I agree with Lee. As I've mentioned to Dave, I'm somewhat concerned about
creating yet another math DTD fragment when choices like AAP and ISO 12083
exist, but I'm even more concerned with using, as Lee puts it, the murkier
aspects of SGML to theoretically make it "easier" for people to create
HTML in vi. Maybe it is a reasonable goal to continue to make it feasible
to edit textual HTML with vi, but as you add tables and math and other
complex constructs, that goal become infeasible. And delving into more
complex markup minimization to reduce keystrokes does not guarantee that
you make it easier for people to understand how to code it correctly.
I'm quite sure the trend will be toward user-friendly tools to create
the more complex HTML constructs such as complex equations [of course,
I would be, wouldn't I], and I don't see the advantage of complicating
the language and parsers to save a few keystrokes.