{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-08-28-015",
    "source_story_id": "tmp-sidebar-agentic-math",
    "edition_id": "mp-2026-08-28-morning-0050",
    "edition_url": "https://themachinepress.com/edition/2026-08-28",
    "position": 17,
    "story_type": "ticker",
    "section": "benchmarks-evals",
    "editorial_classification": "editorial",
    "headline": "Equal Answers Hid Different Mathematical Skills",
    "slug": "equal-answers-hid-different-mathematical-skills",
    "dek": "A process benchmark decomposes agentic mathematics into planning, action and feedback capabilities.",
    "summary": "A process benchmark decomposes agentic mathematics into planning, action and feedback capabilities.",
    "body_text": "Models with similar final-answer accuracy showed materially different capability profiles, supporting more diagnostic evaluation than one end score. The trajectories include controlled model rewriting, so the benchmark's fine-grained annotations remain part of what must be audited.",
    "why_it_matters": "A process benchmark decomposes agentic mathematics into planning, action and feedback capabilities.",
    "limitations": [
      "The trajectories include controlled model rewriting, so the benchmark's fine-grained annotations remain part of what must be audited."
    ],
    "importance": 6,
    "canonical_url": "https://themachinepress.com/story/mp-2026-08-28-015/equal-answers-hid-different-mathematical-skills",
    "json_url": "https://themachinepress.com/story/mp-2026-08-28-015.json",
    "first_published_at": "2026-08-28T09:00:00.000-04:00",
    "modified_at": "2026-08-28T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-08-28-015-001",
        "text": "A process benchmark decomposes agentic mathematics into planning, action and feedback capabilities.",
        "source_ids": [
          "source-2026-08-28-017"
        ],
        "qualification": "The trajectories include controlled model rewriting, so the benchmark's fine-grained annotations remain part of what must be audited."
      }
    ],
    "source_ids": [
      "source-2026-08-28-017"
    ],
    "tags": [
      "mathematics",
      "agent evaluation",
      "process metrics"
    ],
    "image_url": null,
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-08-28-017",
      "title": "arXiv preprint 2608.26950",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2608.26950",
      "canonical_url": "https://arxiv.org/abs/2608.26950",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-08-27T06:52:06.000-04:00",
      "accessed_at": "2026-08-28T08:15:00.000-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-08-28-015-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "Equal Answers Hid Different Mathematical Skills",
    "publisher": "The Machine Press",
    "published_at": "2026-08-28T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-08-28-015/equal-answers-hid-different-mathematical-skills"
  }
}