Our archive and filter snapshot

How we measure

Our automated pipeline checks selected feeds. This is not every UK source or a representative sample of UK journalism. The logs record selected rejection stages; they do not contain every source item, every decision or every publication route.

The archive count comes from article Markdown files. Daily counts use the date in each filename; the total counts files present when the snapshot was generated. This includes older archive pages and pages marked noindex, as well as deliberately indexable reporting. It is not a count of newly reviewed or Google-indexed stories.

Filter counts are raw rejection-log rows, not deduplicated articles. Repeated runs or repeated titles can create more than one row, and some rows lack a title. Missing logs do not imply that no filtering took place. Automated keyword rules assign the displayed theme labels from recorded text; labels can be incomplete or mistaken.

Direct editorial submissions follow a separate review process. Their archive files can appear in these totals without belonging to the automated feed-filter cohort. Adding archive files to rejection-log rows does not establish how many unique candidates were considered, or the proportion that passed a shared process.

Snapshot: 2026-10-18 10:08 UTC. Later articles and log rows are included only after regeneration. Comparisons can change with feed selection, repeated runs and recording coverage; this is a dated operational snapshot, not a live national index.

The JSON download retains legacy fields for compatibility: published means archive-file counts, filtered means recorded rejection rows, and considered adds those two counts. Its ratio fields are not a valid shared-cohort approval rate or a measure of UK news sentiment.