If there's a question and an accepted answer for it, the same question basically should not be asked again. However, it's not unusual for the field to advance, and thus some q/a pairs might be outdated. Should the moderator who decides to close the duplicate consider this fact and possibly turn the question into a community wiki?
(related, but not duplicate: Why don't we allow repeat questions?)
In the cases where the question itself is not outdated, but the answers are, I think the best solution is to simply add an answer to the original question (and thus, for any duplicate questions to be closed as duplicates). The closed duplicates will still act as breadcrumbs to the main question, and by adding new answers over time the question will be further enhanced with a historical progression of valid techniques, which is interesting.
The downside is that the original author will likely never bother to update the "accepted" answer, so upvotes by the community will be the primary way of indicating that a newer answer is good. I'm not sure there's too much of a problem with that, though.
I feel that trying to introduce some kind of discretionary duplicate handling policy would be difficult, error-prone, and tedious -- for very limited benefit.
In the cases where the question itself can become outdated to the point of non-usefulness, I still mainly that's a case for closing due to the topic being too localized.
We do have the experimental "greatest hits" route which should highlight the questions that are seen from the outside the most:
https://gamedev.stackexchange.com/questions/greatest-hits
And of course bounties were improved to solicit updates for questions that need it:
http://blog.stackoverflow.com/2011/09/bounty-reasons-and-post-notices/
Leaving the new question open defeats the entire purpose of SE. If I find the newer one it might have a link to the older one, but if I find the older one, hardly ever has someone gone back to update it with information from the new one. This is exactly the kind of problem that happens on forums, which SE was designed to solve, and why questions can be closed as duplicates at all.
You could, I suppose, delete the entire old question after the new one has been approved. That solves that problem, but raises a host of others. A better approach might be closing the old question as a duplicate of the new one, but that's definitely going to be confusing sometimes.
Making the question CW seems irrelevant. Why would you do that? You can edit (or propose edits for) any question/answer. CW is more about questions where rep generation or "the single right answer" makes little sense.
I still don't see a concrete problem here. "The field advances", yes, but not very fast, and rarely in a way that invalidates truly good answers. (Admittedly, I view that somewhat tautologically - one characteristic of a truly good answer is that it rises above the immediacy of "this exact thing is what you do right now with this API-of-the-week/month/year".)
I know this is not an answer, but it is very closely related to the question. I am not sure whether I should open a new question or simply continue in discussion here.
Recently there was a questions about engines in LUA: https://gamedev.stackexchange.com/questions/57964/lua-3d-game-programming
The problem is that google search for Lua 3D ranks the outdated question as first! Now this is a great example of SEO (exploit), but it doesn't really help. The next results is of bigger help I believe.
The asker of the original question actually tried some the answers from the "duplicate" question, but it did not help him.
SE has a motivation system based on reputation, but community wiki seems to be based on good-will only. How do you motivate anyone to update community wiki? Who's responsibility it is to keep an eye on the outdated information?
