Patrick Collison — Why Silicon Valley's most talented should leave

Patrick Collison questions whether young people should universally move to San Francisco. The city celebrates independent founders who reject conventional wisdom, a culture that has produced figures such as Steve Jobs and Bill Gates and, in a different way, Stripe. But he argues that the world also needs people who spend decades developing deep technical expertise. He points to Genentech co-founder Herb Boyer, whose recombinant-DNA work made inexpensive insulin possible only after years of scientific training, and to Patrick Hsu's bridge-editing research at Arc Institute. These examples require sustained study and laboratory skill, not just entrepreneurial speed.

Collison avoids prescriptive advice about which path any individual should take. He praises the host's work as a 23-year-old podcaster because disseminating information can create value by connecting people to opportunities, such as the Scroll Prize. He suggests that readers drawn to his career may choose a similar path, but stresses that many inventions and accomplishments he values came from trajectories unlike his own. He also cautions against San Francisco's unusually high status for entrepreneurs: entrepreneurship is valuable, but should not become the only respected ambition.

The exchange distinguishes principled independence from reflexive contrarianism. Collison says scientific fields often have too many or too few adherents, and the same may be true of effective altruism. But automatically opposing the prevailing mood is just following it with the sign reversed. The herd is powerful, and resisting trends is difficult in practice. For long apprenticeships, he recommends finding the place where learning is steepest rather than automatically choosing the most credentialed or conventional route. Biology, for example, requires extensive bench skills and factual knowledge; top laboratories provide practical experience that few pure autodidacts acquire on their own.

Collison draws on the book Apprentice to Genius to explain how scientific mentorship transmits more than facts. One crucial skill is selecting problems that are important enough to matter if solved but tractable enough to make progress. Another is learning what high standards actually look like by working alongside people and institutions that embody them. For a young person, the practical task is to identify interesting fields, then seek environments where excellence is visible firsthand.

The host challenges whether marginal improvements to institutions such as the NIH can overcome ideas becoming harder to find. Collison objects to treating money as a fixed input-output lever. The cost of a factory or bridge depends on how it is built, and the binding constraint may be permits, labor, or other factors rather than funding. Comparing research budgets alone therefore risks mistaking spending for scientific output. He notes that the United States had roughly one percent as many professional scientists before World War II as after it, yet it produced substantial discoveries in the earlier period; research spending later grew by more than two orders of magnitude without an obvious linear relationship to output.

Collison favors examining concrete conditions at the level where research happens. A Fast Grants survey found that 79 percent of recipients said their research agenda would change substantially if they could direct current funding more freely. Rather than asking only how much the NIH budget should increase, he asks how constrained grantees should be in choosing their research. Even a fivefold improvement from changing that constraint is conceivable, though he does not claim it would happen.

The host distinguishes money from the broader question of how inputs become outputs. Across sectors, including areas with no direct connection to NIH, researchers increasingly need exponentially greater effort to sustain progress. He introduces one-time 'level effects' such as the COVID response, which can cause a sharp temporary increase in output. The unresolved question is whether such bursts raise long-run growth rates or merely shift the level of activity while underlying scientific progress continues to slow.