[
  {
    "section": "s1",
    "kind": "author_analysis_or_verification_proposal",
    "sources": [],
    "claim": "1. Predictive coordinates need not be causal variables"
  },
  {
    "section": "s2",
    "kind": "source_based_review",
    "sources": [
      "dl3-c05-1",
      "dl3-c05-2",
      "dl3-c05-3",
      "dl3-c05-4",
      "dl3-c05-5",
      "dl3-c05-6"
    ],
    "claim": "2. What the primary records establish"
  },
  {
    "section": "s3",
    "kind": "author_analysis_or_verification_proposal",
    "sources": [],
    "claim": "3. Identifiability is always relative to a model class"
  },
  {
    "section": "s4",
    "kind": "author_analysis_or_verification_proposal",
    "sources": [],
    "claim": "4. Observational rotation gives a transparent ambiguity"
  },
  {
    "section": "s5",
    "kind": "author_analysis_or_verification_proposal",
    "sources": [],
    "claim": "5. A labeled mean shift adds directional information"
  },
  {
    "section": "s6",
    "kind": "author_analysis_or_verification_proposal",
    "sources": [],
    "claim": "6. Intervention diversity determines the visible subspace"
  },
  {
    "section": "s7",
    "kind": "author_analysis_or_verification_proposal",
    "sources": [],
    "claim": "7. An intervention changes a mechanism, not merely a label"
  },
  {
    "section": "s8",
    "kind": "author_analysis_or_verification_proposal",
    "sources": [],
    "claim": "8. Nonlinear reparameterization changes intervention semantics"
  },
  {
    "section": "s9",
    "kind": "author_analysis_or_verification_proposal",
    "sources": [],
    "claim": "9. Temporal order provides structure but not automatic causality"
  },
  {
    "section": "s10",
    "kind": "author_analysis_or_verification_proposal",
    "sources": [],
    "claim": "10. Auxiliary variables can carry shortcuts"
  },
  {
    "section": "s11",
    "kind": "author_analysis_or_verification_proposal",
    "sources": [],
    "claim": "11. Metrics may hide permitted transformations"
  },
  {
    "section": "s12",
    "kind": "author_analysis_or_verification_proposal",
    "sources": [],
    "claim": "12. Counterfactual claims require additional commitments"
  },
  {
    "section": "s13",
    "kind": "author_analysis_or_verification_proposal",
    "sources": [],
    "claim": "13. A proposed study separates identification from optimization"
  },
  {
    "section": "s14",
    "kind": "author_analysis_or_verification_proposal",
    "sources": [],
    "claim": "14. Intervention effects should be evaluated on observables too"
  },
  {
    "section": "s15",
    "kind": "author_analysis_or_verification_proposal",
    "sources": [],
    "claim": "15. A causal variable is an experimentally grounded abstraction"
  }
]