{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-11-07-012",
    "source_story_id": "tmp-story-blindfolded-controller",
    "edition_id": "mp-2026-11-07-morning-0120",
    "edition_url": "https://themachinepress.com/edition/2026-11-07",
    "position": 12,
    "story_type": "dispatch",
    "section": "research",
    "editorial_classification": "editorial",
    "headline": "A Decision Model Gets Reinforcement Learning to the First Reward",
    "slug": "a-decision-model-gets-reinforcement-learning-to-the-first-reward",
    "dek": "DM-guided training initializes sparse-reward policies, then lets environment feedback improve beyond the initial controller.",
    "summary": "DM-guided training initializes sparse-reward policies, then lets environment feedback improve beyond the initial controller.",
    "body_text": "Decision Model-Guided Training gives a model a task and environment description, uses its behavior to initialize a reinforcement-learning policy and then continues learning from reward. Across seven reported tasks, the hybrid beat reinforcement learning from scratch in all seven and the decision model alone in five. The result supports initialization as an exploration shortcut; it does not establish a general advantage outside those tasks or descriptions.",
    "why_it_matters": "DM-guided training initializes sparse-reward policies, then lets environment feedback improve beyond the initial controller.",
    "limitations": [
      "The result supports initialization as an exploration shortcut; it does not establish a general advantage outside those tasks or descriptions."
    ],
    "importance": 8,
    "canonical_url": "https://themachinepress.com/story/mp-2026-11-07-012/a-decision-model-gets-reinforcement-learning-to-the-first-reward",
    "json_url": "https://themachinepress.com/story/mp-2026-11-07-012.json",
    "first_published_at": "2026-11-07T09:00:00.000-05:00",
    "modified_at": "2026-11-07T09:00:00.000-05:00",
    "content_status": "updated",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-11-07-012-001",
        "text": "DM-guided training initializes sparse-reward policies, then lets environment feedback improve beyond the initial controller.",
        "source_ids": [
          "source-2026-11-07-012"
        ],
        "qualification": "The result supports initialization as an exploration shortcut; it does not establish a general advantage outside those tasks or descriptions."
      }
    ],
    "source_ids": [
      "source-2026-11-07-012"
    ],
    "tags": [
      "reinforcement learning",
      "decision models",
      "sparse rewards"
    ],
    "image_url": null,
    "corrections": [
      {
        "correction_id": "correction-2026-11-07-001",
        "revision": 2,
        "summary": "Publication note: This November 7 edition was assembled before its scheduled 9:00 AM America/Detroit reveal but is being published late after recovery from a local build interruption. Its original reporting, image-generation times and scheduled publishAt are preserved.",
        "published_at": "2026-11-07T09:00:00.000-05:00",
        "affected_story_ids": [],
        "affected_claim_ids": []
      }
    ]
  },
  "sources": [
    {
      "source_id": "source-2026-11-07-012",
      "title": "arXiv preprint 2611.07934, original v1",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2611.07934v1",
      "canonical_url": "https://arxiv.org/abs/2611.07934v1",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-11-05T07:38:10.000-05:00",
      "accessed_at": "2026-11-07T08:24:00.000-05:00",
      "supports_claim_ids": [
        "claim-mp-2026-11-07-012-001"
      ]
    }
  ],
  "corrections": [
    {
      "correction_id": "correction-2026-11-07-001",
      "revision": 2,
      "summary": "Publication note: This November 7 edition was assembled before its scheduled 9:00 AM America/Detroit reveal but is being published late after recovery from a local build interruption. Its original reporting, image-generation times and scheduled publishAt are preserved.",
      "published_at": "2026-11-07T09:00:00.000-05:00",
      "affected_story_ids": [],
      "affected_claim_ids": []
    }
  ],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "A Decision Model Gets Reinforcement Learning to the First Reward",
    "publisher": "The Machine Press",
    "published_at": "2026-11-07T09:00:00.000-05:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-11-07-012/a-decision-model-gets-reinforcement-learning-to-the-first-reward"
  }
}