[feedback] shared-model-judges: Two same-model judge seats produced largely disjoint findings (a strength) but shared one model-level bias (FB

Context. Solo-operator project using Claude Code (desktop app, Windows 11) with a lead session on Claude Fable 5.1 orchestrating Opus 5.5 / Sonnet subagents (worktree-isolated), a JSON task store, PreToolUse/SubagentStop hooks and a harvest-train commit flow. This finding was measured on or before 2026-09-22 and filed on 2026-09-28 at the operator's request; numbers are from tool output. No secrets, identities or private paths.

Class / model / severity. shared-model-judges / Opus (judge seats) / MEDIUM

Finding. Two same-model judge seats produced largely disjoint findings (a strength) but shared one model-level bias (FB-003); a seat reversed its view after reading the other but kept its recorded verdict (correct).

Evidence (measured). reports/j_M4_M5_M3_r2_seat1 and seat2; the concurrence reads.

Minimal reproduction. Run the same review with two instances of one model vs two models.

Ask. For safety adjudication, diversify instrument and model, not only instance.