Summary

  • AFRINIC’s News & Announcements route returned HTTP 200 on 31 August and carried a news title, visible heading, current announcement list and news-specific JSON-LD.
  • In that same response, the canonical URL, ordinary description, Open Graph title and URL, and Twitter title identify the separate Privacy Policy page.
  • AFRINIC’s sitemap lists news and privacy separately. RFC 6596 says a canonical target must be duplicate content or a superset, so the declaration conflicts with the two pages’ captured roles.
  • This proves a publication-identity mismatch, not deindexing, traffic loss or a misrendered social card. The smallest control is a versioned route identity receipt across visible and machine-facing surfaces.

A 200 response answers only the first question

At 05:46 UTC on 31 August, AFRINIC’s news route returned HTTP 200. Its browser title was “News & Announcements - AFRINIC - Regional Internet Registry for Africa”. The visible H1 said “News & Announcements”, and the body opened a 2026 list containing announcements about the Appeal Committee, the CEO designate, the stakeholder survey and other current matters.

The page’s JSON-LD agreed. It named a News & Announcements page and supplied a URL ending in news.html. Breadcrumbs and navigation pointed to News as well. A reader arriving in a browser therefore receives the expected archive. This is not an outage story, and it is not evidence that the announcements disappeared.

Reachability, however, is only one state. The same response also tells automated consumers what object the page claims to be. That second state is where the identities divide.

The head names a different object

The canonical link in the news page points to https://afrinic.net/privacy.html. Its ordinary meta description says it is the official AFRINIC Privacy Policy page. Its keywords begin with “privacy policy”. Both Open Graph title declarations use the Privacy Policy title; the absolute Open Graph URL points to the Privacy Policy; and the Twitter title and description do the same.

This is not a harmless difference between www and the apex domain. The destination is a different content role. The separately reachable Privacy Policy has its own Privacy Policy browser title, visible H1 and self-canonical. AFRINIC’s sitemap also lists /news and /privacy as distinct resources.

The result is one HTTP representation with two machine-readable identity stacks. The title, H1, body and JSON-LD describe news. Canonical and social metadata describe privacy. Checking that every field exists would pass; comparing what the fields mean would fail.

Canonical is a relationship claim, not a redirect

RFC 6596 defines the canonical relation as a preferred identifier among resources with duplicative content. More formally, the target must contain content that duplicates or is a superset of the referring page. The RFC describes why applications may focus processing on the canonical resource and update references to it.

News announcements and a privacy policy do not become duplicates because one document declares the other canonical. The captured bodies give them different titles, headings and purposes, and AFRINIC’s own sitemap keeps them apart. The defect is therefore in the relationship declaration, not in the existence of either page.

That distinction also limits the claim. A canonical link is not an HTTP redirect. The RFC says an application can ignore an improper declaration and apply its own heuristics. Without search-console or platform evidence, it would be wrong to say that a search engine removed the news page, consolidated its signals, reduced traffic or lost any announcement. The served source proves the instruction; it does not prove every recipient’s response.

Social identity fails along a separate path

The Open Graph protocol gives two of the affected fields a precise role. og:title is the title of the object as it should appear in the graph. og:url is the canonical URL used as that object’s permanent graph identity. On AFRINIC’s news response, both fields describe the Privacy Policy.

Again, this is evidence about source metadata. It does not prove that Facebook, X or another named service actually fetched the page, cached it or displayed the wrong card. A platform may choose other fields, preserve an older cache or reject the declaration. Search identity and social-preview identity should remain separate evidence lanes even when the same copied values caused both mismatches.

The JSON-LD adds a third lane. Unlike the canonical and social fields, it still calls the object News & Announcements. A machine extracting several vocabularies from the page can therefore receive contradictory answers without encountering a missing tag.

A route needs one reviewed identity record

The smallest durable control is a publication-surface identity receipt for each public route. It should bind the requested and resolved URL to the reviewed content role; HTML title and H1; canonical URL and the reason the target is duplicate or superset content; description; Open Graph and Twitter identity; structured-data type, name and URL; sitemap membership; exact build or byte hash; validation result; approved exceptions; and correction history.

The validation must be semantic. A canonical check that only asks whether the target returns 200 would accept this pair. A better check asks whether the target represents the same content or a documented superset. An Open Graph check that only counts required properties would also pass; a better one compares the identity they announce with the reviewed route role.

For the current page, a repair need not touch the news archive or the Privacy Policy body. It can align the news page’s canonical, description and social fields with the content that is already visible, then preserve before-and-after hashes and a correction time. That gives future reviewers evidence of what changed without pretending that external caches changed at the same instant.

Keep four states apart

The case is easier to govern when four states are not collapsed:

  1. Reachability: the news route returned 200 and served a substantive archive.
  2. Source identity: canonical and social metadata identified the Privacy Policy.
  3. External processing: no search or social outcome was observed in this evidence set.
  4. Correction: no later source representation has yet been used here to prove repair.

Those separations protect both sides of the record. AFRINIC gets credit for a functioning news page and a clear public archive. Readers also get a precise account of the machine-facing mismatch. Repair can then be verified as a bounded release event rather than inferred from a refreshed browser tab.

Sources