SEO & Distribution Law — the site's auditable perception of how to reach the world
The disclosure
This site is trying to reach the world. That intent is not implied by its behavior — it is stated here, versioned in this file, enforced mechanically at a public endpoint, and open to critique by any reader, human or model. What this site believes about being found is exactly the list below; if the belief is wrong, the place to say so is the comment door, and the fix lands here first.
The mechanical gate
GET https://miscsubjects.com/api/seo/score — the site fetches its own live surfaces and grades itself against every practice in this file, returning per-check evidence a visitor can re-run. Gaps the site knows about and has not closed are returned as declared_gaps, not hidden. The channel inventory (installed vs. installable, with the credential each needs) lives at GET https://miscsubjects.com/api/syndication. Arguing with any of it takes two calls at the comment door: mint a keyless token, write the critique onto any article, and the build answers in the thread.
- Self-score: https://miscsubjects.com/api/seo/score
- Channel inventory: https://miscsubjects.com/api/syndication
- Critique door: https://miscsubjects.com/api/comments
The practices this site holds itself to
Each line names its outside source. A practice with no source is an opinion and does not belong in this list.
Being understood by engines
- Every public page carries canonical, meta description, Open Graph, and Twitter card —
a share that renders as a bare link is a lost referral. (https://ogp.me/, https://developer.x.com/en/docs/x-for-websites/cards/overview/abouts-cards)
- Every article carries typed JSON-LD (Article / TechArticle / MedicalWebPage) with
citations and a BreadcrumbList; the homepage carries WebSite + SearchAction. (https://developers.google.com/search/docs/appearance/structured-data/article)
- robots.txt admits search and AI crawlers alike; llms.txt serves the model-native
self-description. Being read by models is a distribution lane, not a threat. (https://llmstxt.org/)
Being found fast
- The sitemap includes the root and submits only the sourced spine — curation beats volume;
two thousand thin pages read as a content farm and poison the sharp set. (https://www.sitemaps.org/protocol.html, owner curation law in functions/sitemap.xml.js)
- Change is announced, never awaited: IndexNow on every publish/edit/delete of a spine page,
WebSub push on the feed. The key file existed for months while nothing pinged it — plumbing without a caller is the same as no plumbing. (https://www.indexnow.org/documentation, https://www.w3.org/TR/websub/)
- Every public HTML surface states an explicit cache-control policy; timestamps never lie
(a lastmod that says "changed daily" on every page teaches engines to distrust the site). (https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Cache-Control)
Being carried by people
- Every article shows a visible share row for the majors — X, LinkedIn, Facebook, Reddit,
Hacker News, Pinterest (hero as pin image), WhatsApp, Telegram, email — as plain intent URLs, no SDK, nothing loaded before the reader chooses. Social referral is a first-class traffic lane, not a garnish.
- Every article carries an open signed comment thread. Criticism on the page is
distribution: it is the reason a model or reader links back.
- Network posting follows the network's own physics: Pinterest is a search engine (pins
compound for months), Reddit and HN punish automation (operate, don't post blindly), TikTok/YouTube require video the site does not yet produce (a declared gap, not a secret), X/LinkedIn reward the argument in the post with the link behind it.
Being honest about the loop
- Announce lanes without read-back are half a loop: Search Console and Bing Webmaster
read-back comes before adding more announce lanes. Until wired, the site says it is publishing blind — in its own self-score.
Exhibits (why this skill exists)
- 2026-08-06: the IndexNow key file had been served since June with no caller — discovered
only during a full audit. Practice 5 and the /api/seo/score check indexnow_key exist so that class of dead plumbing cannot sit invisible again.
- 2026-08-06: the homepage carried no Open Graph, no Twitter card, no canonical — while every
article did. Practice 1 now covers every public page and the check home_social_identity refuses the drift.
- 2026-08-06, owner order: distribution strategy that came back "Bluesky and Mastodon" when
the question was the majors. Practice 9 and the ranked next_by_leverage list in /api/syndication exist so the answer starts from traffic physics, not from what is easiest to wire.
Operating rule
Before shipping any change to discovery, sharing, or syndication surfaces: run the self-score, ship only if the score does not drop, and update this file and the score endpoint together — they are one object in two projections. A practice changed here without its check, or a check added without its practice, is drift.