Standards compliance needs to be built into RFP's from the get-go and then enforced by companies who pay the web-dev's.
An example of bad code: http://web1temp.cheme.cornell.edu/courses/cheme112/
Whoever made that site is basically having their bad code justified by browsers that actually display it, even though it has no doctype, no html tag, no head tag, unclosed p tags... etc etc etc. If that person was doing the same thing in Java or C++, the compiler would spit out a bunch of wierd messages and he/she would realize he/she doesn't know what he/she's doing.
If browsers didn't parse bad code, that would probably stop a lot of xangas and myspaces from working... I can dream, can't I?
No html tag! I can't get over that one!
