mattur - 3:53 pm on Mar 29, 2013 (gmt 0)
A real standard (one that's not a moving target) is e.g. needed in contracts between a web developer and a site owner. You can't validate the work if the standard you agreed upon shifted (maybe or maybe not) during the work.
You're assuming "standard" = "implemented interoperably in browsers". This has never been the case, and part of our job is explaining this to clients.
That's another of those "bright" ideas: they dot rid of that pesky dtd as it was enabling us to say "invalid" all too easy.
A DTD isn't required for validation, and in fact DTD's aren't sufficiently rich enough to express the complexity of HTML/XHTML anyway. This is why the XML world also largely abandoned DTD's years ago, in favour of RELAX NG, Schematron etc.
The tag soup promotion in the html5 specs and the attitude of everything is optional is just pain wrong.
Late 90s cargo-cult XHTML-fandom seems to be the problem here :) The spec defines HTML and XHTML serialisations, so there's nothing to stop you using XHTML, but most people prefer to spend their time on stuff that actually matters.
W3c HTML5/HTML5.1 makes XHTML1 appendix-C style void elements conforming in HTML, so authors don't need to worry about whether to use:
<meta name="pinterest" content="nopin" />
<meta name="pinterest" content="nopin">
both are fine when using <!doctype html> and this makes things a lot simpler for everyone. Think of it as graceful error-recovery for the XHTML years. ;)