recently i wrote about HTTP Philosophy and how the Semantic Web now runs into some very fundamental problems: if the
Web part of its name should be more than just clever branding (which it certainly is), then it has to be built on the assumption that semantic descriptions can be somehow combined based on commonly used concepts. and the really challenging question (which is not so much a technical issue) is how these concepts, or, more specifically, their identities and identifiers, are created, defined, described, understood, used, managed, and made available.
the W3C just published a report about suggestion:
If you do not want to import another ontology in its entirety (e.g. because it would be too large, too buggy or would introduce unnecessary constructs), you can still 'copy & paste' portions of the ontology into your own.
isn't that the exact opposite of what the Semantic Web should be about? if people start copy&pasting ontologies or parts of them, then any identity that might have existed on the semantics level will not be accessible on the Semantic Web level. however, i do see the pragmatics behind such a suggestion, because very often it might be impractical to properly reuse a complex ontology.
but still, things like that point to the very fundamental issue of how much embedding into some external semantics an application scenario wants to have or is able to afford. real-world applications may very often run into the situation where they start copy&pasting (parts of) ontologies. which in the end creates many Semantic Webs with no connections between them, which would kind of defeat the idea of the Semantic Web...