Read:http://old.opendocumentfellowship.org/Articles/Int roductionToTheFormatInternalshttp://old.opendocumentfellowship.org/Articles/For matODFVsMSXML
And let me know if you still think the ODF is merely a 'memory dump in angle brackets'. Maybe they could have reused a good chunk of CSS, but that would also require another type of basic parser in implementations. I imagine you've heard of expat, but can you name a standard CSS parser library? I can't, and once upon a time, I had CVS checkin privs on mozilla. Looks simple enough, but ask a web developer if they've ever heard of any major browser having CSS parser bugs.
And it looks like ODF's style definitions could maybe be generously described as CSS in XML, too. Regardless, I think you could make a pretty compelling argument that the layout needs that have historically driven CSS are a little different than a word processor's needs.
Back when I worked on Abiword, the native format was very similar to XHTML/CSS. Some arbitrary element renamings -- I believe our equivalent to the span tag was a single letter. The XML->XHTML conversion could probably have been handled by a simple sed script.
For styling, we reused as much CSS as possible. I learned about a lot of nifty stuff in CSS3 back then. I hope I get to use some of that stuff in browsers some day. But we were well on our way to the first draft of a hypothetical CSS3 Wordprocessor Module, too.
The OOXML format does strike me as a brain dead C struct to XML encoder, however. And I know the doc format pretty well, having written some non-trivial bits of wvware and the Abiword importer based on it. We actually once got a post on the mailing list from someone looking for technical details on the doc format, and they had been forwarded to us by someone on the Word team at Microsoft. They had their time-tested, battle-worn libraries, but we apparently understood the actual bytes better than anyone still in Redmond willing to help a customer.
But we all knew that the eventual Microsoft XML format was going to be silly. Actually, it's better than I expected. I had considered the occasional base64 encoded binary data structure wrapped in data tag to be a very real possibility.
In my mind, the most astonishing thing is that they just arbitrarily reimplemented -- and generally very badly -- dozens of standards, including many ISO ones. I believe they have several novel timestamp definitions, in addition to ISO's.
I'm pretty shocked anyone is even pretending OOXML is being seriously considered as a standard. I think some people in Redmond had an April Fools' joke get out of hand. If this gets standardized, I expect the next anti-trust case is going to reveal internal Microsoft emails with text such as "holy shit, ISO just accepted our format!"
PS: I don't even read slashdot that often anymore, and I very rarely post. The few times I do, I generally don't even bother to login. But it would seem that several years of random hobbyist open-source contributions have made me quite likely one of the top few dozen or so domain experts on the planet regarding your specific post. I thought that was kind of amusing myself. I don't know if anyone actually cares, but my name is Justin Bradford, and I imagine google retains sufficient evidence of what I claim.