The TAG's Town Hall
December 10, 2003
As a relatively new resident of the East Coast, living two hours south of Philadelphia, and as someone who finds Philly one of the most interesting, most American of American cities, I didn't hesitate when Edd Dumbill suggested I join him at IDEAlliance's XML Conference 2003 at the Philadelphia Convention Center.
When I begun looking at the schedule of events, I was immediately interested in the W3C's Technical Architecture Group's plans to hold a Town Hall meeting. Since I've written about the TAG often I thought it might be interesting to hear TAG members talk about their understanding of their work. I was also interested because I'd heard rumors that the TAG might be moving its Architecture of the World Wide Web (AWWW) to Last Call soon.
The TAG's efforts to create a document which describes the architecture of the Web, which I've written about in two XML Deviant columns, " Identity Crisis" and " TAG and the Web's Architecture", are laudatory and important. I haven't always agreed with every decision, and I suspect that trend will continue; but all the members of the TAG don't agree with every decision that's been reached, a point which I elaborate below.
What I didn't realize, looking at the schedule a few weeks ago, is that the TAG would announce Last Call for its AWWW at the Town Hall held yesterday evening. About a dozen people showed up to the Town Hall meeting, a figure which reflects more, I think, the interestingness of Philly, rather than the interestingness of the TAG's work. I for one had to drag myself out of Chinatown, relying only on a sense of professional obligation.
The TAG members in attendance at the Town Hall included Paul Cotton, Tim Bray, Chris Lilley, Dan Connolly, and Norm Walsh. Paul Cotton chaired the very informal session, prefacing the Q&A session with a few remarks. One of the interesting things about the TAG is that it's unlike an ordinary W3C Working Group. For example, as Cotton pointed out, the AWWW has entered Last Call with unresolved, open issues. That can't happen in ordinary Working Groups.
The AWWW itself is interesting in that it's so completely belated, which is to say that the Web has been a vast success for a rather long time, and no one has ever sat down to document, in anything like a formal or rigorous way, the architectural principles which it expresses. (Tim Berners-Lee has written about the architecture of the Web, but I don't believe he ever intended those writings to be formal or especially rigorous.) In that sense the AWWW was called by various TAG members, during the Town Hall, an "emergent architecture" -- a very odd, yet interesting term, if you think about it much. Imagine an architect drawing up plans for a building 10 years after it's been inhabited.
The AWWW is more like a document written by a group of architects as a kind of retrospective of how they conceive of and build habitations. It's more like a document purporting to describe the principles of architecture, per se, rather than the architecture of a particular building or project. I'm not sure this architecture analogy really enlightens much, but I think the belatedness of this document is worth thinking and talking about. It suggests, at the very least, something interesting about global information systems.
Various TAG members pointed out some of the warts (either of implementation, design or specification) of the Web's architecture, a list which should be familiar to anyone who's thought seriously about the Web. They include user-agent content sniffing; namespace document "waffling" (to use Tim Bray's term); fragment IDs, particularly in the context of multiple resource representation types; extensibility and versioning of vocabularies, formats, and protocols (though, as Chris Lilley and others suggested, no one really does this well); and, in various ways, qualified names.
While I intend to write an XML Deviant in the near term about the TAG and the AWWW in particular, let me offer a few commendations about what they've done so far. First, as a matter of process and community civility, the decision to extend the AWWW's Last Call comment period until the first week of March is very welcome and very prudent. Given the fact that the AWWW will be something that people are pointed out for a long time to come, it is crucial that the community is free to provide as robust a vetting of the AWWW as possible. Second, I'm happy that the AWWW clearly punts on what is called these days the "social meaning" issue, but which is really a whole cluster of issues (see "Social Meaning and the Cult of Tim" and "The Social Meaning of RDF" for discussions of this issue cluster). This issue cluster will eventually have to be solved, or at least finessed, but the prudent thing to do was to put in the context of the AWWW.
The most interesting part of the Town Hall was the discussion about whether the AWWW should have had a conformance section. Since many of the points the AWWW makes are really directed at people, rather than at the systems which people design, it's not clear what it would mean to specify conformance requirements. One could say, in general, that systems which instantiate the principles expressed by the AWWW are conformant, but that doesn't have much bite. So perhaps one could translate the AWWW's statements of best practice into statements of conformance constraint or requirement? Well, perhaps, but it's not clear that the TAG has intended any of those statements of good or best practice to have any normative weight whatever. Yet, for all these and other difficulties, some of the TAG members at the Town Hall seemed intrigued and excited by the idea of an AWWW conformance section.
Also in XML-Deviant
The final point I'd like to make about the TAG Town Hall and the AWWW Last Call is twofold. First, there are a set of TAG findings which amplify various of the principles and practices expressed by the AWWW. This means, at the least, that the AWWW is a deceptively short document. But it also means that there is more backing and foundational material available if you find some parts of the AWWW thin or wrong or otherwise inadequate. Second, I encourage everyone who's at all interested in the Web, and who follows these sorts of issues at all, to read the latest draft of the AWWW. Having read it carefully, do what the TAG has asked that we, as a community, do -- pass on comments, questions, suggestions, improvements, and the like. I intend to do this, and I encourage others to do it, too.
A last bit of news from today, Wednesday, at the conference: IDEAlliance awarded the XML Cup, 2003, to Adam Bosworth and C.M. Sperberg-McQueen in recognition of their contributions to the development of XML and to the industries surrounding it.