{
  "title": "Sanitary paper data for AI ply and package records",
  "source_page": "https://fiund.com/industries/tissue-ply-count-pack-record-data",
  "authorship": "Original fictional Fiund editorial exercise. No actual operating records or third-party source assets are included.",
  "example": {
    "title": "Ply count is not sheet count",
    "context": "Inert paper and an original carton mockup provide visual and ledger evidence for a fictional pack.",
    "records": [
      [
        "T1",
        "Pack print",
        "Pack K reads 20 sheets, 2-ply."
      ],
      [
        "T2",
        "Conversion tally",
        "Pack K received 20 finished sheets."
      ],
      [
        "T3",
        "Peel photo",
        "One sampled sheet separated into three visibly distinct layers."
      ],
      [
        "T4",
        "Sampling log",
        "Only one of the 20 sheets was peeled."
      ]
    ],
    "question": "Which claim is contradicted and at what scope?",
    "answer": "T3 contradicts the two-ply claim on T1 for the one sampled sheet. T2 independently supports the 20-sheet count. T4 prevents asserting all 20 sheets have three plies; preserve the observed sample scope and keep the two count dimensions separate.",
    "failureModes": [
      "Counting each separated layer as a finished sheet",
      "Generalizing from one sampled sheet to the full pack"
    ],
    "fields": [
      [
        "finished_sheet_count",
        "Tracks sale units apart from constituent layers"
      ],
      [
        "observed_ply_count",
        "Records layers on the tested sheet"
      ],
      [
        "sample_n",
        "Limits the scope of the observation"
      ]
    ],
    "holdout": "Use one sheet with a partially torn third ply; label the visible section and uncertainty before generalizing across the sheet.",
    "boundary": "Use owned mock packages and no hygiene, consumer labeling or commercial quality claims."
  }
}