adds is the most interesting idea in this stretch. A space optimist, in his sense, can hold a very long horizon without becoming obsessed with safety, because the payoff from settling galaxies is multiplicative and enormous. Cowen says plainly that he is not a space optimist. He treats the speed of light and the difficulty of travel as binding constraints, allows that there may be vacations on the moon, and concludes that the resource base is Earth.

I’m not a space optimist.