I know this has been discussed to death before, and I apologize beforehand for rehashing this old topic. Please feel free to close or ignore this thread. In any case, I am wondering what the current site policy is regarding the automatic "community wiki" (CW) status.

Here is the background for what prompts this post: I asked a question (http://mathoverflow.net/questions/119619/homotopy-type-of-embeddings-versus-diffeomorphisms), which received a partial answer by Tom Goodwillie. I then placed a bounty on the question. A little while later, I apparently solved the question, and posted my answer (http://mathoverflow.net/questions/119619/homotopy-type-of-embeddings-versus-diffeomorphisms/120186#120186). Being very picky, I edited it several times (for wording, exposition, technical details, etc) but without introducing any significant global changes. So the question got marked CW automatically (it had 3 votes at the time). It was only after this happened that I learned about the automatic switch to CW after too many edits. For the record, it seems to have occurred at least once before with another answer of mine.

In any case, it seems people are accepting my answer, and Tom Goodwillie also endorsed it; so I ended up marking it as accepted (the question now has 9 votes; I assume Tom's endorsement has somewhat swelled the vote count).
In the end, the bounty I placed, and the extra votes got "absorbed" by the CW status. The reputation points are not so much the issue, but the fact that this behaviour of the software seems to have unintended and unwanted consequences:
(1) It marks some answers as community wiki which are otherwise no more worthy of that status than any random answer on mathoverflow. Perhaps I am interpreting the term "community wiki" too literally, but one could say the automatic switch constitutes a misclassification.
(2) It also may discourage beneficial editing and improvement of answers and questions, due to its effect on reputation. One may certainly argue this is a small effect, as it is probably a rare occurrence for most users of mathoverflow. Nevertheless, reputation is partly meant to encourage behaviour which is beneficial for and desired by the community. In this light, the automatic change to CW after lots of edits does not seem entirely appropriate.

Can you thus enlighten me regarding current (and future) policy on the automatic change to CW status after too many edits by the author?

@quid: Thanks for the information. Regarding the bounty points for an answer by the poster of the question, the default behaviour you described seems appropriate. Does it hold for the whole bounty? I.e. if the user answers his own question, does he still forfeit the reputation he placed on the bounty?

Personally, I'm in favor of the policy that too many edits leads to CW status, and I want to discourage minor edits because I don't want to constantly be bumping old questions to the main page. I figure that anyone who is, as you say, "very picky" with edits to their own answer would not care about the loss in reputation points. On the other hand, without something to discourage edits, I can imagine a world where low reputation users periodically make meaningless edits just to bump their old questions/answers and get a few extra votes from people who had not seen those questions/answers before. In my opinion, this is worse than the former alternative, because it means truly new questions remain on the front page for less time. There's already a big vote bias towards content which has been around for a long time (regardless of quality). So searching based on vote count is not a good way to find the best content on the site. I don't want old, lower quality content to clog up the new content which is coming in, and I fear it would chase away our best members if MO got to be like that.

Anyway, quid's point that we can't control the software is the more important one, but I wanted to add my 2 cents that in this case at least I agree with the default software (even if their reason for making it that way isn't the same as my reason for liking it).

@Todd Trimble: I tend to agree regarding the automatic acceptance. However I think OP should definitely still loose the bounty points if no answer is accepted (except perhaps if there is none at all).

If not, there could be endless discussions in case of non-acceptances if actually something should have been accept or some might think so.

I think of a bounty as serving a twofold purpose: offering extra points to somebody that might answer but also giving general extra visibility. The latter is at least as important, IMO. (Because for the former, I mean how likely is it somebody will not answer something they could answer, but will be convinced to answer by, say, 100 points extra. Just seems unlikely to me.)

Thus, one has to 'pay' for this extra visibility, like for an advertisement. (In the end I am even in favor of loosing points in case of no answer, else people could place bounties on essentially unanswerable things and just clutter up the "featured" category.)

@David: I certainly agree that it is not desirable to have posts bumped up just because people are editing them constantly. However, the "threat" of CW status does not seem like a particularly reasonable way to counter that. After all, a CW post is supposed to be a *good* thing, and serves in my mind a different purpose altogether, namely to be a community creation. The above argument was repeated on several meta stackoverflow threads on this issue, particularly this one: http://meta.stackoverflow.com/questions/8654/so-is-too-eager-to-turn-my-edited-answers-into-community-wiki . In that stackoverflow thread, it was suggested: (1) to have cool off periods, in which the thread would not be bumped up by repeated edits by the same person, or (2) an extra option allowing edits which do not bump the question up, among other suggestions.

@Todd: I completely agree with you. We do not want to knowingly accept wrong answers, and absolutely not without some simple mechanism to revert it. But I also agree with quid that it seems reasonable, even fair, for the user to pay for the bounty, even if it comes up relatively empty.