The practice of potentially stopping a study early based on the results so far. Also sometimes called “continuous monitoring”. This is invalid in much of modern statistics, and solving this problem is a cornerstone of safe, anytime-valid inference (SAVI). It’s made possible with e-processes.