Hindsight Bias

Definition

Hindsight bias is the tendency, once an event has occurred, to believe you could have predicted it. Memory does not faithfully preserve our past states of uncertainty: it rewrites them in light of what we now know.

Baruch Fischhoff demonstrated this experimentally in 1975: participants told the outcome of an event systematically estimated they would have assigned it a high probability beforehand, even when they had not.

Why it matters

This bias is particularly dangerous in the evaluation of past decisions. It makes post-mortems inaccurate: decisions are judged with the certainty of hindsight, without reproducing the uncertainty of the moment. It also fuels overconfidence: if we feel we “knew” in advance, we mistakenly believe we are good forecasters.

In organizations, it leads to biased retrospectives (“it was obvious that project would fail”) and a poor learning culture.

Concrete examples

Stock market crash: “Everyone could see it was going to collapse.” A retrospective certainty that erases the genuine uncertainty of the time.

Election result: “With a candidate like that, the result was never in doubt.” Yet nobody announced it with confidence before the vote.

Production bug: “We should have tested that case anyway.” But the case was not obvious before the bug occurred.

Industrial accident: after the fact, investigators underestimate how ambiguous the warning signals were before the event.