docs/experiments/README.md says it is the index of this directory, "newest first". It is not current:
- Six records are absent entirely —
2026-09-01-cusum-event-first-measurement.md, 2026-09-01-triple-barrier-first-measurement.md, 2026-09-02-equities-cost-fidelity.md, 2026-09-03-equities-dca-benchmark.md, 2026-08-27-pooled-review-preview.md, and there is no ## 2026-09 section at all.
- The file's own preamble makes the index a claim about the directory: "Index is newest first, by the date each document carries in its filename." An index missing a quarter of its records is worse than none, because it reads as complete.
This matters more here than in an ordinary docs directory: the README is the entry point a reader uses to find what has been measured, and three of the missing records are the ones that supersede figures other documents still carry.
Done when
docs/experiments/README.mdsays it is the index of this directory, "newest first". It is not current:2026-09-01-cusum-event-first-measurement.md,2026-09-01-triple-barrier-first-measurement.md,2026-09-02-equities-cost-fidelity.md,2026-09-03-equities-dca-benchmark.md,2026-08-27-pooled-review-preview.md, and there is no## 2026-09section at all.This matters more here than in an ordinary docs directory: the README is the entry point a reader uses to find what has been measured, and three of the missing records are the ones that supersede figures other documents still carry.
Done when
.mdindocs/experiments/has an index entry.pydrivers and.jsonlartifacts are named beside the record that cites them, per the directory's stated convention