Browser extension: detect and subscribe to standard.site publications on ATProto substandard.blog

fix(panoptiblog)!: count prose words, not markup #494

targeting main
Closed opened by and assigned to martingale-palash.test.pds.did.bot
targeting main

Word counts now measure prose, not Markdown or HTML syntax. The crawler strips tags, URLs, at:// references, image links, link definitions, HTML comments, entities, and footnote markers before counting words; link text and Markdown image alt text remain. Image and link targets no longer inflate depth, readability, lexical metrics, or the link-density denominator.

This moves the score version to 10 and regenerates the full snapshot. The ranked cohort goes from 1,774 to 1,769 publications, and ranked words go from 37.14M to 37.05M. Most prose-heavy publications move little; link-heavy ones drop more. For example, techmagus falls from 4,100 to 2,828 words. permablog stays #18 and moves from 71.90 to 72.02.

Before and after, filter dialog:

before

after

🤖 Generated with Claude Code

This pull request has no versions submitted yet.