The joy of HTML is the simplicity of typing <b> for bold; the curse of HTML is that style and presentation is inseparable and almost indistinguishable from content. The medium is not the message, the content is. I know from writing a couple of screen-scraper HTML-to-RSS feeds, that the separation would be a Good Thing.
Jon’s description of feeding XML right into a database and mixing XML and SQL and XPath is pretty intriguing as well. Looking forward to having that access in my database program of choice.