Where: .github/ISSUE_TEMPLATE/new_explainer.yml line 8.
The gap: the line reads: "Check the explainers/ folder first - proxy variables, equalized odds, sampling bias, SHAP values, disparate impact, fairness metric conflicts, and calibration are already done." That's the first 7 explainers ever added to the repo - it was never updated as the project grew.
Repro:
$ ls explainers/*.md | wc -l
53
$ grep -n "already done" .github/ISSUE_TEMPLATE/new_explainer.yml
8: ... proxy variables, equalized odds, sampling bias, SHAP values, disparate impact, fairness metric conflicts, and calibration are already done.
53 explainers exist; the template's "already done" list names only 7. scripts/check_explainer_count.py (which guards 5 other "N explainers" mentions across README/CONTRIBUTORS/METRICS/ROADMAP) doesn't scan .github/ISSUE_TEMPLATE/*.yml, so this list is completely unguarded against drift.
Why it matters: a contributor proposing a new explainer topic - e.g. "demographic parity" or "equal opportunity" (both already published) - has no template-level signal that 46 other topics are already covered, undercutting the template's own stated purpose of avoiding duplicate proposals.
Suggested fix: drop the hardcoded list in favor of "check explainers/ for existing coverage" plus a link to the site's explainer index, since a static list will always drift again as more explainers ship.
Where:
.github/ISSUE_TEMPLATE/new_explainer.ymlline 8.The gap: the line reads: "Check the
explainers/folder first - proxy variables, equalized odds, sampling bias, SHAP values, disparate impact, fairness metric conflicts, and calibration are already done." That's the first 7 explainers ever added to the repo - it was never updated as the project grew.Repro:
53 explainers exist; the template's "already done" list names only 7.
scripts/check_explainer_count.py(which guards 5 other "N explainers" mentions across README/CONTRIBUTORS/METRICS/ROADMAP) doesn't scan.github/ISSUE_TEMPLATE/*.yml, so this list is completely unguarded against drift.Why it matters: a contributor proposing a new explainer topic - e.g. "demographic parity" or "equal opportunity" (both already published) - has no template-level signal that 46 other topics are already covered, undercutting the template's own stated purpose of avoiding duplicate proposals.
Suggested fix: drop the hardcoded list in favor of "check
explainers/for existing coverage" plus a link to the site's explainer index, since a static list will always drift again as more explainers ship.