Skip to content

Changelog

Substantive changes to the analyzer are recorded here, newest first.

  1. Change

    Playlist analyses are shared between visitors for up to 6 hours

    • The analyzer now keeps the public figures for a playlist on our server for up to 6 hours and reuses them for the next visitor who analyzes the same playlist. Popular playlists were being fetched from YouTube again for every visitor, which used up YouTube's daily quota and took the analyzer offline.
    • View and like counts in a report can therefore be up to 6 hours old. Reports that use your own API key, or a video range, skip the cache. The cache is keyed by playlist ID and holds no personal data; the privacy policy, terms, methodology and the homepage FAQ (in every language) now say so.
    • Also fixed the quota error message, which showed Google's raw text, and stopped retrying requests that cannot succeed.
  2. Change

    Reference pages now refresh less often, to protect the YouTube quota

    • Course figures are now kept for up to 14 days (was 7) and Discover figures for up to 3 days (was 1) before they are read from YouTube again. They still stay well inside YouTube's 30-day limit for stored public data, and each page shows the date its figures were read.
    • Builds that are not production no longer call YouTube for these pages, so test builds cannot use up the daily quota the live site depends on. The terms, privacy policy and methodology pages were updated to match.
  3. Fix

    Corrected our statement about caching YouTube data

    • The terms said we do not store or cache YouTube content. That stopped being accurate when the course and Discover pages began keeping short-lived summaries of public playlist figures. The terms and privacy policy now say exactly what is kept: up to 7 days for course pages and up to 1 day for Discover pages, then refreshed from YouTube.
    • Both pages now link the YouTube Terms of Service and the Google Privacy Policy, and state that the site uses YouTube API Services with public data and an API key only.
  4. Change

    Added Discover: popular music playlists with live stats

    • New Discover section with nine hand-picked music chart and genre playlists. Each page shows the playlist's real length, genre mix, top artists, least-watched tracks, upload era and days to listen through, read from YouTube every day.
    • No song is named by hand, because chart playlists are rebuilt every week. Each page shows the playlist's title and channel exactly as YouTube reports them.
    • Added a Genre mix, score and persona section to the methodology page, and expanded its definitions of every figure.
  5. Change

    Real results on the homepage and tool pages, hub pages, consent choice

    • The homepage and the sorter, exporter and comparison pages now show a real result for a public playlist, read from YouTube, so the tools can be understood before anything is pasted.
    • Expanded the sorter, shuffle, exporter, comparison, summary card and Rate My Playlist pages with how each tool works, its limits and, where it applies, how the score and personas are calculated.
    • Course length pages now show how a course is paced, from the real spread of video lengths.
    • Added /tools and /guides index pages, which previously returned a not-found error.
    • The cookie banner now controls Google tags: Google storage is denied by default and only allowed after you choose Accept All. Declining leaves it denied.
    • Removed a request for an author photo that did not exist.
  6. Fix

    Fixed video range selection and stopped logging secrets

    • Fixed a bug where requesting any video range beyond the first 50 returned an empty result. The fetch limit was computed as the range's length rather than its end position, so items before the start were never paged past.
    • Range-limited analyses now recompute every statistic — longest video, upload timeline, totals — against the selected range. Previously only four figures were updated and the rest still described the whole fetched prefix.
    • Stopped writing the YouTube API request URL to the server log. That URL contains the API key as a query parameter, including keys supplied by users.
    • Stopped logging submitted playlist URLs and full API responses.
  7. Change

    Corrected indexing of untranslated pages

    • Only the homepage is genuinely translated, but every page was being published at ten locale URLs and marked indexable, producing roughly 405 near-duplicate English URLs.
    • Pages are now indexed in non-English locales only where the content is actually translated, and the sitemap lists only indexable URLs.
    • Sitemap entries now carry real content-change dates instead of the build timestamp, which had made every page appear freshly updated on every deploy.