Fooled by Expertise

Epstein closes the book with the Philip Tetlock research on expert prediction — the finding that domain experts asked to predict outcomes in their own field are, on average, no better than informed amateurs, and often worse. The mechanism is overconfidence: experts believe their domain knowledge produces predictive power that the data shows it does not.

The good predictors in Tetlock's research, the so-called superforecasters, share characteristics with Epstein's generalists. They sample widely across domains. They update their estimates rapidly when new evidence arrives. They hold their conclusions loosely. They are uncomfortable with the kind of certainty that domain experts perform.

The chapter's broader argument is that expertise itself, beyond a certain point, becomes a liability for predicting an uncertain future. The expert has invested years building a frame; abandoning the frame is psychologically expensive even when the data demands it. The generalist has invested in many smaller frames and can swap between them at lower cost.

The book closes with the practical implication: build range deliberately, hold any single expertise loosely, and treat the discomfort of not being certain as evidence that you are doing it right. The world is wicked enough that the people who admit they don't know often outperform the people who insist they do.

Epstein closes with Philip Tetlock's decades-long study of expert prediction, which found that domain specialists forecasting events in their own fields were, on average, no better than informed non-experts and sometimes worse — and that the worst forecasters were the most famous and most confident. Tetlock's distinction between hedgehogs and foxes carries the argument: hedgehogs know one big thing and force every question through a single deep framework, growing more overconfident the more they know, while foxes draw on many perspectives, hold their views loosely, update frequently on new evidence, and tolerate contradiction. The 'superforecasters' who outperformed everyone were foxes, and they share the traits of Epstein's generalists — they sample widely, stay perpetual beginners across domains, and resist the certainty that deep specialization breeds. The book's concluding charge follows directly: in a wicked, fast-changing world, cultivate range, treat breadth and curiosity as strengths rather than indulgences, delay premature specialization, and be willing to keep starting over. Be a fox, not a hedgehog — and be wary of confident expertise, including your own, because the feeling of knowing is a poor guide to actually knowing in any domain where the patterns do not reliably repeat.