The "HTML is an error condition" quip was actually what lead me to that piece. And the fact that you need a specific
validator to actually check the source is actually a pretty strong indicator of the magnitude of the problem. Browsers simply don't care, or worse,
have to not care, that what they're getting is valid HTML. Rather, it's that they will, and pretty much
must, render pretty much any random crap.
See my recent comment about Web fonts, and additional discussions about Web accessibility for the disabled (a very good friend has exceptionally limited vision), and think over how a truly semantically constructed Web document might be just slightly easier to parse by assisting software or systems.
https://news.ycombinator.com/item?id=10326649
https://plus.google.com/104092656004159577193/posts/f6mP6XqH...
A rigorous document structure, strict adherance to it, and penalties (e.g., lower SERP ranking) for invalid / poorly constructed pages would help.
As for Saoirse Moen, her (last) day job was in fact working on a Web browser. Dealing with crap HTML, JS, CSS, etc., etc., etc.