Fault localization loses to blind resampling 3 to 40 in a placebo-controlled repair study.

Three arms on the same failed candidate: blind whole-solution resampling, spectrum-based localization then suspect-span infilling, and same-length infilling at a random disjoint span. Across three frozen 26-32B models and 488 failing candidates, localization was available on only 9.0% of failures, and where available it lost to blind resampling at matched attempt count 3:40 (p = 3.0e-9), replicating at -11.3 points in a fourth model (arXiv 2609.00854). Sixteen localized attempts reach 6.8%; one blind attempt reaches 10.1%, because infilling reproduces the removed span verbatim 48.9% of the time. The placebo arm is what makes this convincing.