{
  "_comment": [
    "The method state register (delivery plan, phase 3; audit item: designed vs",
    "operating).",
    "",
    "The methodology page describes how CRMAA measures, verifies and certifies.",
    "Almost all of it is true as a description of the method. Almost none of it",
    "was true as a description of what is currently happening, because the page",
    "was written in the present tense throughout: CRMAA 'collects' responses and",
    "'ingests' operational signals and 'publishes' annual editions, when no",
    "instrument has been fielded, nothing has been ingested, and no edition has",
    "been published.",
    "",
    "That is the audit's central complaint in one page. The fix is not to delete",
    "the method - a described method is exactly what an institution like this",
    "should publish - but to say, next to each claim, which of four things it",
    "is. A reader can then tell a safeguard that runs from a safeguard that is",
    "written down, which is the distinction the present tense erases.",
    "",
    "Each claim here is matched to an element on methodology.html by its id.",
    "check-method-state.js fails if a claim is declared and not shown, shown and",
    "not declared, or claims a state its cited evidence does not support: an",
    "'operating' or 'implemented' claim must name a file in this repository that",
    "exists, and a 'designed' or 'planned' claim must say what is missing."
  ],
  "version": 1,
  "updated": "2026-11-14",
  "states": {
    "operating": {
      "label": "Operating",
      "meaning": "Running now. Something real passes through it, and the evidence names where."
    },
    "implemented": {
      "label": "Built, not yet in use",
      "meaning": "The mechanism is written and tested in this repository. Nothing has gone through it yet, because nothing has been fielded or certified."
    },
    "designed": {
      "label": "Designed",
      "meaning": "The rule is decided and written down. It is not yet built, and nothing enforces it automatically."
    },
    "planned": {
      "label": "Planned",
      "meaning": "Intended, and not yet decided in enough detail to build. Stated so the reader does not read it as current practice."
    }
  },
  "reader_note": "Nothing CRMAA publishes has been fielded. Where a claim below is marked as operating, what operates is the mechanism, not a stream of real measurement.",
  "claims": [
    {
      "id": "instrument-families",
      "claim": "CRMAA operates two families of instruments: accreditation standards and research indices.",
      "state": "implemented",
      "evidence": [
        "GCRMES_Portal/assets/js/gcrmes-data.js",
        "data/crmaa-methodology.json"
      ],
      "note": "All four standards and all index engines exist and score. No organisation has been certified and no index has been fielded."
    },
    {
      "id": "evidence-adjusted-scoring",
      "claim": "A requirement's credit is adjusted by how widely it is deployed and how confident the evidence is, and gated by whether it produces an outcome.",
      "state": "implemented",
      "evidence": [
        "GCRMES_Portal/assets/js/gcrmes-engine.js"
      ],
      "note": "The adjustment is in the scoring engine and reproduces against the published test vectors. It has only ever been applied to self-assessment input, never to assessor-verified evidence."
    },
    {
      "id": "mandatory-gates",
      "claim": "Mandatory requirements and per-domain floors can hold back a programme that scores well overall.",
      "state": "implemented",
      "evidence": [
        "GCRMES_Portal/assets/js/gcrmes-data.js",
        "PXE_Portal/assets/js/pxe-engine.js"
      ],
      "note": "The gates are declared per standard and enforced by the engines."
    },
    {
      "id": "survey-integrity",
      "claim": "Responses are collected through a shared survey engine with consistent instruments, so figures are comparable tool to tool and edition to edition.",
      "state": "implemented",
      "evidence": [
        "Survey_Studio/index.html"
      ],
      "note": "No responses have been collected through it, so there is nothing yet to compare and no second edition to compare it against.",
      "supersedes": "The page previously stated this in the present tense, which read as a description of fieldwork in progress."
    },
    {
      "id": "evidence-library",
      "claim": "Supporting artefacts are uploaded once, verified, and reused across assessments with a full chain of custody.",
      "state": "implemented",
      "evidence": [
        "backend/app/evidence.py",
        "backend/app/routers/evidence.py"
      ],
      "note": "The library, the verification states and the custody chain are built and covered by the API test suite. No assessment has used it, because none has run."
