Hindsight on acceptance percentage related to the hypothesizing seen in this question: Will the “Accept rate” score encourage unwanted behavior?.
I've had a couple of accepted answers now where it was clear that they really shouldn't have accepted the answer -- they wound up doing something else than what my answer said to do, and they said so.
How prevalent has this behavior become since the implementation of the acceptance percentage?
Feature request: Is there a way we could indicate that an accepted answer shouldn't have been accepted so that those interested in data-mining SO can figure out if the acceptance percentage thing is a net win or loss?
EDIT: To clarify: Would it be worthwhile to allow us to mark an accepted answer as "erroneously accepted" so that data mining could tell us if we think we're getting too many of these? I don't think it's something that needs to be shown to all users or anything; perhaps implement it similar to flagging for moderator review. Just have it available in the data dumps so we can try to figure out if it's a net win or not.