{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-09-14-006",
    "source_story_id": "tmp-story-deictic-shift-verification-pipelines",
    "edition_id": "mp-2026-09-14-morning-0067",
    "edition_url": "https://themachinepress.com/edition/2026-09-14",
    "position": 6,
    "story_type": "dispatch",
    "section": "developer-tools",
    "editorial_classification": "editorial",
    "headline": "The Verifier Changed What ‘Previous’ Meant",
    "slug": "the-verifier-changed-what-previous-meant",
    "dek": "In a synthetic draft–verify–revise test, balanced accuracy ranged from 0.156 to near-perfect as models and reasoning settings changed.",
    "summary": "In a synthetic draft–verify–revise test, balanced accuracy ranged from 0.156 to near-perfect as models and reasoning settings changed.",
    "body_text": "Draft–verify–revise pipelines pass language from one model role to another, creating an opening for context-dependent words such as ‘previous’ to change referents. A study built ten base examples in three controlled conditions and tested six models across 21 reasoning configurations. Balanced accuracy ranged from 0.156 to near-perfect. GPT-5.2 rose from 0.156 without reasoning to 0.942 at its highest tested effort, while Gemini 3 Pro stayed above 0.94 across settings and achieved its low-effort result at roughly 5% of the reported per-trial cost of GPT-5.2 at xhigh. The narrow synthetic design supports an engineering warning, not a universal model ranking.",
    "why_it_matters": "In a synthetic draft–verify–revise test, balanced accuracy ranged from 0.156 to near-perfect as models and reasoning settings changed.",
    "limitations": [
      "The narrow synthetic design supports an engineering warning, not a universal model ranking."
    ],
    "importance": 8,
    "canonical_url": "https://themachinepress.com/story/mp-2026-09-14-006/the-verifier-changed-what-previous-meant",
    "json_url": "https://themachinepress.com/story/mp-2026-09-14-006.json",
    "first_published_at": "2026-09-14T09:00:00.000-04:00",
    "modified_at": "2026-09-14T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-09-14-006-001",
        "text": "In a synthetic draft–verify–revise test, balanced accuracy ranged from 0.156 to near-perfect as models and reasoning settings changed.",
        "source_ids": [
          "source-2026-09-14-006"
        ],
        "qualification": "The narrow synthetic design supports an engineering warning, not a universal model ranking."
      }
    ],
    "source_ids": [
      "source-2026-09-14-006"
    ],
    "tags": [
      "AI agents",
      "verification",
      "context engineering"
    ],
    "image_url": null,
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-09-14-006",
      "title": "arXiv preprint 2609.12162",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2609.12162",
      "canonical_url": "https://arxiv.org/abs/2609.12162",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": null,
      "accessed_at": "2026-09-14T08:24:00.000-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-09-14-006-001"
      ]
    }
  ],
  "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 Verifier Changed What ‘Previous’ Meant",
    "publisher": "The Machine Press",
    "published_at": "2026-09-14T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-09-14-006/the-verifier-changed-what-previous-meant"
  }
}