No, it just makes things worse. If people are going to follow the spec, they should follow it and reject bad input. Otherwise, you get a mess of code that tries to guess at what to do. This leads to inconsistencies, possibly security bugs.

Imagine if invalid HTML had been flat-out rejected by browsers since the beginning. Cross-browser compat would be, at least, somewhat easier due to less bad HTML surviving in the wild, since it simply would not have worked. In XML, for instance, I've never heard of the need to deal with invalidly formatted messages.

There's enough problems interpreting semantics, why do folks want to add syntax to the problem, too?