Thinking in Bets

Duke played professional poker for two decades after leaving a doctoral programme in cognitive psychology, and she argues that decisions should be treated as bets, meaning commitments of resources to an uncertain outcome based on incomplete information. The framing matters because it separates decision quality from outcome quality.

Her central concept is resulting, the error of judging a decision by how it turned out. She argues that in any domain with substantial variance, good decisions produce bad outcomes and bad decisions produce good outcomes regularly, so an outcome-based evaluation trains the wrong lessons. She also covers the asymmetry in how people attribute outcomes, taking credit for good results and assigning bad ones to luck, and she notes that this pattern destroys learning from both.

Her practical recommendations include stating confidence levels explicitly as percentages rather than as binary beliefs, offering to bet on a stated position as a test of how much you actually believe it, and forming a truth-seeking group with explicit rules that reward accuracy over agreement. She draws on Tetlock's forecasting work and on Mercier and Sperber's argument that reasoning evolved for persuasion rather than for accuracy.

She also covers temporal tools including backcasting, premortems and what she calls a decision swear jar, and the practice of separating your present self from your future self when making choices. The book is short and repeats its central point. The poker examples do not transfer perfectly, since poker has defined probabilities and fast feedback that business decisions lack. There is little original research. The separation of decision quality from outcome quality is not new but is stated here more usefully than in most sources.