Fin Moorhouse - Longtermism, Space, & Entrepreneurship

Fin Moorhouse is a research scholar at Oxford's Future of Humanity Institute, an assistant to Toby Ord and host of Hear This Idea. He says he has failed to specialize and dabbles in projects that look underrated. He joined effective altruism (EA) because he loved philosophy, realized after reading Peter Singer that he was not living out the ideas, and found a likable student community. The pandemic gave him a break to read and learn to code before low-odds applications worked out.

Moorhouse helped set up two prize schemes: five $100,000 prizes for blogs on EA-related ideas, and five $20,000 prizes for criticism of EA. His argument is that a movement aiming for near-correct beliefs must be anti-fragile about being wrong and celebrate public changes of mind.

On timing, EA is starting to do large things and funders want ambitious, scalable projects. At that inflection point criticism is cheaper to absorb early, a point he credits to Will MacAskill. Other movements ossify into socially rewarded beliefs whose questioning becomes too costly. Shutting a project down after poor feedback should count as success.

The host suggests a $20,000 prize could help someone like SBF allocate about $10 million better, and asks if any other movement funds its own critics. Moorhouse has no good answer beyond political liberalism, and says following through matters more than claiming openness.

the proof is in the pudding here

For-profits learn quickly when the world is not excited, while non-profits lack that loop, so charity evaluators supply a market-like third-party check. The host expects mistakes in a ten-year-old movement. Moorhouse says good criticism can add a missed distinction, and defends progress in non-empirical fields as conceptual engineering: building useful structures on distinctions.

The host asks why the idea that the future is vast took thousands of years. Moorhouse cites historian Tom Moynihan and says we underrate the conceptual apparatus we take for granted. Formal probability and the idea that humanity can prevent its own premature end are recent. He admits his answer was weak.

a bit like the water we swim in now and so it's hard to notice

Asked whether something as big as AI alignment may still be undiscovered, Moorhouse says the most useful concepts were probably the lowest-hanging, but calls this weak and vague. He adds that progress per person looks more linear, and, citing David Deutsch, that new ideas always generate new problems. Preventing catastrophic risks therefore buys time for further progress and stays robustly important.

The host doubts physical limits constrain things, since limits such as computation per matter were undefinable before the theory of computation, and knowing them would not reveal what is possible. Moorhouse concedes some risks may be missed because they cannot be articulated, though roughly we know what matters most.