Concept

win / tie / loss and win-rate

For each question: did the judge prefer BOAR's answer, the reference, or neither? BOAR won 0 of 51 bounty-style questions.

From the record

In technical terms: Per-item combined winner over both orders (disagreement = tie). Reported, never gated (since 2026-10-10). 0 wins on every Claude-judged row of V-para and v0.2; isolated wins on v0.1 dev only.

Win-rate against the frontier reference: zero on both bounty-relevant sets under the Claude judge; a few isolated wins exist only on the public dev set.Win-rate against the frontier reference: zero on both bounty-relevant sets under the Claude judge; a few isolated wins exist only on the public dev set.0 wins on every row and seed of V-para and v0.2 (Opus); v0.1 dev: 1/57 under Opus for the deep R450, LES and 4B rows; 2/57 under Haiku on the 12 GB-class phone runSetV-para; v0.2; v0.1Statusdev-style / dev / public devJudgeClaude Opus; Claude Haiku L1 (phone run)n51 / 110 / 57SeedsallDate2026-10-14CaveatPublic text that says "0 everywhere" (boar-research README.md:42, boar-bench TRACE-FORMAT.md:70) overstates the scope.Sourceresearch record reports/BAR-TABLE.md:80research record arms/device-x6/DEVICE-X6.md:27Open the data file
How the win-rate condition changed: secondary metric in the plan, briefly part of the bar in arm reports, then "reported, never gated".How the win-rate condition changed: secondary metric in the plan, briefly part of the bar in arm reports, then "reported, never gated".2026-10-02 plan: W/T/L secondary; 2026-10-08/09 arm reports read the bar as "floor-adjusted CI low > 0.50 AND win-rate > 0"; 2026-10-10 BAR-TABLE: win-rate reported, never gatedDate2026-10-10Kinda record, not a measurement (set, status, judge, n, seeds: not recorded)Sourcebenchmark bench/v0.3/plan/analysis-plan.md:30research record reports/BAR-TABLE.md:7Open the data file