What is our policy regarding closing an old challenge as a duplicate of a new one?

I know this happens on other sites (at least SO), if the newer question is generally better or usually if it gathers better answers (such that the old one acts as a "redirect" to the canonical answer). Now this doesn't really apply here, but today there was a case where I'd be tempted to do just that.

The challenges in questions are:

- http://codegolf.stackexchange.com/questions/32475/write-a-word-equation-solver
- http://codegolf.stackexchange.com/questions/34926/write-a-fast-word-equation-solver

The former is mostly a [tag:code-golf] question, but with an odd rule regarding how whitespace is counted - which basically made the actual task fairly irrelevant and turned it into "decode a whitespace-encoded program in the least amount of characters".

The latter has exactly the same task as the form, but is a [tag:fastest-code] challenge. Now Peter Taylor (and apparently four others) was of the opinion that the new one is a duplicate of the old one, since all of the old answers are *valid* on the new one. For starters, I'd say that's irrelevant if none of them are competitive, but that's not the point of this question here. [(I address that over here.)](http://meta.codegolf.stackexchange.com/a/1876/8478)

The more important problem is, that due to the old question's scoring being easy to abuse, any competitive answer to the *new* one could still be turned into a winning answer for the old one. So now you could indeed craft an answer that works on both of them. But to me, this is just saying that the new challenge is really decent, and the old one has some serious problems, and (since answers can be copied from the new to the old one, but not the other way round) that kinda makes the old one a duplicate of the new one.

So what is the community's opinion on this? Do we close old challenges in favour of better new ones? Or is there a problem with that in some way (like being punishing for the original creator of the challenge concept)? Have there been any precedents?