CHAPTER 18
Taming Intuitive Predictions
Read It
Predicting Julie’s college grades, we hold only the evidence of her early reading. Intuition says: she’s that smart, so she’ll be that good. But the correlation between evidence and outcome is imperfect, often weak. The real dilemma is that following intuition systematically overshoots extreme evidence, while full regression to the baseline throws away all information. Kahneman’s correction doesn’t kill intuition; it discounts it by the quality of the evidence. The weaker the evidence, the closer the prediction should sit to the average.
Open full image ↗Draw It
Prose tends to collapse “evaluating” and “predicting” into one act. Judging how impressive Julie is now and forecasting how far she’ll go are different questions. Intuition treats the first as the second, carrying the present impression straight into the future. Once the relationship is drawn, the two nodes separate, and a “correlation” valve appears between them. How open that valve is determines whether the prediction leans toward intuition or toward the baseline. The structure makes “why can’t I just trust my gut?” concrete rather than a vague warning.
Rethink It
In promotion reviews, one candidate’s outstanding year easily becomes a prediction that next year will be just as strong. But how much of that year was luck from environment, project, or team fit? If those factors correlate weakly with individual ability, the “same again next year” forecast should regress toward the average. That’s not dismissing the candidate; it’s matching judgment to evidence quality. Otherwise every promotion pays for extreme performance and systematically overestimates how stable people are.
Take It With You
How much to adjust an intuitive prediction depends on the correlation between evidence and outcome, not on how strongly the evidence feels. The stronger the feeling, the more I need to ask: how much of this is signal, and how much is just a smooth story?