I found both Ben Hammersley's XML, metaphorically speaking and Tim Bray's Markup, Namespaces, and Meaning to be good perspectives on the use and usefulness of namespaces in XML.

I also just found Shelley Power's interesting response to Tim's piece, Bray and Symbols and Grounding. And Jon Udell has been in the fray as well.

One of my own suspicions / conclusions about the creation of context that both Ben and Tim are pointing to is that the definition of each context ends up being some kind of document, and usability of that document is very important.

The parallel of namespaces in the iCite net are attributes and attribute sets, which can be mixed and matched in any iCite. And, the basic thing to say about them at this point is that they are themselves iCites (which, in the iCite net, is parallel to the concept of a document).

This isn't vastly different than the idea of creating an XML parsable document that acts as a schema or specification for a namespace. But, the assumption in the iCite net is that the attribute / attribute set document "includes" (links to) all document instances, and this relationship altogether defines the context of schema meaning.

So, picture a namespace declaration in an XML document: in an iCite, the equivalent attribute declaration is a link to an accessible resource, and that link is a two-way link.

As an example use of such a link, in language, it is often useful to know the first the time a term is used to mean a particular thing, and to see its context in use. A dictionary definition for a term could cite such usage.

With XML namespaces, part of the issue that people are wrestling with, I think, is that there isn't an equivalently rich semantic concept about how to use more than one namespace's elements together, or how such uses might clarify or otherwise alter the meaning of each namespace. The uses of namespace elements don't feedback into the namespace—namespace practices are left to being somewhat anecdotal.

As Tim points out in his piece, if people decide on the story of what a namespace element is for, that's very significant. What I am suggesting though is that the whole XML/namespace system doesn't provide structured data that can be used to help make this kind of decision. It is not that such structured data couldn't be created—someone could create a catalog of namespaces uses, but such a catalog is neither built into the system nor part of the standard itself.

In the iCite net, this catalog is built-in—in some sense, the iCite net is this catalog. It will be possible to query what attributes are being used with what other attributes, and to query the context of their uses. And, in fact, the iCite net might end up being a decent system for cataloging XML namespaces.

What I think is an exciting potential in the iCite net is the ability to "pick up" on attributes people are using and come up with new uses (combinations) yourself. If your friends start using bad in a particular context to mean really good, you can too.

And, hey, sorry to everyone who wishes people wouldn't do things like that! I personally love the hip semantic. I just figured someone needed to be working on the slang semantic web!