The problem is that there is no-one who is willing to stand up in a university and say "we will not use Microsoft formats no matter what desperate deal they offer us". If that happened then so much proprietary stuff could actually be cleared out.

Essentially what we need is someone who understand documents and XML to make the decision about how to switch to XML. But that may well not happen.

If all or even just most, other office suites agreed to respect a XML-dreivative (hopefully one which is an extension of the XHTML standard, rather than a new standard altogether), then sufficient pressure could be brought to bear upon M$ that they are forced to follow the standards (as is the case with IE, where thier support for standards is gradually imporving). The format shoudl also support BibTeX bibiliographies.

The XML-based format should be based on the concept of keeping the structure distinct from the markup, by using CSS to handle all the formatting. This would allow an ODF2LaTeX converter to be used, effectively providing a 3-format world of (La)TeX, ODF, and PDF (or an equivalent open format), and allow simple conversion between the two source formats, thus allowing word processing, typesetting, and fianl documents to be used easily and interoperably.

Remember that LaTeX provides much more power than is available from any word processor, whilst still being easy to use. There are plenty of GUI front-ends, and some provide a word processor like preview, although to use all the power of LaTeX you still need to be able to handle the code. THe code is simple to learn, and very consistent, making it much simpler than writing a static HTML+CSS web page (although I would say that tables are a pain when hand-coding either), since virtually all the "styles" are predefined in templates available for download.

It's way too early for MS to submit OOXML to any standards organization. It's specs will change with every service pack, and the next generation of Office will likely use something called OOXML that is significantly different than the current, incredibly bloated implementation.