New phenomenon: Rage Unaccepting

We should wait and see if "rage unaccepting" becomes a systemic issue.

And we should be very careful before we conclude something constitutes "rage unaccepting."

People who are unsatisfied with existing answers should unaccept them.

One user appears to be unaccepting vindictively, and we're considering a policy change?

This doesn't indicate a trend and it's silly to think something must be done about this, unless there is more information to indicate this is becoming common.

Furthermore, of those three posts (and the unaccepted meta post), it's not clear that any of them are really rage unaccepts.

I don't see any strong indication in any of those questions that the OP was actually satisfied by the answers. It seems perfectly plausible--and, by Occam's razor, pretty likely--that s/he had felt pushed into accepting answers. Now that they don't feel pushed into accepting answers dishonestly, they may simply be doing the thing that is both good for them and honorable toward the community: unaccepting answers they should never have accepted in the first place.

I have seen this happen on Ask Ubuntu, where users admit to accepting answers (often total crap answers) just to make their accept rate increase. As the social flogging for low or medium accept rates let up, they unaccepted answers they felt they'd been bullied into accepting originally. (And the social pressure to accept answers on Ask Ubuntu was considerably lighter than it sometimes was on Stack Overflow.) Unaccepting those answers was the right thing for them to do.

Reputation we earn from wrongful accepts is dishonest reputation. It doesn't reflect badly on us (because it is not usually our fault). But it harms us. Like serial upvoting, it dilutes the real-world meaning of reputation.

From time to time, someone accepts an answer of mine when it's quite clear it didn't solve their problem. For example, they might accept but comment about how it didn't work. When this happens, I explain that they should unaccept the answer. This is not a selfless act; it is not motivated just by my concern for them. It is both practically and morally bad for me to have reputation that comes from baseless acceptances.

Losing reputation from such unacceptances is like the sting of cool water on sunburned skin. The injury was already there, and as it starts to be fixed, we become aware of it.

If people really are "rage unaccepting," then that is (at least a little) bad and harmful.

But what we have is one possible rage unaccepter.

Now, let's suppose that the answers this user unaccepted helped the user and should have remained accepted.

Does that mean the user is malicious? Or that s/he has just judged poorly? Or that s/he doesn't understand how accepting answers is supposed to work because s/he never learned to accept answers properly?

Anecdata: After this change went live, I personally went through all my own questions and unaccepted a substantial number of answers. Why?

I wasn't angry. But they were answers I had accepted but with which I was never quite satisfied - they were the best I was going to get at the time and the displayed accept rate irked me, so I accepted the 'best' answer - even when the best wasn't quite good enough for my own standards. This change made me feel free to apply the accept when I believed it was earned - and to do so retroactively. I think jumping to the conclusion that his motives were malicious - especially when the change to accept rate display is so recent - is probably a bit unfair.

TL;DR: No rage is indicated. Lower accept rates may be appearing as advantageous, since January 2013. People prefer interaction to tumbleweed.

As the research recorded here in the other answers shows, there is no indication of rage, revenge, or malicious intent here. The user appears to be a cooperating citizen who distinguishes good and bad content and is willing to take actions to provide feedback in some cases. (Look at his/her voting structure.)

I see this incident as confirming a theory I had already when learning that accept rate would no longer be displayed, predicting some drop in total acceptance rates.

The (only?) reason of asking a question is getting people see it and provide relevant information on it (answers, comments, votes, votes and comments on various answers, views). The frequency of all such events goes down when the question is considered to have an accepted answer.

Supposing that the user does not feel bored by the prospect of getting even more interaction related to his/her questions, we are seeing prima facie economic behavior here.

I know that unaccepting also entails the loss of 2 points of reputation per answer, which might kind of compensate for most people, not speaking of all kinds of ethical, aesthetical, or interaction quality reasons against completely stopping to accept.

• This user may not know that they are also losing reputation
• This user may not care that they are also losing reputation
• This user may expect that the higher exposure of their questions will lead to a mixed voting traffic more than compensating for this loss in the long run.

Who knows.

[Edit: only now I noticed that user's comment "Please consider that as self-vandalism." Good, this won't be a very frequent motivation here.]

Maybe the January change to displaying accept rates was a little overdone.

After enough time passes, we should check the database to see how widespread this slightly damaging behavior is and see whether there we have a problem with making the aggregated accept rate completely unaccessible to other users, or not.

My solution to prevent this would be to only allow switching the acceptance to a different answer, and not removing it entirely after a certain time.

Being able to change the accepted answer even after a long time is important, but I don't see a realistic use case for unaccepting an answer completely after a few months or so.

If you find an error much later, you could still work around this by posting your corrected solution as an answer and self-accept that one.

I'm not yet convinced that this is a problem that happens often enough to warrant such a solution, this is the only case I heard about so for.