{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-08-25-010",
    "source_story_id": "tmp-story-executable-rubrics",
    "edition_id": "mp-2026-08-25-morning-0047",
    "edition_url": "https://themachinepress.com/edition/2026-08-25",
    "position": 10,
    "story_type": "dispatch",
    "section": "benchmarks-evals",
    "editorial_classification": "editorial",
    "headline": "The Rubric Became a Program",
    "slug": "the-rubric-became-a-program",
    "dek": "ExecRubrics compiles evaluation logic into inspectable scoring functions instead of leaving every criterion to a black-box judge.",
    "summary": "ExecRubrics compiles evaluation logic into inspectable scoring functions instead of leaving every criterion to a black-box judge.",
    "body_text": "Across HealthBench, HelpSteer and ArgQuality, executable rubrics matched or improved natural-language rubric baselines at best preference accuracies of 53, 78 and 92 percent while cutting latency by as much as 320 times. The approach makes dependencies, penalties and override conditions explicit and editable.",
    "why_it_matters": "ExecRubrics compiles evaluation logic into inspectable scoring functions instead of leaving every criterion to a black-box judge.",
    "limitations": [],
    "importance": 8,
    "canonical_url": "https://themachinepress.com/story/mp-2026-08-25-010/the-rubric-became-a-program",
    "json_url": "https://themachinepress.com/story/mp-2026-08-25-010.json",
    "first_published_at": "2026-08-25T09:00:00.000-04:00",
    "modified_at": "2026-08-25T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-08-25-010-001",
        "text": "ExecRubrics compiles evaluation logic into inspectable scoring functions instead of leaving every criterion to a black-box judge.",
        "source_ids": [
          "source-2026-08-25-010"
        ],
        "qualification": null
      }
    ],
    "source_ids": [
      "source-2026-08-25-010"
    ],
    "tags": [
      "evaluation",
      "rubrics",
      "verifiable scoring"
    ],
    "image_url": null,
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-08-25-010",
      "title": "arXiv preprint 2608.22559",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2608.22559",
      "canonical_url": "https://arxiv.org/abs/2608.22559",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-08-23T15:11:59.000-04:00",
      "accessed_at": "2026-08-25T08:17:00.000-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-08-25-010-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "The Rubric Became a Program",
    "publisher": "The Machine Press",
    "published_at": "2026-08-25T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-08-25-010/the-rubric-became-a-program"
  }
}