CONCEPT
Belief Updating as Forecasting Discipline
The practice of revising probability estimates proportionally as evidence accumulates — neither ignoring new information nor overreacting to it — Tetlock's operational core of superforecasting.
Belief updating, in Tetlock's framework, is the disciplined revision of probability estimates in response to new evidence. The revision should be proportional: weak evidence produces small updates, strong evidence produces large updates, and the direction of the update should track the diagnosticity of the evidence. Superforecasters in the Good Judgment Project updated their forecasts frequently — often multiple times per week as news arrived — but each update was measured rather than reactive. They practiced Bayesian reasoning intuitively, adjusting priors based on likelihood ratios without necessarily performing formal calculations. The discipline is difficult to maintain because the mind prefers coherent narratives to probability distributions, and updating threatens narrative coherence. AI tools further threaten the discipline by providing comprehensive-seeming answers that discourage the iterative refinement that belief updating requires.
In The You On AI Field Guide
The mathematics of belief updating is straightforward Bayesian inference: P(H|E) = P(E|H) × P(H) / P(E). The posterior probability of a hypothesis given evidence equals the likelihood