Pre-registration

Writing down the hypothesis, metric, sample size, and pass/fail threshold, plus a dumb baseline to beat and an iteration budget, before any testing. Borrowed from clinical trials, it converts a multiple-testing problem into a single honest test because the rules are fixed before you can see which way to bend them, and it commits you to reporting a failure as a failure.

First used in Lesson 2.8 · Overfitting statistics and pre-registration — the lesson that makes this term real.