Irrationality

Stuart Sutherland

6 ideas

1. Interviews predict worse than simple formulas

   Unstructured interviews give selectors vivid but unrepresentative impressions, and they weight these more heavily than objective records such as past performance. Studies comparing clinical judgement with simple actuarial formulas find that the formula usually matches or beats the expert, even when the expert has access to the same data. Sutherland argues that institutions should therefore replace intuitive selection with mechanical scoring wherever outcomes can be measured.

2. Availability distorts estimates of frequency

   People judge how common or likely an event is by how easily examples come to mind. Recent, dramatic, or personally witnessed events are overweighted. This is why plane crashes and murders seem more frequent than they are, while undramatic causes of death such as diabetes or strokes seem rarer than they are.

3. Seek disconfirmation, not confirmation

   People test beliefs by looking for cases that fit and seldom look for cases that would refute them. In Wason's selection task most subjects choose the card that can only confirm the rule and ignore the card that could falsify it. Asking 'what would I expect to see if I were wrong?' exposes how little evidence many firmly held beliefs actually rest on.

4. Sunk costs trap rational abandonment

   People keep investing in failing projects because of what they have already spent, even though past expenditure cannot be recovered and should not affect the decision. Escalation is worse when the decision-maker is publicly responsible for the original choice, because withdrawing looks like admitting a mistake.

5. Obedience experiments show situational override of conscience

   In Milgram's studies, ordinary volunteers gave what they believed were dangerous electric shocks to a protesting stranger simply because an experimenter in a lab coat told them to continue. About two-thirds went to the maximum voltage. Sutherland uses this to argue that deference to authority, not individual wickedness, explains much harmful conduct in hospitals, the military, and bureaucracies.

6. Base rates are neglected for vivid specifics

   When given both statistical base rates and a descriptive case, people judge by how representative the case seems and largely ignore the base rate. The error is systematic among trained professionals, so expertise alone does not protect against it.

Save and mark ideas in the app