PROVENANCE AUDIT · 29 AUGUST 2026

The catalogue names one council. The source names another.

Across three public records, data.gov.uk displays North Tyneside as publisher while the description and source page point to Leeds, Liverpool or Newcastle. What changed, exactly?

Short answer The pages show a metadata mismatch, not proof that the files were rewritten. The cause and the scale still need checking.

3 page pairs · manual check · official sources

01 / THE FINDING

The answer in 30 seconds

The observation is clear enough for an audit and too narrow for an accusation.

catalogueOne displayed publisher

All three data.gov.uk pages display North Tyneside Metropolitan Borough Council.

sourceThree named organisations

The Data Mill North pages name Leeds City Council, Liverpool City Council and Newcastle City Council.

limitThe cause is not on the pages

We do not know whether this is a harvest issue, a metadata migration or something else. Three cases are not a prevalence estimate.

3records checked
3publisher/source mismatches
0causes established

02 / THE COMPARISON

What the three records show

The rows keep the catalogue statement and the named source statement separate.

Static comparison checked on 29 August 2026. Links open the original pages.
DatasetCatalogue publisherWhat the summary saysPublisher on sourceCatalogue UpdatedSource last update
Blue badge applications
Harvest 20j71
North Tyneside Metropolitan Borough CouncilMonthly totals processed by Leeds City CouncilLeeds City Council15 August 2026about 7 years ago
Prompt payment performance statistics
Harvest e5l3g
North Tyneside Metropolitan Borough CouncilPrompt payment performance for Liverpool City CouncilLiverpool City Council15 August 20269 years ago
Newcastle libraries enquiries
Harvest v8go1
North Tyneside Metropolitan Borough CouncilMonthly branch-level enquiries for Newcastle City CouncilNewcastle City Council15 August 20262 years ago

The catalogue `Updated` value is a metadata label. We do not treat it as proof that the CSV changed on the same day.

03 / REPRODUCTION

How the check worked

A useful audit leaves a short path back to the original material.

  1. 01We fixed the record. For each case we noted the data.gov.uk URL, displayed publisher, `Updated` label, summary and Harvest GUID.
  2. 02We followed the named source. We opened the Data Mill North page with the same identifier and noted its organisation, displayed age and coverage.
  3. 03We separated observation from explanation. We counted only visible mismatches. We did not turn a coincidence into a cause.
  4. 04We kept the evidence. The consulted pages are archived locally, while QA checks hashes and calculations from `data/evidence.json`.

04 / THE BRAKE

What this page does not establish

Rigor is partly about what we refuse to infer.

  • AIt is not proof that the data was rewritten. A new catalogue label can describe metadata, not file contents.
  • BIt is not a directory-wide rate. These three rows are a bounded audit chosen for verifiability.
  • CIt is not an assignment of blame. Public pages do not show whether the mismatch came from harvest, migration or another operational step.
  • DIt contains no contact data. We publish organisations and official links, not email addresses or people.
The question still open

Which job or metadata record attached these three datasets to the displayed publisher? Only logs and administrative history can answer that.

05 / THE USE

Why this matters to public-data users

An unclear publisher changes the trust path around a dataset, even if the file itself is unchanged.

findReach the right steward

The publisher is the first lead for questions, updates and editorial responsibility.

citeKeep catalogue and source apart

An analysis can preserve the index page, the data source and the time each was observed.

maintainReport a verifiable issue

A useful ticket contains the URL, GUID, divergent field and captures from the same check.

06 / THE ACTION

A checklist for the next audit

Use these steps before citing a dataset in a story, analysis or decision.

  1. Save the URL and check time. The catalogue and source can change independently.
  2. Compare publisher and summary. Look for the organisation on the source page, not only in the title.
  3. Record the Harvest identifier. The GUID can help a catalogue team find the technical entry.
  4. Do not name a cause without a log. Ask for metadata history before saying that the file or source changed.

07 / VERIFICATION

Sources and versions

Links point to the public pages consulted. Local captures and hashes record the version used for this build.

data.gov.uk records

The check answers “what is visible now?” and leaves “why?” open. That separation is what keeps the audit useful.