The Four-Layer Evidence Stack

Canonical-string ownership is not sufficient by itself. Its strategic value is precise: it is the one layer of a properly executed authority stack that a competitor cannot recreate through execution once it is controlled elsewhere. The case rests on four layers — the first three are the better-supported ones, and the fourth is the distinctive strategic prediction.

Layer 1 — Per-position retrieval advantage

Mechanistic-interpretability research shows a cross-encoder computes relevance through a shared, BM25-like mechanism: matching heads detect exact and semantic matches, an IDF-like component weights them, and relevance-scoring heads aggregate. A canonical-string position on a discriminative compound concept can sit at a favorable point on that curve. The boundary is firm — the build is constitutive. Registration buys the coordinate; the build supplies the token mass and document structure the system can compute against.

Layer 2 — Within-property cluster depth

Comprehensive topical coverage materially expands citation reach. Query fan-out and query rewriting mean deep cluster coverage produces citation breadth well beyond a single page. This is documented at the single-property level: canonical-string ownership improves the position a build starts from, and within-property depth turns that position into topic-wide reach.

Layer 3 — Entity-level signal aggregation

Coordinated declarations across properties — consistent naming across the domain, legal entity, and content property, reinforced by Organization markup and sameAs — help systems resolve and stabilize a single entity record rather than fragmented properties. The mechanism for entity coherence is not speculative; the magnitude across a coordinated multi-domain portfolio is the open question.

Layer 4 — Coordinated namespace coverage

A coordinated set of positions can cover more of the sub-query surface than any one property can alone. This is the distinctive strategic prediction, not the most directly measured element. The mechanism is plausible; the marginal magnitude is a production question a flagship build would test.

The honest characterization, until that build measures it: option value plus denial value, with upside to durable advantage. If the contribution proves non-trivial, irreplicability converts it into moat; if it proves marginal, the layer still retains option value anchored in Layers 1 through 3.