v12 opened this week as a copy of v11 and immediately absorbed a fix we had been circling for a while. Cloudflare Pages serves every page both with and without its .html extension, and our canonical tags pointed at the extensioned form — a canonical/redirect contradiction the platform’s redirect rules could not cleanly resolve. So the site resolved it instead: all 279 canonical tags, og:url values, structured-data URLs and sitemap entries rewritten to the clean extensionless form.
Machine-readable identity
The first enrichment phase of v12 finished on 24 May: Person schema for the principals plus sameAs entity links — Wikidata and MusicBrainz identifiers — on all 279 pages. Search engines no longer have to guess that our “Paul McCartney” is that Paul McCartney; the pages say so in a form built for machines.
Alongside it, a see-also system built on an actual relatedness graph rather than hand-picked links: 214 songs scored against each other, top-30 edges kept per song — 6,420 edges cached as a build artifact that future passes can reuse.
Citations: destination, not funnel
The week’s most consequential decision was editorial. Our citation policy is now locked as a hybrid: inline plain-text attribution to primary sources — Lewisohn, the EMI/Apple archival record, the Anthology materials — backed by an internal bibliography page, with zero outbound links to competing Beatles reference sites. Authority for a reference work comes from primary-source attribution and entity linking, not from link exchanges. This site competes for the same searches as the established fan references; it should be a destination, not a referral funnel.
Also this week: ship cadence settled (v12 ships as a batch once its enrichment phases are complete), and an AdSense application drafted — and deliberately held until the site is at its best. Behind both decisions sits the same instinct: do the work first, then present it.