Related link: http://monasticxml.org/mirage.html
XML is powerful stuff, but developers need to take information seriously rather than expecting XML to solve their problems if they want to get worthwhile results.
About a month ago, I put up “monasticxml.org“, a deliberately unpretty site that looks at how markup works and how an understanding of markup structures might better inform the ways that people use XML. I guess it’s reasonably clear that I’m disenchanted with how most people use the stuff, not to mention the enormous baggage the W3C seems intent on piling on top of it. (At least the hype wave is fading.)
Underneath all the dreck, however, there are still some sparkling gems. Markup as a means of sharing labeled and structured information is pretty powerful stuff, though figuring out the labels and structures is still hard work, and stuff XML itself doesn’t address. Developers can make excellent use of XML, if they’re willing to stand back and look at their information separate from markup syntax and processing expectations. Create open and extensible information structures, and XML can follow.
Does ‘monastic’ XML make sense, or do developers really need a ‘decadent’ XML toolkit that does their work for them?
Excellent site
It's a message that needs to be heard by many of the people working with XML today.
One thing it doesn't cover that I personally think is also a problem is the effort to encode the semantics of a document using it's syntax. One of the largest problems, obviously, with exchanging xml documents, especially between orgainzations, is agreeing on a common understanding for what the data means. Currently programmers seem to be proceeding down the road of defining schemas etc., searching for some Holy Grail where an appropriate syntax can remove all semantic ambiguity. This is just not possible. In fact, to be a bit pompous about it, Godel's Incompleteness Theorem proved it's not possible. Any system (syntax) contains truths (semantics) that can not be proved within the system. I think people need to face up to this directly and recongize that at some point in the process of defining the data exchange human intervention may be required.