[dragonwriter]

> A large part of the complexity of HTML is simply quirks and compatibility.

Actually, the HTML spec already addresses those (that's why it is incredibly complex; unlike HMTL4 and previous, the WHATWG HTML Living Spec -- and possibly the W3C HTML5 spec, though I can never keep straight what that was in the WHATWG spec at the time W3C kept and what it didn't -- contains a complete specification of how compliant user agents should parse anything purporting to be HTML even if it actually isn't valid HTML (IIRC, a compliant parser may throw an error on invalid HTML, but if it is tolerant of errors, the spec specifies how it is to be tolerant, specifically to avoid the pre-HTML5 issue of different browsers parsing the same thing different ways. Modern browser either have converged or are converging -- some might still be lagging -- on that consistent model.)