Governance Decision Record · GDR-004
Documentary correction · no numerical change

C01 United States Country Volume: Unknown-Path Review Status Correction

The published United States volume stated that the unknown-path review remained pending at the time of publication. That statement was incorrect: the review had been completed before publication. No dataset, classifier, denominator or headline figure changes under this record.

RecordGDR-004
Decision typePublished-document provenance correction
Raised28 August 2026
DetectionSelf-detected internally
1 · Decision type

What kind of correction this is

The classification matters, because it bounds what changed.

This is a published-document provenance and methods-description correction.

It is not a methodology transition, a dataset revision, a classifier revision, a numerical correction, or a version event. No dataset, methodology, classifier, row classification or numerical value changes under this record.

C01 measures crawler access only. It does not measure AI visibility, AI ranking, AI citation, AI recommendation, digital authority or commercial performance. The figures reproduced later in this record are shown unchanged for reference; they are not restated as findings here.

2 · The defect

Published wording, and the correct status

Quoted from the published volume, §12 Limitations.

“Unknown-path review status. At the time of publication, the unknown-path review remained pending. Accordingly, the expanded restriction measure should be interpreted as a conservative upper bound…”

The defective clause is the second sentence: “the unknown-path review remained pending.”

Correct status: the US unknown-path review had been completed before publication.

3 · Evidence

How the correct status is established

Three independent retained artefacts, plus row-level confirmation.

The count of 279 is independently confirmed from row-level source: the frozen US dataset c01_us_808_v15.csv contains 279 domains with one or more unknown-path entries.

4 · Disposition

What the review did, and what it changed

The review was substantive; its outcome was that nothing required alteration.

The US unknown-path review inspected the unknown-path log of 279 domains and found no new substantive content category — the unknown paths were overwhelmingly locale-prefixed variants and CMS or platform infrastructure. It changed no row classification, caused no classifier change, retained the existing conservative secondary default, and required no numerical recomputation.

MeasureValue — unchanged by this record
Policy observed624 of 808
Strict restriction24/624
Meaningful restriction67/624
Expanded restriction309/624

No figure was recalculated as a result of the review, and none is recalculated by this record.

5 · Cause

Why the wording was wrong

A publication-generation defect, not a measurement defect.

The four FINAL country manuscripts were generated in a single batch within a four-second window. That batch applied the same “review remained pending” wording to all four volumes.

The wording was correct for Great Britain, Australia and Singapore, whose reviews were genuinely outstanding at publication, and incorrect for the United States, whose review had already been completed. The defect therefore lies in how the document was generated, not in how the data was measured, classified or aggregated.

6 · Not affected

Great Britain, Australia and Singapore

Three volumes are explicitly outside this correction.

The Great Britain, Australia and Singapore volumes also state that their unknown-path reviews remained pending. Those statements accurately describe their publication states and are correct as published.

They must not be altered or harmonised with the United States correction. Their six published artefacts were hashed before any correction work began, so their unchanged state remains independently demonstrable.

7 · Scope of the correction

Explicitly out of scope

Stated so the record cannot be read as more than it is.

The historical published US volume is not replaced or silently edited. It stands as published, with this record carried alongside it. That is the programme’s standing treatment for documentary corrections: the artefact is preserved and the correction is published beside it, rather than the record of what was published being altered after the fact.

Related records

Where this sits

C01 United States Country Volume — the affected publication.

C01 integrity and freeze record — frozen artefact hashes and the reproducibility boundary.

Five-market comparative findings — the current published comparison across all five markets.

GDR-002 Correction 001 · GDR-002 version transition