Should a self-answer be accepted if the best answer doesn't paint the whole picture?

Question

There recently was a case where OP, after initially accepting my answer, later posted their own, quite extensive answer and consecutively accepted that answer. The new answer references my post and adds some (potentially useful) information, so that's fine with me.

I was in a similar situation a while back, but decided to leave the original answer marked as accepted, as it was the most helpful for me while solving the problem.

This made me wonder if there exist a guidelines on whether accepting a newer, more complete answer is better. As I see it, a comprehensive answer will be more useful to future visitors and should be made easy to find. On the other hand, I prefer to give credit where credit is due, even at the risk of burying important information in a non-accepted self-answer. Is that wrong? How should one handle those situations in general?

Answer 1

The metaSO site has this Q&A on the subject: How does accepting an answer work?.

Within a bullet list there in the accepted answer (ironically) is this:

• The bottom line is that you should accept the answer that you found to be the most helpful to you, personally.

So it's entirely up to the OP to accept whichever and it's just something you have to accept when providing answers on the SE sites. I've had many OP's take what I've written and make it much more specific to their situations in the form of an answer and mark it as accepted.

In looking at the OP's answer vs. yours, their's is the superior answer now, even though they needed yours to develop it. So it makes sense to me that it be the accepted now as well. The OP also put a bold acknowledgment at the top giving you credit for helping to develop their answer.

Always try and keep perspective on your motives on why you're answering. I generally answer for a few reasons.

• help others
• help future me and others
• fill in holes in the collective knowledge on the Internet

It's only human, IMO, to have the response you're having, since I've had this exact same thought creep in from time to time as well. I use the above 3 bullets as my reminder(s) as to why I'm here and what my actual objectives are.

I leave it to you to determine what motivates you, but I'd hope your list is similar to mine 8-).

Answer 2

I agree with part of @slm's answer, specifically:

The metaSO site has this Q&A on the subject: How does accepting an answer work?.

Within a bullet list there in the accepted answer (ironically) is this:

• The bottom line is that you should accept the answer that you found to be the most helpful to you, personally.

So it's entirely up to the OP to accept whichever and it's just something you have to accept when providing answers on the SE sites.

However, there is some loss of data when the situation described by the OP occurs, specifically the history of the original Q&A process (think of a person coming to the post later via a web search or something). As many here are familiar with, this would be undesirable in terms of source control/scm.

I would propose there is a missing feature on the stackExchange family of sites which would help preserve the history of the Q&A process. Perhaps a feature like "Author's Final Comments" which would always be optional (and becomes attached to the OP), but fits the bill for situations like these (also allowing the original answer to be maintained for future reference, among other things, and providing a good spot for the author to add fine or extra detail).

Additionally, I would suggest a time limit on changing accepted answers. Similar to a bounty period, although globally-standardized and much more generous (perhaps around 3 months?), as this would allow sufficient time for better answers to come in. This is similar to already-established closing of old threads in forums. After the period expires, the history is "locked down" and preserved.