People spend way to much time worrying about HTML Validation. It's easy to understand why: when the client knows just a enough to be dangerous, the valid benchmark resolves disputes.
But, for most sites, what you want is just that the damn thing work in the major browsers. Lots of mildly invalid HTML works just fine. And it's quite possible to write valid code that doesn't.
Still, for one brief shining moment, the cover page of markbernstein.org is 100% valid. This is mostly a stunt: I wanted to be sure there wasn't anything in Tinderbox that keeps you from building an absurdly elaborate page that is still 100% valid.