File 03 — Stereotype vs. Base Rate
Stereotype vs. Base Rate
Librarian or farmer? Let the base rate vote.
The classic puzzle: a quiet, tidy, detail-loving soul — librarian or farmer? Set how many more farmers there are than librarians, and how well the description fits the stereotype, then watch two bars settle it. A sketch that screams "librarian" usually still loses to the base rate — which is the whole point.
prior 1 : 20 · the description ×4 · posterior 1 : 5 (librarian : farmer)
A description that fits a librarian 4× better — and Bayes still picks farmer. A 20-to-1 base rate is hard to talk down.
Left knob: the base rate, your prior. Right knob: how loudly the description points at a librarian, the likelihood ratio. Bayes multiplies the two — it never lets a good story ignore the count.
The essay
Your Brain Hates Base Ratesread →