Jeff explores the tricky idea of expanding the discussion of web standards to include acknowledgment of things that happen before any pointy brackets get involved, and so raises the question of what exactly the “structure” of the data is; as someone who daily develops applications which use a whole lot more output formats than just HTML, I really appreciate this sort of holistic view of how the web works.
Also be sure to read Wilson’s comment. It’d be great if he had a blog or something where he could regularly write such insightful stuff ;)