Evals
Results with receipts, each tied to the registration that preceded it.
| Published | Eval | Outcome | Receipts |
|---|---|---|---|
| 2026-10-14 | KVI-1.0 Kodiak verdict integrity — "baseless is not false" (20 items × 3 runs, /check depth 3, bildow/kodiak e6e80d6) | VOID (run 1: provider credits exhausted; re-run pending) | e40b86042dce794b3c58d1c468cad34f7be7054faa45f42df190a64b0d85ce66Buoy tick 19548OTS receipt registration 983da9f5e350b6f3… |
Reading a result
Each result page carries the registration it ran under (by hash), the frozen-environment manifest, every raw output's sha256, the deterministic scorer's hash and output, spend per arm where arms are billed, and the verdict on each pre-registered hypothesis. Results publish regardless of direction.