Predicting the Future Is Possible. These ‘Superforecasters’ Know How.
About this episode
Guest host Julia Galef interviews psychologist Philip Tetlock about the forecasting research that became 'Superforecasting.' In an IARPA-sponsored tournament, teams of ordinary volunteers — later dubbed superforecasters — predicted geopolitical events more accurately than CIA analysts with classified access, beating them by roughly 30 percent. Tetlock explains what separated the best: thinking in probabilities, updating beliefs like Bayesians when evidence changes, breaking hard questions into Fermi-style parts, seeking base rates and outside views, and keeping score with tools like Brier scores. They revisit his earlier finding that famous pundits perform worse than chance ('Expert Political Judgment'), discuss why the intelligence community and other institutions resist accountability, and consider what forecasting can teach policymakers about humility, decision hygiene, and preparing for the next pandemic.