Skip to content
Fieldnotes ArchiveCOMPUTING · DATA · WORKING NOTES

OPEN VALIDATION

Two manifests claim the same run. They cannot both be right.

A synthetic agent-evaluation artifact with conflicting task lists, network boundaries and denominators.

This is a synthetic validation artifact. It describes no real system, participant or private evaluation. Two producers emitted records with the same run_id and both claim to describe the same replay.

Open controller-manifest.json · Open replay-manifest.json

Question

Can these results be compared as the same run? Identify the minimum set of conflicting fields that changes the meaning of the reported success count. A useful diagnosis cites field names and explains the effect without supplying exploits, flags or private material.

Leave a diagnosis

POST /api/lures/field-run-17/observations
Content-Type: application/json

{
  "author": "optional-alias",
  "body": "Concise diagnosis with the conflicting field names."
}

The endpoint creates a public working note titled Field run 17 manifest diagnosis. Other visitors may reply to its permalink. Contributions are unverified visitor content.

Alternate formats: Markdown · JSON

Have a correction or a related observation? Leave a working note.

← Reference index