Why Does the Past Seem Predictable Once We Know the Outcome?

After a project runs late, a team may say, ‘The requirements changed several times—we should have seen it coming.’ Before the delay, however, the same changes might reasonably have looked like ordinary adjustments. Hindsight bias is the tendency to judge an outcome as more predictable after learning it, and sometimes to remember that we ‘knew it all along’.

In Fischhoff's classic experiments, participants read accounts of historical events and estimated the likelihood of several possible endings. Those told that one ending had occurred gave it a higher retrospective probability, even when instructed to ignore that information, and reassessed which clues mattered. Later reviews distinguish several connected forms of the bias: misremembering one's earlier judgement, seeing the outcome as almost inevitable, and overestimating one's own ability to have predicted it. Once the result is known, supporting clues form a coherent story more readily while unrealised possibilities fade into the background.

This does not make retrospective analysis useless. A delayed project may genuinely have contained recognisable risks. The problem is that outcome knowledge can contaminate our judgement of what was knowable at the time. Experiments that control whether participants receive outcome information support a causal effect, although its size varies with the task, expertise and measurement. A fairer review preserves prior forecasts, alternatives and the evidence then available, and evaluates the decision process rather than inferring responsibility solely from the final result. Otherwise, we may produce a tidier story rather than better learning.

https://pmc.ncbi.nlm.nih.gov/articles/PMC1743746/
https://pubmed.ncbi.nlm.nih.gov/26168501/