It would be nice to be able to run a message through a validator to get some indication of _major_ screw ups. This would not have to be part of your script language.
I don't think it's plausible to expect we could prevent screw-ups by blacklisting scripts. What are the chances that the same mistake will be made ever again? Next time it will be some other unforeseen error. And if you use heuristics for filtering, then what if there is a bug in the filter itself? I'm mostly impartial though, just not convinced about the usefulness of adding this particular mistake to the protocol specification. (I guess others wrote the very same reply.)