{
  "title": "Compost manufacturing data for AI sample and report context",
  "source_page": "https://fiund.com/industries/compost-sample-maturity-report-data",
  "authorship": "Original fictional Fiund editorial exercise. No actual operating records or third-party source assets are included.",
  "example": {
    "title": "A composite taken before the pile split",
    "context": "A fictional site tracks clean paper pile cards. Report stubs give sample identifiers only.",
    "records": [
      [
        "1",
        "Pile history",
        "Pile P7 divided Tuesday into P7-A and P7-B."
      ],
      [
        "2",
        "Sampling ticket",
        "Composite S8 collected Monday across P7."
      ],
      [
        "3",
        "Report index",
        "M8 describes S8; report timestamp Wednesday."
      ]
    ],
    "question": "Can M8 be treated as a separate report for P7-A?",
    "answer": "No. Record 2 places collection on the undivided P7 before the event in record 1. Record 3 connects M8 to that composite, even though the report arrived later. A child-specific observation would require separately scoped sampling evidence; the report date does not establish it.",
    "failureModes": [
      "Treating report date as sample date",
      "Copying a composite result to each later child"
    ],
    "fields": [
      [
        "pile_state_at_collection",
        "Fixes identity at sampling"
      ],
      [
        "sample_coverage",
        "Describes the area represented"
      ],
      [
        "report_reference",
        "Connects the report to its ticket"
      ]
    ],
    "holdout": "Use separately logged pre-split portions with a recorded child map and require exact part lineage.",
    "boundary": "No real compost compositions, environmental conclusions, remediation or agronomic recommendations."
  }
}
