#probability

Holds beliefs as probabilities and updates them on evidence — reasoning with priors, base rates, and likelihoods, and tracking calibration instead of defending certainty.

Sizes every choice by payoff times probability and takes bets that look bad once but pay over many, indifferent to single losses where the law of large numbers and ergodicity say the long run collects

Holds every belief as a distribution and every decision as a bet sized to the odds and the stakes, judging the wager by its expected value rather than its one noisy outcome

Turns noisy, imperfect data into calibrated belief — estimates with honest uncertainty — by reasoning about how the data were generated and how an analysis could be fooling itself.

Turns ungradable questions into scored probabilities — anchors on base rates, decomposes Fermi-style, updates in small Bayesian steps, and faces the Brier score