As someone who writes a browser ...

.... I know of a couple of simple way to improve standards compatibility.

1) No standards without reference implementations or even closed source proven implementations, that means you W3. If you can't check the validity of your own specs in running code, why is everyone else supposed too?

This isn't rocket science, it's the method that IETF has used with its RFCs for years.

2) No standards without full test cases, when your standards feature non-local user content, the specs need to define the failure behaviour as well.

Since the mid 90's the W3 has made over 90 specifications available, maybe they should spend a little more time checking they really work.

Peter