{
  "taxonomy": {
    "boundary_tiers": {
      "A": {
        "en": "task operation"
      },
      "B": {
        "en": "service maintenance"
      },
      "C": {
        "en": "lifecycle"
      },
      "unknown": {
        "en": "boundary cannot be determined (the source states no accounting scope)"
      }
    },
    "component_status_values": {
      "included": "measured or accounted for in the figure",
      "excluded": "not in the figure at all",
      "neutralized": "appears in the formula but a neutral value was substituted (e.g. PUE fixed at 1.0), so effectively excluded",
      "unknown": "the source does not say",
      "not_applicable": "physically not part of this measurement",
      "not_extracted": "the source may state it, but it has not been pulled yet — distinct from unknown, where the source is silent"
    },
    "components": {
      "gpu_compute": "A",
      "gpu_idle_colocated": "A",
      "host_cpu": "A",
      "dram": "A",
      "local_storage": "A",
      "model_load": "A",
      "network_egress": "B",
      "facility_overhead_pue": "B",
      "cooling_water": "B",
      "idle_fleet_capacity": "B",
      "training_amortized": "C",
      "finetuning_amortized": "C",
      "embodied_hardware": "C",
      "datacenter_construction": "C",
      "client_device": "C"
    },
    "components_note": {
      "en": "cooling_water is used for boundary description only. Water consumption itself is out of scope (dec-F)."
    },
    "knowledge_state_values": {
      "extracted": "value taken from the source",
      "not_extracted": "the source has it; we have not pulled it yet — actionable TODO with a known location",
      "not_reported": "the source does not contain it — do not re-search that source",
      "not_investigated": "nobody has looked yet",
      "investigated_not_found": "we searched and it does not appear to exist publicly — do not repeat the search"
    },
    "reported_by_values": [
      "independent_academic",
      "independent_journalism",
      "vendor_self_reported",
      "regulator_filing",
      "third_party_audit",
      "unknown",
      "industry_survey",
      "independent_analysis"
    ],
    "verification_depth_values": {
      "primary_source_read": "the record's author read the cited source directly",
      "relayed_from_editor_not_independently_reread": "values were relayed by the editorial team and structured without re-reading the source",
      "unverified": "not checked",
      "relayed_from_investigator_not_independently_read": "an investigating AI read the source and reported it; the ledger keeper and the editor did not open it"
    },
    "disclosure_items": {
      "inference_energy_per_request": {
        "definition_required": [
          "per_request_or_per_token",
          "pue_included",
          "model_specified"
        ],
        "en": "Energy per inference request"
      },
      "training_energy_total": {
        "definition_required": [
          "which_model",
          "pretraining_only_or_including_runs"
        ],
        "en": "Total energy used in training"
      },
      "datacenter_pue": {
        "definition_required": [
          "fleet_average_or_site",
          "measurement_period"
        ],
        "en": "Data centre PUE"
      },
      "water_usage": {
        "definition_required": [
          "onsite_or_including_offsite_power"
        ],
        "en": "Water use"
      },
      "grid_mix_or_carbon": {
        "definition_required": [
          "market_based_or_location_based"
        ],
        "en": "Electricity mix or carbon intensity"
      },
      "hardware_embodied": {
        "definition_required": [
          "amortization_period"
        ],
        "en": "Embodied share of the hardware"
      },
      "third_party_assurance": {
        "definition_required": [
          "assurance_level",
          "assurer_name"
        ],
        "en": "Third-party assurance"
      }
    },
    "disclosure_items_note": {
      "en": "water_usage remains a disclosure-survey item. The report does not treat water quantities themselves (dec-F). Whether a company discloses it and whether we cover it are separate questions."
    },
    "staleness_policy": {
      "disclosures_days": 90,
      "measurements": "re-verify if the preprint has a newer version than read_version"
    },
    "units": {
      "Wh": {
        "si_factor_to_wh": 1,
        "en": "watt-hour"
      },
      "kWh": {
        "si_factor_to_wh": 1000,
        "en": "kilowatt-hour"
      },
      "W": {
        "en": "watt (power)"
      },
      "TWh": {
        "si_factor_to_wh": 1000000000000,
        "en": "terawatt-hour"
      },
      "GW": {
        "en": "gigawatt (installed capacity)"
      },
      "percent": {
        "en": "percent"
      },
      "factor": {
        "en": "multiplication factor"
      },
      "gCO2e_per_kWh": {
        "en": "grams of CO2 equivalent per kilowatt-hour"
      },
      "gCO2e": {
        "en": "grams of CO2 equivalent"
      }
    },
    "units_note": {
      "en": "Store the source's own unit. Do not convert. Consumers must read the unit field and never assume Wh."
