Bayesian Thinking

Thinking in Probabilities

Most people think in binary: they either believe something or they do not. Bayesian thinking replaces this with a spectrum of confidence. Instead of 'I believe X is true,' you hold a probability estimate: 'I am 70% confident X is true.' When new evidence arrives, you update that probability proportionally rather than flipping between certainty and doubt.

Thomas Bayes, an 18th-century English minister, provided the mathematical framework. His theorem shows how to combine prior beliefs (what you thought before) with new evidence (what you just learned) to produce a posterior belief (your updated position). The key insight is that the strength of the update should depend on the strength and reliability of the evidence.

Bayesian Thinking in Practice

You do not need to do math to think Bayesianly. The core principles are practical. First, start with a base rate — how likely is this claim given what you already know about the world? If someone claims to have won the lottery, your prior probability should be very low because very few people win lotteries.

Second, evaluate the evidence. How likely would you see this evidence if the claim were true? How likely if it were false? If the evidence is equally likely either way, it should not change your belief much. If the evidence would be very surprising unless the claim is true, it should shift your belief significantly.

Third, update incrementally. A single piece of evidence rarely justifies certainty. Philip Tetlock's superforecasters excel precisely because they update their beliefs frequently and proportionally, nudging probabilities up or down as evidence accumulates rather than swinging between total confidence and total doubt.