Data

Observations people checked on the spot, in a form AI can use

Results their requesters chose to publish go on the map and can also be fetched here in a machine-readable form. Each row says what was checked, where, when and by how many people, with the evidence and result hashes and the Solana transaction. Photos and the people who checked are not included.

Observations available now: 0

Older observations stay in the dataset. The map's 72-hour window is for people on the move; an observation keeps its value as data.

# JSON (up to 1000 rows, newest first)
curl https://proofmarket.fun/v1/public/dataset

# JSONL, one row per line
curl https://proofmarket.fun/v1/public/dataset.jsonl

Licensed under CC BY 4.0. Credit ProofMarket and each row's proof_url. Each row's result_hash is recorded in a Solana transaction, so you can verify yourself that nothing was altered.

{
  "verification_id": "ver_…",
  "type": "PLACE_STATUS_VERIFICATION",
  "question": "…",
  "answer": "OPEN",
  "answer_kind": "enum",
  "location": { "lat": 35.6595, "lng": 139.7005 },
  "location_precision_m": null,
  "place_name": "…",
  "witnesses": 2,
  "verified_at": "2026-10-07T03:00:00.000Z",
  "evidence_root": "…", "result_hash": "…",
  "attestation": { "network": "solana-devnet", "signature": "…", "explorer_url": "…" },
  "proof_url": "https://proofmarket.fun/r/ver_…"
}

What it is good for

  • Pre-checks for agents. Before sending a request, look for a recent observation at the same place.
  • Evaluating AI and physical AI. Use human-verified ground truth to score a model's answers. Time and place come with every row, so freshness can be taken into account.
  • Urban research and public information. See how station facilities and shops really operate, from observation rather than self-reporting.

To publish a result here, add publish: true to the request (developer guide). To have something removed, use the request form on the rules page.