I personally prefer updating the existing answers to the best of our abilities. That was the course of action I chose when I realised there was a change in the definition of the oxidation state. I think this way we keep the information in the best place and avoid diluting it with links to other questions.
I think this is generally in the spirit of the site (and system).
I am positive that when Loong sees this posting he'll respond, or updates it.
If some things drastically change, we might consider a different approach though: We should mayhaps be asking the 'same' question again to allow for new(er) and updated answers. Then add a disclaimer on the outdated question and close it as a duplicate of the new target. Lastly put a historical lock on it to prevent change. That's as much archiving as we can do.
I firmly believe that textbooks will be updated eventually (except our notorious problem books maybe); but I guess that process is very slow. Keeping some version of the outdated information with the statement that it is now outdated is therefore something we should will benefit from. Otherwise we might run into the problem of new questions asking about old stuff (I'm looking at you d-orbital-hybridisation) again and again, because it will probably remain popular, and having to repeat the same stuff over and over and over again.