A protocol for making one author a machine-readable entity

A protocol for making one author a machine-readable entity

The question: when Google's systems encounter your byline, can they resolve it to a single, persistent entity rather than a string of characters that collides with seventeen other people sharing the name?

The disambiguation playbook, in order of dependency:
— Mint one canonical author URL on your own domain. This is the entity anchor; everything else points back to it.
— On that page, add Person schema with sameAs pointing to profiles you control and that already carry an established identity (LinkedIn, ORCID for academics, a verified Crunchbase or Muck Rack profile).
— Ensure the name string is byte-identical across every property. "J. Smith" on one site and "Jonathan Smith" on another fragments the entity.
— Cross-link bidirectionally: the LinkedIn 'website' field should point back to the canonical author URL.
— Where the person is genuinely notable, pursue a Wikidata item; the QID becomes a stable cross-platform identifier.

Why this order matters: the 2024 Quality Rater Guidelines repeatedly ask raters to research "who is responsible for the content," and reputation research is performed about an entity, not a string. If the entity can't be resolved, the reputation evidence can't be attached to it.

Caveat: sameAs is a hint, not a directive. Google has stated it does not guarantee consumption of structured data, and entity resolution leans heavily on corroborating off-site signals. The schema lowers the cost of resolution; it does not manufacture an identity that the open web doesn't already support.

What we still don't know: the threshold of corroboration required before Google's Knowledge Graph treats two profiles as the same node, and whether a self-asserted sameAs with no third-party confirmation moves that needle at all.