Seattle R.A.R.E. — the Records Accountability and Review Engine — is a read-only review of public records. Every figure on this site traces to the master source dataset. This page documents the source, the scope, the definitions, and the discipline that keeps flags separate from findings.
The review is built from the City of Seattle SDCI complaint and enforcement records published on the Open Data Portal, covering 2003 through December 31, 2025. The dataset is treated as the authoritative boundary: every count, rate, and chart on this site is computed from it directly and can be reproduced from the same source. Nothing is estimated, projected, or supplemented with outside figures.
The master dataset mixes land use, construction, vacant building, weeds, and many other record types alongside landlord–tenant matters. The landlord–tenant (LLT) view isolates records that carry housing/tenancy categories and the keyword signatures of habitability, tenant rights, heat, plumbing, infestation, and related conditions. All-record figures and LLT-filtered figures are labeled distinctly throughout — they are never silently mixed.
The most important discipline on this site is the separation between a recorded status and a conclusion about what happened. The definitions below govern every page.
Record-type flags in the source data. They mark how a record is classified, not independent enforcement events.
A data-visibility flag. It means the record carries no inspection date or result — not proof that no inspection ever occurred.
A status flag indicating a Notice of Violation record. It does not by itself prove enforcement liability or a final legal outcome.
Rows without a resolved council district are shown separately, never merged into a district to inflate or deflate its totals.
Where a metric's measurement method has not yet been validated to publication standard, the number is withheld rather than published provisionally — even when a draft figure exists. Case studies report what captured public evidence shows and what it does not show; they do not assert legal liability, agency wrongdoing, or individual fault, and they do not publish private correspondence or personal contact information. When two reporting streams could be confused, they are kept separate unless a source explicitly links them.
The dashboard lets anyone filter the same underlying counts by district and year and watch the figures recompute. The district dossiers quote complaint text verbatim. The case study evidence index lists each source document by ID, year, and role. The intent is that a journalist, a tenant, or a court can follow any claim on this site back to the record it came from.