Canonical fragility
Canonical fragility designates the vulnerability of a declared truth when its authority depends on too narrow an anchoring: a single page, a single format, a single access path, or a single signal type. In an interpreted web, a fragile canon can be accurate, but remain difficult to activate, easy to invisibilize, or simple to parasitize.
A robust canon does not depend on a single point. It exists across multiple surfaces, with coherence, controlled redundancy, and enforceable evidence.
Definition
Canonical fragility is a situation where:
- canonical information is correct and public;
- but its authority rests on a unique or overly concentrated anchoring;
- and this concentration makes the canon vulnerable to invisibilization, neighborhood contamination, capture, or collision.
Canonical fragility therefore measures the activation resilience of the canon in automated interpretation systems.
Why this is critical in AI systems
- A page can be ignored: a unique canon may never be activated in responses.
- A page can be reframed: dominant neighborhood and standard vocabulary override the intent.
- A page can be imitated: canonical mimicry and signal parasiting.
Common forms of canonical fragility
- Single-point fragility: a single URL carries the definition.
- Format fragility: truth contained in a format that AI activates poorly (PDF, image, buried section).
- External dependency fragility: truth depends on third parties (wikis, profiles, aggregators).
- Evidence absence fragility: no enforceable trace allows preferring the canon.
Practical indicators (symptoms)
- The canon is accurate, but rarely mobilized in responses (invisibilization).
- A small query variation is enough to make the canonical definition disappear.
- Secondary sources dominate the subject’s explanation.
- A competing page “takes over” the official definition.
What canonical fragility is not
- It is not merely a problem of editorial quality. It is an activation fragility.
- It is not an obligation for massive duplication. Redundancy must remain governed.
- It is not a lack of content. It is a weakness of structure, evidence, and surface.
Minimum rule (enforceable formulation)
Rule CF-1: any high-impact statement must be protected against canonical fragility by at least: (1) an explicit canonical definition, (2) governed negations against probable confusions, (3) multiple activation surfaces (links, graphs, satellite pages), and (4) enforceable evidence (interpretation trace, fidelity proof, version power).
Example
Case: an organization publishes a single “definition” page. AI systems still explain the concept according to more dominant secondary sources.
Diagnosis: canonical fragility (single point + absence of evidence + dominant neighborhood).
Expected correction: reinforce the canon through surfaces and links, produce evidence, and neutralize contamination.