    },
    {
      "id": "coverage-gating",
      "claim": "Operational signals feed the shared indicators only once they meet a minimum sample-coverage threshold.",
      "state": "implemented",
      "evidence": [
        "backend/app/models.py"
      ],
      "note": "The threshold and the representativeness flag exist in the data model. CRMAA is not currently ingesting any operational signal, so nothing has been gated by it.",
      "supersedes": "The page previously said CRMAA ingests live operational signals. It does not."
    },
    {
      "id": "human-decision",
      "claim": "An accreditation decision is made by an independent human auditor, not by a readiness score.",
      "state": "designed",
      "what_is_missing": "No auditor has been appointed and no decision has been made. The rule is published; the independence it describes has not yet been tested by a real decision.",
      "note": "The applicant-facing software can indicate readiness and is explicitly barred from deciding."
    },
    {
      "id": "right-of-response",
      "claim": "Applicants can see the auditor's findings and respond before a decision is finalised, and may appeal one.",
      "state": "implemented",
      "evidence": [
        "backend/app/routers/accreditation.py"
      ],
      "note": "The findings, response and appeal routes exist and are tested. No applicant has used them."
    },
    {
      "id": "separation-of-duties",
      "claim": "Preparation and self-assessment are kept distinct from independent review and the certification decision.",
      "state": "designed",
      "what_is_missing": "The separation is a published rule, not an enforced one: there is no second organisation, and no roster of assessors independent of CRMAA."
    },
    {
      "id": "conflict-of-interest",
      "claim": "Consulting an organisation toward readiness and certifying it are treated as distinct engagements.",
      "state": "designed",
      "what_is_missing": "CRMAA sells consulting and awards accreditation, and mitigates that by disclosure rather than by separation. Every such relationship is disclosed permanently on the register and the certificate, and no individual who delivered or sold the consulting takes part in the assessment. A buyer may still reasonably weigh that.",
      "note": "Stated here rather than described as independence, for the same reason it is stated on the transparency page."
    },
    {
      "id": "public-register",
      "claim": "Accredited organisations appear in a public register, so any certification claim can be checked at source.",
      "state": "operating",
      "evidence": [
        "accredited-directory.html",
        "backend/app/routers/accreditation.py"
      ],
      "note": "The register is live and queryable, and currently empty, because nothing has been certified. An empty register that anyone can check is the honest state of a body that has certified nobody."
    },
    {
      "id": "versioned-standards",
      "claim": "Each standard publishes its dimensions, requirements, levels and tiers, so what is measured is open to scrutiny.",
      "state": "operating",
      "evidence": [
        "methodology-instrument.html",
        "data/crmaa-methodology.json",
        "data/crmaa-formula-vectors.json"
      ],
      "note": "Generated from the code that computes each score, and published with test vectors an outsider can reproduce."
    },
    {
      "id": "annual-editions",
      "claim": "Indices are published as dated editions, with a live feed between editions.",
      "state": "planned",
      "what_is_missing": "No edition has been published and no feed is running. Fieldwork, consent, a frozen edition and a correction route all come before the first one.",
      "supersedes": "The page previously described editions and live feeds as current practice."
    },
    {
      "id": "comparable-method",
      "claim": "Instruments keep a consistent structure across editions, so movement reflects real change rather than a changed ruler.",
      "state": "planned",
      "what_is_missing": "There is one edition of each instrument and no movement to measure. What exists today is the mechanism that would make movement meaningful: frozen instrument versions, published weights and test vectors that fail the build if a weight moves without being declared."
    },
    {
      "id": "privacy-practice",
      "claim": "Data minimisation, essential browser storage only, and member control over their own records.",
      "state": "operating",
      "evidence": [
        "privacy.html",
        "data/crmaa-data-handling.json"
      ],
      "note": "Declared per product, checked on every build by check-data-handling.js, and the deletion route is real."
    }
  ]
}
