I did some digging into the upcoming Open XML formats and found a great article by Ted Pattison explaining the changes MS have made. Essentially the new docx format is a zip file containing a bunch of other files mostly XML files that make up a word document ms refer to this as the 'Package'. Ted has written a great article that explains how it all hangs together in some detail.
So why did MS go to all this trouble? Well folks like me want to be able to programatically work with Word documents on the serverside so we can automate certain processes. The old binary formats supported by previous versions were unfriendly so we tended to use the word object model or VBA to do this. These were not very performant and less than 100% reliable. So according to Ted the new formats will allow manipulation of word files without installation of Word via some new .Net classes. To start with this will not be straightforward, first you will have to familiarise yourself with how a docx package is structured, but I imagine some coding gurus have begun work on this already and by the time Office 2007 is released you will be able to purchase a 3rd party component that will make indenting paragraphs a snip.
We have a need to manipulate bookmarks so we can inject data into documents prior to converting them to PDF's. I am on the lookout for worthy open source projects that are attempting to undertake this task, I would gladly make a contribution to their efforts in order to reap the rewards )
Another great article that goes into more detail
A project with example code on how to create a simple docx from scratch.