Bayyinah al-Amāna: Provenance Integrity for Public Open Data — Verifying Declared Provenance, Freshness, and Schema Against the Substrate, with a Reproducible Worked Example on New York MTA Datasets

Every civic-accountability analysis built on public open data silently assumes the data is what the portal claims: official rather than derived, current rather than stale, stable rather than silently rewritten. This note applies Structural Honesty Verification to that assumption itself, verifying a dataset's declared provenance, freshness, and schema against its substrate — its provenance classification, row-level creation and update timestamps, and served field inventory. It defines three tiers of what can and cannot be verified and demonstrates the method on two New York State datasets reporting the same MTA procurement records: a community view (gpsc-qqsz) and the first-party official dataset (twsw-2mqa). Reproducible from a single live pull, it establishes their divergence at three resolutions, a wholesale post-close batch load of history in both, and a schema divergence — and refuses to judge whether any was legitimate. It is a meta-layer that makes the data-integrity assumption underlying other accountability work checkable rather than inherited on faith. Converged after three consecutive independent cross-instance audit passes.