If we've learned anything from HTML/CSS it's that, declarative languages (like XML) work best to describe User Interfaces because:
It's easy to build code preprocessors that can template the code effectively.
The code is in a well defined well structured (ideally) format so it's easy to parse.
The technology to effectively parse or crawl an XML based source file already exists.
The UIs scripted code becomes much simpler and easier to understand.
It simple enough that designers are able to design the interface themselves.
Programmers suck at creating UIs so it should be made easy enough for designers.
I recently took a look at the meat of a WPF application (ie. the XAML) and it looks surprisingly familiar to the declarative language style used in HTML.
It's blindingly apparent to me that the current state of desktop UI development is largely fractionalized, otherwise there wouldn't be so much duplicated effort in the domain of user interfaces (IE. GTK, XUL, Qt, Winforms, WPF, etc).
There are 45 GUI platforms for Python alone
It's painfully obvious to me that there should be a general purpose, open source, standardized, platform independent, markup language for designing desktop GUIs. Much like what the W3C made HTML/CSS into.
WPF, or more specifically XAML seems like a pretty likely step in the right direction. Why hasn't anyone in the Open Source community (AFAIK) even scratched the surface of this issue.
Now that the 'browser wars' are over should we look forward to a future of 'desktop gui wars?'
Note: This topic is relatively subjective in the attempt to be 'future-thinking.' I think that desktop GUI development in its current state sucks ((really)hard) and, even though WPF is still in it's infancy, it presents a likely solution to the problem. Has no one in the OS community looked into developing something similar because they don't see the value, or because it's not worth the effort?