Hindsight bias
Researchers have tested this many times, in different places, over many years, and they keep finding the same result. This one is safe to trust.
In 1972, before President Nixon flew to China, two researchers ran a quiet experiment. They asked students to write down probabilities for specific outcomes. Would Nixon meet Mao. Would the trip be declared a success. The students committed their predictions to paper. Then the trip happened, and months later the researchers asked the same students to recall what they had predicted. The students misremembered, and they misremembered in one direction. Whatever had actually happened, they now recalled having given it a higher probability than they really had. Outcomes they had doubted, they remembered predicting. One of the researchers called the broader phenomenon creeping determinism, and it became known as hindsight bias. Once you know how something turned out, memory quietly rewrites what you believed beforehand, until the outcome feels like it was obvious all along.
Every debrief and every post mortem is built on memories that have already been edited to fit the result. The person who backed the failed project now genuinely remembers having had doubts. Nobody is lying, and that is what makes it hard to catch, because the reconstruction is automatic and it feels like recall. This is the reason you cannot audit your own judgment from memory, and it is the strongest practical argument for writing predictions down before you know the answer. A dated line on a page is the only version of what you believed that memory cannot quietly improve after the fact. The companion error, judging a decision by how it happened to turn out, is outcome bias.
Hindsight bias rewrites your memory of what you predicted, so the outcome feels like it was always obvious. Outcome bias rates the decision itself by how it turned out, even with identical information about how it was made. They arrive together in every post-mortem and they need different fixes. Hindsight is beaten by writing the prediction down before the result is known. Outcome bias is beaten by asking two questions in order: was this a good decision given what was known, and separately, did it work. The written note is what makes the first question answerable.
Source: Fischhoff and Beyth, I Knew It Would Happen: Remembered Probabilities of Once-Future Things, Organizational Behavior and Human Performance, 1975. One of the most reliably reproduced findings in the field, confirmed by meta-analysis in 1991.
Michael Lewis, 2016
On the two psychologists whose work produced this and much of the rest of the library, and on how they thought.
Draw your own card. It does not take long, and it rewards taking your time.