> and the savings costs associated with invalid markup
> (bandwidth, time, etc.) are very easily enumerated.
Please, enumerate. I guess you are not aware that a lot of things you consider invalid are actually allowed in HTML: unquoted attributes, tags omission, etc. This was allowed even in HTML4, HTML5 gives us even more options there.
You could remove DOCTYPE (which is very short compared to HTML4 or XHTML versions), but then you'd throw your browsers to quirks mode, and that is a pain. The whole purpose of DOCTYPE in HTML5 is to force standards-compliant rendering mode, that's it.
And it only looks that browsers have little problems parsing invalid markup. Making sense of tag soup parsing (and making it consistent) was one of the major tasks for HTML5 people.
I suppose I'm referring more to 4.01, since 5, as far as I know, doesn't constitute the majority of websites. Things like nonstandard attribute tags and wrapping inline elements around block elements. These are what I thought of when I thought of "dated".
Majority of websites can be converted to 5 just by replacing doctype. Nonstandard attributed don't help you to save bytes though.
Also, from what I've seen sites which were built without even thought about validity and standards tend to be of much lover quality and hence are more bloated: lots of unnecessary div's, classes, wrappers, etc.
Say, how would invalid markup help http://kod.as/lab/valid/ be even smaller? Sure, that is pretty meaningless document, but take a look at a source and you will see some stuff that can be applied to normal sites. It's valid: http://validator.nu/?doc=http%3A%2F%2Fkod.as%2Flab%2Fvalid%2... http://validator.w3.org/check?uri=http%3A%2F%2Fkod.as%2Flab%...
You could remove DOCTYPE (which is very short compared to HTML4 or XHTML versions), but then you'd throw your browsers to quirks mode, and that is a pain. The whole purpose of DOCTYPE in HTML5 is to force standards-compliant rendering mode, that's it.
And it only looks that browsers have little problems parsing invalid markup. Making sense of tag soup parsing (and making it consistent) was one of the major tasks for HTML5 people.