Nick at work sent a link to an article that is a follow-up on Tim Berners-Lee rebooting of the HTML standards process. The interesting bit is some commentary from Ryan Paul (my emphasis below):

There are also doubts about W3C's ability to create a usable forms standard. Tim Berners-Lee claims that there "are many implementations and users of XForms." As a developer with first-hand XForms experience, I have to say that the word "broken" best characterizes the standard. Although the concept looks great on paper, it just doesn't work for real-world projects. No widely used browser supports the standard by default, and free XForms plug-ins available for Internet Explorer and Firefox are so incomplete that they can't be used for anything other than experimentation. The standard itself also has considerable failings, and some server-side XForms implementors have actually deviated from the standard and invented new elements to work around the holes. Based on my own experience, I think that better support for automated validation control and XML serialization in XUL (Mozilla's XML-based interface development language) would be far more effective than XForms.

In my opinion, XForms failed as a standard because it wasn't developed with implementation in mind and it didn't have early support from browser developers. The W3C has clearly learned from the experience, and Tim Berners-Lee hopes that the new HTML working group will avoid those pitfalls: "Some things are very clear. It is really important to have real developers on the ground involved with the development of HTML. It is also really important to have browser makers intimately involved and committed," writes Berners-Lee. If the working group addresses the most pressing concerns that impact web developers and Internet users, the organization could regain relevance and bring credibility back to the web standards process.

Right on. Now, I'll admit I'm biased because I work on Microsoft InfoPath which is an XML forms editor down to its very core and most often compared against XForms (which... doesn't really seem to exist as a competitor but, hey, it's a standard).

Note, however, that InfoPath is built upon a foundation of web standards: XML, XSLT, HTML, and XSD. The last one is the bitter pill. XSD represents to me the shift from useful standards to over-intellectualized "who in the world understands all this?" - let alone dealing with vendor implementations + subtle differences. You can design an XSD schema by hand that is pure and correct but the various implementations of XSD validators will choke on, or at least not all agree upon.

But anyway, InfoPath is built on W3C standards. I've been thinking as of late the last time I remember TimBL bringing down the hammer: it was for XML namespaces. He more or least knocked some heads together and told people they had to do better.

XForms is a failed standard. You read it and your inner geek says, "Man, I don't know... sure seems  like there are a bunch of carts in front of the horse here. Can we go back to the begining and work on this?" XForms should be abandoned and replaced with something that organically makes sense and that IE and other browsers are capable of implementing well - and motivated to do so.

This would send an excellent message about how not to do things.