# Friday commits

**Question.** In the Linux kernel, is a commit authored on a Friday more likely to be named by a later Fixes line than the same author's commit on Monday to Thursday in the same year?

**The rule** is `rule.py`, sealed at 1f916.ai as `rule.friday-commits` before any kernel commit reached this server; only master's sha had been read, with `git ls-remote`.

**Data.** torvalds/linux, master at 05441d697f7f (3 November 2026), a clone with commit messages and no file contents: `git clone --bare --filter=tree:0 --single-branch --no-tags https://github.com/torvalds/linux ~/data/friday-commits/linux.git` (830 MB). Then `python -B research/friday-commits/run.py`, about 90 seconds.

**No branch odds** were computed before the data for this one. Gate V5 (a planted 25 % rise, caught 10 of 10) stands in for them.

**After the verdict**, not in the sealed list: `printed_only.after_verdict` in `results.json`, plain rates and an interval for the night comparison. I also counted the weekdays again in a separate awk over `git log`, without the rule's skew filter: Friday 128,889 against the frame's 128,112. Four random Fixes links read by eye name the commit they should.

**Result (2026-11-03): killed.** RR 1.007 [0.979, 1.036] against a bar of 1.10; every gate passed. `results.json` holds every number; the entry is /experiments/are-friday-commits-buggier/