    },
    "numeric_use_values": {
      "allowed": "may feed derivations",
      "forbidden": "display only; must never feed a calculation or a comparison denominator"
    },
    "date_verification_values": {
      "confirmed_on_page": "the date was read on the page or document itself",
      "access_snapshot_created_by_investigator": "the source is continuously updated and this date identifies the investigator's saved/accessed snapshot, not a publication date",
      "from_search_metadata_not_confirmed_on_page": "taken from search-result metadata; not confirmed in the body",
      "relayed_from_editor_not_confirmed": "supplied by the editorial team; the ledger keeper did not confirm it",
      "not_investigated": "nobody has checked where the date came from",
      "not_confirmed_source_not_opened": "reached only as a citation in another document; the source itself has not been opened",
      "relayed_from_investigator_not_confirmed": "an investigating AI reported the date; neither the ledger keeper nor the editor confirmed it on the page"
    },
    "disclosure_status_values": {
      "disclosed_full": "published with a defined accounting scope",
      "disclosed_partial": "published but incomplete or partially defined",
      "disclosed_but_undefined": "a number exists with no stated accounting scope",
      "not_disclosed_explicitly_declined": "asked or publicly addressed, and declined",
      "not_disclosed_no_statement": "no statement found either way",
      "not_investigated": "nobody has looked yet"
    },
    "surveyed_companies": {
      "criterion": {
        "en": "Provides a general-purpose AI service that ordinary users can use directly. This single criterion only."
      },
      "companies": [
        "google",
        "openai",
        "anthropic",
        "microsoft",
        "meta",
        "amazon",
        "xai"
      ],
      "decided_by": "dec-J"
    },
    "venue_types": {
      "conference": "peer-reviewed academic conference",
      "workshop": "academic workshop",
      "journal": "academic journal",
      "corporate_blog": "a post on the company's own blog",
      "vendor_documentation": "product documentation published by the vendor",
      "trade_press": "industry news outlet",
      "fact_checking_organisation": "independent fact-checking organisation",
      "industry_association": "industry body or trade association",
      "personal_blog": "an individual's personal blog, not an organisation's channel",
      "research_organisation": "independent research organisation, not peer-reviewed",
      "preprint_repository": "repository hosting a manuscript that has not necessarily undergone peer review",
      "consultancy_article": "article published by an environmental or technical consultancy",
      "academic_institution": "university research centre",
      "corporate_report": "a company's own formal report"
    },
    "document_types": {
      "versioned": "has stable versions or a fixed publication date",
      "continuously_updated": "revised in place with no version history; the accessed date is the only anchor",
      "industry_survey": "periodic survey of self-reporting respondents",
      "personal_blog": "post on an individual's blog",
      "vendor_educational_page": "educational material published by the vendor",
      "independent_analysis": "third-party analysis with stated assumptions, not peer reviewed",
      "technical_report": "technical report published by the developer",
      "system_card": "model system card",
      "index_report": "periodic index or scorecard",
      "corporate_sustainability_report": "annual sustainability or environmental report",
      "methodology": "methodology document supporting a report"
    },
    "disclosure_record_status_values": {
      "active": "sourced and usable in the report",
      "draft_pending_source": "content is known but the source record does not exist yet; must not be published",
      "draft_pending_verification": "content is sourced but the primary source has not been read; must not be published",
      "superseded": "historical disclosure record retained under append-only correction; superseded_by points to replacement"
    },
    "disclosure_date_fields": {
      "as_of": {
        "en": "The date the disclosed information itself refers to. This is the column that shows how old each company's newest disclosure is."
      },
      "checked_at": {
        "en": "When the editorial team checked the cell. The 90-day staleness policy is measured against this."
      }
    },
    "reported_by_note": {
      "en": "industry_survey is an industry body's aggregation of self-reported figures: neither vendor self-report nor independent research. The population and response rate drive the result."
    },
    "disclosure_intake_contract": {
      "note": {
        "en": "The minimum a single cell needs to enter the ledger. A batch arriving in this shape will pass validate."
      },
      "required": {
        "id": "disc-<company>-<item>, derived mechanically from the two fields below",
        "company": "one of taxonomy.surveyed_companies.companies",
        "item": "one of taxonomy.disclosure_items",
        "disclosure_status": "one of taxonomy.disclosure_status_values",
        "status": "one of taxonomy.disclosure_record_status_values",
        "publishable": "true or false; true requires at least one source_id",
        "source_ids": "at least one, unless the cell is quarantined (see below)",
        "checked_at": "the date the cell was checked; the 90-day staleness policy runs on this",
        "checked_by": "a participant key",
        "as_of": "the date the disclosed information itself refers to, or {value:null,state:...}",
        "definition_completeness": "an {answer, state} object for every key in this item's definition_required",
        "boundary_claimed": "{tier, components}; tier may be unknown"
      },
      "quarantine_route": {
        "when": "the content is known but its source record does not exist yet",
        "shape": "status=draft_pending_source, source_state=not_extracted, publishable=false",
        "visibility": "validate reports the count of these on every run"
      },
      "do_not": [
        "Do not invent a source record to satisfy source_ids; quarantine the cell instead.",
        "Do not put a third-party figure in notes or description (conventions.free_text_policy).",
        "Do not fill a cell by inference from what similar companies disclose."
      ]
    },
    "measurement_status_values": {
      "active": "verified and usable",
      "draft_pending_verification": "the primary source has not been read; publishable must be false",
      "superseded": "replaced by a later record"
    },
    "correction_root_cause_values": {
      "editorial": "a failure in editorial process, framing, instruction design, or enforcement rather than a single participant identity"
    },
    "negative_finding_rule": {
      "en": "A finding of absence is evidenced by the search, not by a document. A cell with knowledge_state investigated_not_found may have no source_ids, but search_paths_taken becomes mandatory. An unsupported claim of absence does not enter the ledger."
    }
  }
}