Math RefereeIndependent assessments and attempt coverage
These are returned final assessments, not accepted campaign conclusions. Initial responses were produced independently; source adjudication and arbitration followed. A packet assessment is not a full-paper review. Failed/invalid attempts supply no mathematical verdict. Later coordinator checks can resolve context complaints left in an initial response.
| Attempt | Scope | Model / effort | Delivery | Final assessment |
|---|---|---|---|---|
| 1 | full-text / full_manuscript | gpt-6.1-sol / high | completed | JSON |
| 2 | local / narrative:94293 | gpt-6.1-sol / high | completed | JSON |
| 3 | local / prop:homogen | gpt-6.1-sol / high | completed | JSON |
| 4 | local / prop:liftingmeasures | gpt-6.1-sol / high | completed | JSON |
| 5 | local / thm:MS | gpt-6.1-sol / high | completed | JSON |
| 6 | local / theorem:superrigiditydichotomy | gpt-6.1-sol / high | completed | JSON |
| 7 | local / thm:main:homogeneous | gpt-6.1-sol / high | completed | JSON |
| 8 | local / thm:main | gpt-6.1-sol / high | completed | JSON |
| 9 | local / cor:kleinian | gpt-6.1-sol / high | completed | JSON |
| 10 | local / prop:liftingmeasures | gemini-4-argon / high | completed | JSON |
| 11 | local / narrative:14240 | gpt-6.1-sol / high | completed | JSON |
| 12 | local / lem:initobj | gpt-6.1-sol / high | completed | JSON |
| 13 | local / targeted packet | claude-opus-5-5 / max | provider_failed_or_incomplete | No accepted assessment |
| 14 | local / thm:dehnfilling | gpt-6.1-sol / high | completed | JSON |
| 15 | local / cor:knots | gpt-6.1-sol / high | completed | JSON |
| 16 | local / thm:BF | gpt-6.1-sol / high | completed | JSON |
| 17 | local / lemma:properness | gpt-6.1-sol / high | completed | JSON |
| 18 | local / lemma:compatible | gemini-4-argon / high | completed | JSON |
| 19 | local / lem:clarifying | gpt-6.1-sol / high | completed | JSON |
| 20 | local / targeted packet | claude-opus-5-5 / max | provider_failed_or_incomplete | No accepted assessment |
| 21 | local / lemma:compatible | gpt-6.1-sol / high | completed | JSON |
| 22 | arbitration-000 / targeted packet | gemini-4-argon / high | invalid_assessment | No accepted assessment |
| 23 | arbitration-001 / targeted packet | claude-opus-5-5 / max | provider_failed_or_incomplete | No accepted assessment |
| 24 | arbitration-002 / targeted packet | gemini-4-argon / high | invalid_assessment | No accepted assessment |
Original output hashes and the sanitized attempt ledger are available in attempts.json. No raw request, provider result, or reasoning stream is included.
The user separately supplied an Astra assessment and a Fable assessment. They were excluded from the initial native prompts. Their theorem-level repair endorsements remain attributed proposals; model IDs and effort settings were not independently authenticated, and subscription costs are unknown. The current report incorporates only the separately checked portions.
Additional supporting archive items are identified in the original report. View the manuscript hashes and repository snapshot.