The Discipline of Changing Your Mind

Two failures look like opposites but share a root. The first is the person who never updates — who holds a belief at full confidence no matter what arrives. The second is the person who over-updates — who lurches to a new certainty with every headline. Both have broken the same instrument: the one that converts evidence into belief at the right exchange rate.

That instrument has a name. Bayes’ rule.

Informally: your new belief is your old belief, moved toward the evidence in proportion to how surprising and how reliable that evidence is. A strong prior resists weak data. Strong data overrides a weak prior. Nothing moves all the way to 0 or 1.

The dashed curves are the prior (left) and the evidence (right). The solid curve is the posterior — it sits between them, pulled toward whichever is sharper. It did not jump to the evidence. It did not ignore it. It moved by the right amount.

Why it feels like weakness

Updating in public looks like inconsistency, and we are punished socially for it. So we calcify: better to be wrong with conviction than right by revision. But conviction is not a virtue — calibration is. The honest question is never “did you change your mind?” It is “did you change it by the weight of the evidence?”

Operating it

• State the prior out loud. A belief you can’t put a number on can’t be updated, only defended.
• Ask what would move you, and how far. Before the data: “If I saw X, I’d shift from 70% to 50%.” Now the update is arithmetic, not ego.
• Distrust loud, cheap evidence. Surprise is not the same as weight. A vivid anecdote is a small likelihood ratio wearing a big costume.
• Don’t go to the rails. Reserve 0 and 1 for logic and death. Everything else is a probability.

Changing your mind well is not flip-flopping. Flip-flopping is un-weighted updating. The discipline is to move — but only as far as the evidence can carry you.

Hold your beliefs like positions, not identities. Mark them to market. Update on the close.