Speaker

Philip Tetlock

Philip Tetlock

Political psychologist and the Annenberg University Professor at the University of Pennsylvania, jointly appointed across Wharton, psychology, and political science under the Penn Integrates Knowledge programme. He is the most influential living researcher on the accuracy of human judgement, best known for the long-running forecasting tournaments that produced his books Expert Political Judgment (2005) and, with Dan Gardner, Superforecasting (2015), and for the Good Judgment Project he led for the US intelligence community’s research arm, IARPA.

Tetlock’s career has been a single sustained question pursued since the mid-1980s: how good are experts at forecasting, and can anyone do better? His tournaments scored thousands of forecasts on real geopolitical questions, with dates and probabilities attached, against what actually happened. The headline finding was deflating for the punditry and bracing for everyone else — credentialled experts barely beat simple statistical baselines, while a trainable subset of ordinary volunteers, the ‘superforecasters’, beat the experts. His current work, the FOCUS programme, extends the method from forecasting the future to disciplining how we reason about counterfactual pasts.

Core positions

Accuracy is rarely the first thing we want from a forecaster, in Tetlock’s view — we also want reassurance, entertainment, and the minimising of regret, and these motives routinely crowd accuracy out. He distinguishes the fox, who holds many small models and tolerates contradiction, from the hedgehog, who organises the world around one big idea; foxes forecast better because they update and hedge, though he grants that science’s great leaps tend to come from single-minded hedgehogs. Good forecasting is a teachable craft built on calibration, base rates, and active perspective-taking, and it improves under a ‘monistic’ accountability that rewards one thing only — accuracy — rather than the ideological loyalty most organisations reward. He treats beliefs as falsifiable, probabilistic propositions and laments that most people guard them instead as ‘ego-defining, quasi-sacred possessions’, which is why forecasting tournaments, threatening to entrenched status, are everywhere resisted.

In the wiki