More interesting posts on the Complexity / Pragmatics of Web Services Protocols

At hinchcliffe.org and this excellent post by Adam Bosworth (Nov, 2004). It's hard to disagree that flexibility and "sloppiness" needs to be accomodated otherwise systems will end up brittle and/or unusable. Ironically this is also what the XML/RDF structures over HTTP activities are all about - bringing in more meaning to reduce dependence on syntax. So:

• Is this just the speedbump before the benefits shine through?
• Is there something fundamentally wrong here?

I have to agree with the sentiment that in the long run more structure is useful/needed. The idea that a few keywords over HTTP does the job is appealing but is flawed in the long run if what expect anything other than humans to configure/code/setup the understanding of the results. The semantics of a message can be defined explicitly (in formal languages which define the meanings of combinations of elements of a message) or implicitly in the processors at either end of the message stream. As long as we have plenty of human engineers at either end talking to each other on how to build those processors (and carry out the actions associated with each keyword) we can survive without making everything explicit. But if we want to reason about the meaning in an automatic way, you need to have everything explicity - which general means complex structures.

Probably at the moment we are falling between two stools - to complex to be really useful to humans and not well defined / implemented enough to really process automatically.

On the other hand Adam's Bosworth's plea for simplicity tells us something else - certain formats catch on if they are flexible and useful - people generally work out how to use them later. Something the Semantic Web still needs to get right...