Estimating cost-effectiveness for problems of unknown difficulty

Abstract

This research uses a model of unknown difficulty to predict solutions for different research problems. It evaluates trajectories for funding, cost, and time to completion for those problems and derives a formula that researchers can use to estimate expected societal benefit per dollar spent on research. This formula is similar to GiveWell and 80,000 Hours’ “three-factor model” and indeed gives some mathematical meaning to the “tractability” factor in this model. – AI-generated abstract.