{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-10-31-007",
    "source_story_id": "tmp-story-dynarev-goal-revision",
    "edition_id": "mp-2026-10-31-morning-0113",
    "edition_url": "https://themachinepress.com/edition/2026-10-31",
    "position": 7,
    "story_type": "dispatch",
    "section": "benchmarks-evals",
    "editorial_classification": "editorial",
    "headline": "Keep the Browser State When the User Changes the Goal",
    "slug": "keep-the-browser-state-when-the-user-changes-the-goal",
    "dek": "DynaRev injects additive, unconditional and runtime-conditioned revisions at semantic milestones across 181 web-agent tasks.",
    "summary": "DynaRev injects additive, unconditional and runtime-conditioned revisions at semantic milestones across 181 web-agent tasks.",
    "body_text": "Web-agent benchmarks usually freeze the user's request before execution, unlike real interactions. DynaRev delivers revisions at task-defined milestones and scores correctness, responsiveness and efficiency across 181 tasks. Five evaluated models often completed only part of the revised requirements; rewriting the instruction while preserving browser state helped, while resetting the browser lost progress. The benchmark supplies a controlled testbed, not evidence that one revision strategy will generalize across deployed sites.",
    "why_it_matters": "DynaRev injects additive, unconditional and runtime-conditioned revisions at semantic milestones across 181 web-agent tasks.",
    "limitations": [
      "Five evaluated models often completed only part of the revised requirements; rewriting the instruction while preserving browser state helped, while resetting the browser lost progress.",
      "The benchmark supplies a controlled testbed, not evidence that one revision strategy will generalize across deployed sites."
    ],
    "importance": 8,
    "canonical_url": "https://themachinepress.com/story/mp-2026-10-31-007/keep-the-browser-state-when-the-user-changes-the-goal",
    "json_url": "https://themachinepress.com/story/mp-2026-10-31-007.json",
    "first_published_at": "2026-10-31T09:00:00.000-04:00",
    "modified_at": "2026-10-31T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-10-31-007-001",
        "text": "DynaRev injects additive, unconditional and runtime-conditioned revisions at semantic milestones across 181 web-agent tasks.",
        "source_ids": [
          "source-2026-10-31-007"
        ],
        "qualification": "Five evaluated models often completed only part of the revised requirements; rewriting the instruction while preserving browser state helped, while resetting the browser lost progress."
      }
    ],
    "source_ids": [
      "source-2026-10-31-007"
    ],
    "tags": [
      "web agents",
      "dynamic goals",
      "benchmarks",
      "computer use"
    ],
    "image_url": null,
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-10-31-007",
      "title": "arXiv preprint 2610.40133, original v1",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2610.40133v1",
      "canonical_url": "https://arxiv.org/abs/2610.40133v1",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-10-28T20:00:00.000-04:00",
      "accessed_at": "2026-10-31T08:25:01.279-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-10-31-007-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "Keep the Browser State When the User Changes the Goal",
    "publisher": "The Machine Press",
    "published_at": "2026-10-31T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-10-31-007/keep-the-browser-state-when-the-user-changes-the-goal"
  }
}