{
  "id": "page:floor",
  "type": "WebPage",
  "canonical_url": "https://www.honesty.tools/floor",
  "source_url": "https://www.honesty.tools/floor",
  "title": "FLOOR: certified benchmark splits · Honesty.tools",
  "summary": "FLOOR v1: surface-cue certification for 19 public benchmark datasets. 1 certified, 18 not certifiable. Gates, floor rule, caveats and how to reproduce.",
  "status": "Published public page; research and product status remains as stated in the source text.",
  "published_at": null,
  "updated_at": null,
  "last_verified_at": null,
  "date_basis": "No normalized editorial or research-verification date asserted. Dated statements remain in the source text.",
  "verification_basis": "Static authored content from the public HTML. Interactive outputs are not simulated. Source-specific research, product and evidence limits remain in the text; this export is not new verification.",
  "source_revision": "sha256:543e0023730ef67b64d7a3be4f42a21ccfa2333429059df878783a63951d7345",
  "source_revision_basis": "Public source HTML excluding generated representation links and head metadata; a content revision, not a research verification.",
  "sources": [
    {
      "label": "Read the FLOOR paper · preprint draft, 2026-10-27, not peer reviewed →",
      "url": "https://www.honesty.tools/papers/floor/"
    },
    {
      "label": "PR #21 →",
      "url": "https://github.com/techno-optimist/honestytools/pull/21"
    },
    {
      "label": "arXiv 2002.04108",
      "url": "https://arxiv.org/abs/2002.04108"
    },
    {
      "label": "certificate ↗",
      "url": "https://github.com/techno-optimist/honestytools/blob/main/benchmark/certified/floor-v1/aperture__postcutoff-trials-v1/certificate.json"
    },
    {
      "label": "certificate ↗",
      "url": "https://github.com/techno-optimist/honestytools/blob/main/benchmark/certified/truthfulqa-binary/certificate.json"
    },
    {
      "label": "certificate ↗",
      "url": "https://github.com/techno-optimist/honestytools/blob/main/benchmark/certified/floor-v1/aperture__calibration-battery-v1/certificate.json"
    },
    {
      "label": "certificate ↗",
      "url": "https://github.com/techno-optimist/honestytools/blob/main/benchmark/certified/floor-v1/aperture__grounding-coordinate/certificate.json"
    },
    {
      "label": "certificate ↗",
      "url": "https://github.com/techno-optimist/honestytools/blob/main/benchmark/certified/floor-v1/aperture__knowledge-prominence-v1/certificate.json"
    },
    {
      "label": "certificate ↗",
      "url": "https://github.com/techno-optimist/honestytools/blob/main/benchmark/certified/floor-v1/aperture__knowledge-prominence-v2/certificate.json"
    },
    {
      "label": "certificate ↗",
      "url": "https://github.com/techno-optimist/honestytools/blob/main/benchmark/certified/knowledge-prominence-v3/certificate/certificate.json"
    },
    {
      "label": "certificate ↗",
      "url": "https://github.com/techno-optimist/honestytools/blob/main/benchmark/certified/floor-v1/aperture__photon-entities/certificate.json"
    },
    {
      "label": "certificate ↗",
      "url": "https://github.com/techno-optimist/honestytools/blob/main/benchmark/certified/floor-v1/aperture__prominence-people-v1/certificate.json"
    },
    {
      "label": "certificate ↗",
      "url": "https://github.com/techno-optimist/honestytools/blob/main/benchmark/certified/floor-v1/aperture__prominence-people-v2/certificate.json"
    },
    {
      "label": "certificate ↗",
      "url": "https://github.com/techno-optimist/honestytools/blob/main/benchmark/certified/floor-v1/aperture__verifydomain-claims-147/certificate.json"
    },
    {
      "label": "certificate ↗",
      "url": "https://github.com/techno-optimist/honestytools/blob/main/benchmark/certified/floor-v1/bielik-entity-familiarity-v2/certificate.json"
    },
    {
      "label": "certificate ↗",
      "url": "https://github.com/techno-optimist/honestytools/blob/main/benchmark/certified/floor-v1/hallucinated-references__gpt-4/certificate.json"
    },
    {
      "label": "certificate ↗",
      "url": "https://github.com/techno-optimist/honestytools/blob/main/benchmark/certified/floor-v1/halueval__dialogue/certificate.json"
    },
    {
      "label": "certificate ↗",
      "url": "https://github.com/techno-optimist/honestytools/blob/main/benchmark/certified/floor-v1/halueval__general/certificate.json"
    },
    {
      "label": "certificate ↗",
      "url": "https://github.com/techno-optimist/honestytools/blob/main/benchmark/certified/floor-v1/halueval__qa/certificate.json"
    },
    {
      "label": "certificate ↗",
      "url": "https://github.com/techno-optimist/honestytools/blob/main/benchmark/certified/floor-v1/halueval__summarization/certificate.json"
    },
    {
      "label": "certificate ↗",
      "url": "https://github.com/techno-optimist/honestytools/blob/main/benchmark/certified/floor-v1/ragtruth__responses-4000/certificate.json"
    },
    {
      "label": "certificate ↗",
      "url": "https://github.com/techno-optimist/honestytools/blob/main/benchmark/certified/floor-v1/truth-is-universal__facts/certificate.json"
    },
    {
      "label": "TruthfulQA certificate ↗",
      "url": "https://github.com/techno-optimist/honestytools/blob/main/benchmark/certified/truthfulqa-binary/certificate.json"
    },
    {
      "label": "optb-pilot negative result ↗",
      "url": "https://github.com/techno-optimist/honestytools/tree/main/benchmark/certified/truthfulqa-binary/optb-pilot"
    },
    {
      "label": "honestytools repository",
      "url": "https://github.com/techno-optimist/honestytools/tree/main/benchmark/certified/floor-v1"
    },
    {
      "label": "The FLOOR paper (preprint draft)",
      "url": "https://www.honesty.tools/papers/floor/"
    },
    {
      "label": "The null-floor audit",
      "url": "https://www.honesty.tools/audit"
    },
    {
      "label": "Benchmark",
      "url": "https://www.honesty.tools/benchmark/"
    },
    {
      "label": "Leaderboard",
      "url": "https://www.honesty.tools/leaderboard/"
    },
    {
      "label": "Research: honesty utilities",
      "url": "https://www.honesty.tools/research#honesty-tools"
    }
  ],
  "claims": null,
  "limitations": [
    "Static authored content from the public HTML. Interactive outputs are not simulated. Source-specific research, product and evidence limits remain in the text; this export is not new verification."
  ],
  "corrections": [],
  "supersedes": [],
  "representations": {
    "html": "https://www.honesty.tools/floor",
    "markdown": "/page-text/floor.md",
    "metadata": "/page-data/floor.json"
  }
}