Due to sidetracking another thread, further discussion is being moved here.
I stated that:
After seeing numerous "newbies" being blasted off these forums by overzealous members going to validator.w3.org almost before going to the very page in question, and then say that making the page validate will probably make the problems go away.Citing validation errors doesn't actually "help" anyone with their problems. Furthermore, it perpetuates the myth that validated markup is the end-all solution to HTML woes.
Yes, this is often true, but it is completely unfair to lambast someone new to this for something they don't know, then refuse to help until they "pay up" and fix their markup. It's a constructive extortion, but still extortion.
And there are massive differences between being "valid" and "well-formed", and having a stable cross-browser rendering. I know for a fact that browsers knowingly implement different means to make up for non-well-formedness, and construct entirely differently DOM's by not completing a tag or closing the wrong tag at positions in the markup. However, failing to validate against a DTD (having invalid markup) is often a "safe" mistake, putting marginleft="0" in the <body> tag isn't going to destroy your rendering in any browser, or making up tags which don't exist in the DTD will just render as unstyled inline elements (and construct identical DOMs, unless you're in IE).
Ultimately, I think you're doing a disservice by lording validation errors over newbies. Nobody has malicious intent when they code a page, so don't act like that they do when validator.w3.org says something isn't quite right. Certainly there are cases when you can't possibly figure out what's wrong because the markup is so terrible, but this doesn't occur as often as forum members simply not caring -- and there are lots of guilty people here (myself possibly included some time ago). There is a lot more to a good page than "validation" (and remember that XML doesn't even need to be validated, for example combining XHTML+SVG+MathML works just fine with or without a doctype), so don't pretend that there isn't.
Or, someone can convince me otherwise below. Let this be an informative thread.