Skip to content

Close decision needs a salvageability axis — defect confidence alone does not predict close-correctness #8962

Description

@JSONbored

Advances #8828.

Empirical finding from the 2026-07-26 backfill adjudication (331 historical decisions, three repos).

The AI defect-finding confidence is calibrated to "is the defect real" — but close-correctness depends on a second, independent axis: is the PR salvageable and the author responsive. The data: 28 contributor PRs held on real, high-confidence defect findings (0.55–0.97) later MERGED — 17 as-is (finding refuted) and 11 after the author fixed the flagged defect (finding right, but a close would still have denied a real contribution). Meanwhile acted closes at the same confidences ran 96.0% correct. Same confidence, opposite close-correctness, split by salvageability.

Implication for the 99.5% goal: raising the confidence floor alone cannot get closes past ~96% — the residual error class is 'real defect, salvageable PR'. The close criterion should combine:

Requirements:

  1. Define the salvageability signal as a deterministic, explainable score (no new AI call) computed at decision time and persisted in the decision record (schema bump).
  2. Gate wiring: high-confidence defect + high salvageability → hold-with-guidance instead of close; high-confidence + low salvageability → close as today. Manifest knob for the salvageability threshold, following the config-as-code pattern.
  3. The decision-audit rubric gains a reason category for the new boundary so future labels measure it.
  4. 99%+ branch-counted coverage on changed src/** lines; golden-corpus entries (eval: versioned golden replay corpus gating every gate/prompt/model/config change #8832) for both sides of the new boundary.

Metadata

Metadata

Assignees

Labels

maintainer-onlyOwner-only work — yields no Gittensor points.

Projects

No projects

Relationships

None yet

Development

No branches or pull requests

Issue actions