Drethelin Gardens (Hut)
Optimal Science Funding
- Scientific philosopher at CMU working on game theory, formal epistemology, metascience, and AI safety — builds interactive models of how groups reason.
3.3/35 attending
Chance I’d attend
Who should grantmakers fund: the people with the strongest track records, the most latent promise, or the biggest resource bottlenecks? We model grantmaking as a Bayesian allocation problem where publication and grant records are noisy evidence about two hidden traits: capabilities and resources. The funder must decide both whom to support and when: early grants can generate output, build capacity, and reveal talent, while later grants benefit from better information. I’ll use the model to frame an interactive discussion about peer review, seed funding, inequality, exploration, and what grant systems should optimize—with the same logic applying, perhaps even more sharply, to automated AI research.