Citations got their final form this week. Inline author-plus-page scaffolding — accurate, but a speed bump in every sentence — is being stripped from verified prose and replaced with book-level superscript markers (1 = Lewisohn, 2 = Ryan & Kehew, 3+ = others in order of appearance) resolving to a clean numbered Sources list, mirrored in the page’s structured data. The conversion is deliberately not automated into a bulk pass: removing a citation lead-in leaves grammatical debris, and quoted text must never be touched, so the work rides the careful per-page remediation lane where a person reads every sentence.
Your verification can lie to you
The week’s hard lesson: two earlier corpus scans had reported every page structurally clean, and both were wrong. A stricter scan — read the raw bytes, strip trailing padding, assert the file actually ends with </html> — found 16 enriched pages whose closing tags had been silently clipped by an editing-tool artifact, each missing only a handful of bytes. Tail-sampling and pattern-matching checks passed them because the breakage was smaller than the checks’ resolution. All 16 repaired deterministically the same day, and the byte-level scan is now the standing integrity gate at every ship. If your verification has never caught your tools lying, it isn’t verification yet.
Presentation systems
A cluster of visible changes shipped alongside the remediation sessions (three to four pages restructured per session all week):
- The brand wordmark aligned to the domain — BeatlesAnswers.org — site-wide.
- The homepage’s song-of-the-day box gave way to a live enrichment-phase radial chart: the project’s own progress, published as a dashboard and refreshed at every session close.
- The top navigation collapsed behind a hamburger on mobile — site-wide markup and CSS, no scripts.
- The CSS skin standard was codified: all styling lives in stylesheets, colours are skin tokens, and no page carries inline style pins. Swap one skin file and the whole site re-skins. The hub pages are already inline-free; content pages shed their legacy inline styles (about 1,400 of them) as each passes through remediation.
None of this is content. All of it is the difference between a pile of pages and a publication.