In earlier specifications, I was always among the camp choosing to take the words “address _may_ include” from the specification more liberally. Admittedly, the new language in the HTML5 spec leave less wiggle room, by explicitly stating the element’s intention: “The address element provides contact information for a document or part of a document.” I do however find the “part of a document” intriguing. Obviously, this could refer to a section of a page with alternate authorship, such as a multiple-author blog or a comment.

However, it seems to make just as just as much semantic sense to me to use to provide contact information for a document _not_ authored by the subject. For example, if I have a page about Stephen King, called Stephen King, how does it make any less sense to use for his contact info? I’ve found very little info about why such a restriction would be beneficial. Perhaps we should think carefully before stating that this approach is definitively “the wrong way”. If that is the W3C intention, however, then I may still exercise discretion in how closely I adhere to it…

the examples I’ve seen are using spans and such to represent multiple values, not semantic tags.