NamespaceDistortion

Namespace distortion is the practice or act of issuing data (such as RDF/XML files) incompatible with the ontologies available from the namespace URIs. (Contrast with BadNameReuse, which is similar, but doesn't focus on namespace documents.)

This can be done by accident (a misunderstanding, a typo), with benign purpose (imagining an improvement having been made) or with even with malice (perhaps for commercial gain).

Contrast with NamespaceSquatting, where there is no authoritative ontology. Here, there is an authoritative ontology, but it is not properly used.

The exact meaning of "incompatible" depends on whether the ontology is fully expressed in a formal language. If it is, then "incompatible" simply means logically inconsistent under the logic of the ontology. If the ontology includes human readible elements (such as with rdfs':comment statements), then compatibility becomes a more complex human/social issue.

Are there any specifications which say authors MUST or SHOULD not engage in Namespace Distortion?

Should there be?

Example/Issue

RDF Semantics says that rdf:first is not a many-to-one (functional) property, (for example, { x rdf:first y, z } does not entail { y = z } ) but that "semantic extensions" may define it as being so. Would defining it to be so be be namespace distortion? Does it matter whether it's done by some third party, the W3C team, or the a W3C working group?