{
  "schemaVersion": 1,
  "policyId": "mde-content-consolidation-v1",
  "dateModified": "2026-08-05",
  "description": "Deterministic consolidation analysis for public articles, concepts, questions and topic hubs. Exact duplicates may produce redirect recommendations; near-duplicates are review-only and no source is deleted automatically.",
  "eligibleRouteClasses": [
    "article",
    "concept",
    "question",
    "topic-hub"
  ],
  "eligiblePrefixes": [
    "/research/blog/",
    "/concepts/",
    "/questions/",
    "/topics/",
    "/research/topics/"
  ],
  "excludedExactRoutes": [
    "/research/blog/convergence-17-openai-broadcom-jalapeno-inference-chip.html",
    "/research/blog/policy-jalapeno-inference-chip.html",
    "/research/blog/template.html"
  ],
  "exactDuplicatePolicy": {
    "minimumBodyWords": 45,
    "automaticAction": "redirect-recommendation-only",
    "preferredCanonicalOrder": [
      "non-template",
      "self-canonical",
      "shorter-route",
      "lexicographic-route"
    ]
  },
  "nearDuplicatePolicy": {
    "minimumBodyWords": 120,
    "jaccardThreshold": 0.82,
    "automaticAction": "none-review-required"
  },
  "topics": [
    {
      "id": "ai-alignment",
      "label": "AI alignment",
      "keywords": [
        "alignment",
        "safety",
        "stewardship",
        "meltdown",
        "constitutional",
        "eden protocol"
      ]
    },
    {
      "id": "recursive-intelligence",
      "label": "Recursive intelligence",
      "keywords": [
        "recursive",
        "recursion",
        "arc principle",
        "inference-time",
        "scaling",
        "self-improvement"
      ]
    },
    {
      "id": "ethics-and-values",
      "label": "Ethics and values",
      "keywords": [
        "ethics",
        "ethical",
        "values",
        "moral",
        "care",
        "love",
        "religious"
      ]
    },
    {
      "id": "evidence-and-provenance",
      "label": "Evidence and provenance",
      "keywords": [
        "evidence",
        "priority",
        "provenance",
        "verification",
        "citation",
        "hash",
        "dkim",
        "falsification"
      ]
    },
    {
      "id": "book-and-concepts",
      "label": "Book and concepts",
      "keywords": [
        "infinite architects",
        "book",
        "chapter",
        "concept",
        "hrih",
        "eden vs babylon"
      ]
    },
    {
      "id": "external-developments",
      "label": "External developments",
      "keywords": [
        "convergence",
        "anthropic",
        "openai",
        "google",
        "meta",
        "nvidia",
        "government",
        "vatican"
      ]
    },
    {
      "id": "engineering-and-operations",
      "label": "Engineering and operations",
      "keywords": [
        "engineering",
        "architecture",
        "pipeline",
        "netlify",
        "worker",
        "control",
        "promotion engine"
      ]
    },
    {
      "id": "funding-and-strategy",
      "label": "Funding and strategy",
      "keywords": [
        "grant",
        "funding",
        "funders",
        "aria",
        "portfolio",
        "investment"
      ]
    },
    {
      "id": "neurodiversity-and-research",
      "label": "Neurodiversity and research",
      "keywords": [
        "adhd",
        "autism",
        "neurodivergent",
        "cognition",
        "creative breakthrough"
      ]
    }
  ],
  "outputs": {
    "clusters": "research/data/content-clusters.json",
    "topics": "research/data/topic-index.json",
    "consolidation": "research/data/content-consolidation.json",
    "redirectRecommendations": "research/data/content-redirect-recommendations.json"
  },
  "excludedExactRoutesReasons": {
    "/research/blog/convergence-17-openai-broadcom-jalapeno-inference-chip.html": "Its visible title contains one of the aiDiscovery.forbiddenTerms hardware words, so publishing an analysis record that quotes the title trips the patent guard. The page is commentary on a public third-party chip announcement, not programme hardware disclosure, but the guard is deliberately blunt and stays that way, so the route is excluded here instead. The term is not repeated in this file, for the same reason.",
    "/research/blog/policy-jalapeno-inference-chip.html": "Its visible title or description contains one of the aiDiscovery.forbiddenTerms hardware words, so publishing an analysis record that quotes it trips the patent guard. These pages are commentary on a public third-party chip announcement, not programme hardware disclosure, but the guard is deliberately blunt and stays that way, so the route is excluded here instead. The term is not repeated in this file, for the same reason."
  }
}