Well now isn't this interesting... I've not cracked it open yet, but one of the big use cases I see for the XML type in Yukon is the storage, query and serving of WordDocs. Now sure, you can use XQuery to shape the XML you from those instances, but sometimes it might just be easier to let XSLT to the job. Of course, the problem is... how you write XSLTs against WordML?
Been there, done that. Got myself quite the headache for my efforts. So when John Durant posted this, it made my day:
Microsoft Word 2003: WordProcessingML Transform Inference Tool
Obviously, WordProcessingML is great because you can run a transform against the XML data files and get nicely formatted Word docs to send out to information consumers. But, what should the transform look like? How should you structure it? To be sure, there is no replacement for cracking open a good text on XSLT [ed: link added] and getting up to speed on it, especially when you want to do advanced things. However, the Inference Tool lets you create a "seed" document in Word where you load one of the raw XML data files into a doc and format it to your liking. You save the Word doc as an XML document marked up with WordProcessingML. Then, you run the tool against your saved file to produce the XSLT.
Ta-Dah! BRILLANT!