It's a well-known problem that new/unregistered users tend to neglect accepting answers. This is usually because they just don't pick up on the concept. If new users wind up sticking around and continuing to participate, they will eventually get clued in and then hopefully go back and accept answers to previous questions.

The real problem occurs with "one-shot" users. We all know the scenario:

  • a 1-rep member-for-today user asks a question on SO, let's call him "timmy."

  • Stack Overflow members help timmy with his question. Comments are exchanged, suggestions thrown out.

  • Eventually somebody nails it on the head, let's call him "Jon Skeet."

  • Up-votes indicate general consensus. timmy emphatically thanks Jon in comments using various emoticons.

  • Then timmy is never to be seen again.

  • Stack Overflow members linger on the question page, and an awkward silence falls. Maybe this time it'll be different they whisper to themselves. Nope - timmy's half across the internet by now.

Now, I'm not pretending this problem hasn't already been discussed (in great detail): Force Accepted Answers on Questions by Inactive Users, but that suggestion is centered on auto-forcing accepted answers based on some algorithms, which would be heavy-handed.

What I'd like to propose is more tempered: the ability to flag an answer for forced acceptance. I realize this would be a pretty radical change, but I think there's merit to it given the issue. The criteria for such a flag would be very strict:

  • Question is some X amount of time old (or OP has not been seen for X amount of time).
  • OP is a user that has never accepted an answer.
  • Flagged answer has at least X upvotes.

And essentially:

  • OP has given some indication of the answer's correctness in the comments.

    These criteria would limit the global effect of this flag. It would not be a site-wide sweep like the auto-force suggestion. Mainly it would be a procedural solution going forward, reserved for situations were there should very obviously be a check mark.

As with the previous suggestion, feel free to add/remove rules to make this viable.

EDIT: I'm getting a mostly negative reaction on this suggestion; fair enough. Before I give up, I thought I'd post some examples of the posts this would be targeting specifically:

Error while calling function

How to get the address of a variable stated in C Macro?

MapKit blue dot not visible when using custom annotations

Android can't send GET request with HttpURLConnection

Is there any way to prevent the trailing ants border effect on links?

How in Linq to entity replace true or false with On or Off?

Here is the query I used to locate these (modified from the query in the linked discussion):

SELECT q.Id as [Post Link] , 
q.Score, users.Id, users.DisplayName, a.Score, q.AcceptedAnswerId
FROM posts q 
INNER JOIN posts a ON (q.Id=a.ParentId)
LEFT JOIN users ON (q.OwnerUserId=users.Id)
INNER JOIN comments c on (a.Id=c.PostId)
WHERE a.Score >= 2 
AND a.Score = ( SELECT MAX(p3.Score) FROM posts p3 WHERE p3.ParentId=q.Id )
AND users.Reputation < 15
AND q.PostTypeId = 1
AND q.AcceptedAnswerId is null
AND datediff(day, users.LastAccessDate, q.LastActivityDate) < 8
AND c.UserId = users.id
AND (c.Text LIKE '%hank%' OR c.Text LIKE '%preciate%' OR c.Text LIKE '%:)%')
GROUP BY q.Id, q.Score, users.Id, users.DisplayName, a.Score, q.AcceptedAnswerId

The query returns several thousand posts, after which some manual filtering is needed to find the ones I'm targeting. The examples were just pulled from the top of the list after looking them over.

Anyway thanks for hearing me out at least.

  • 1
    Okay, I'm seeing some downvotes. Please comment - what is flawed or undesirable about this? Commented Aug 14, 2011 at 11:09
  • 2
    No, there will always be border cases here. What if the user accepts one answer, but leaves 10 unaccepted? And we will have to wait for quite some time, making the "pleasure" of having the answer accepted more of a surprise. Where did those points come from?! It also assumes that the most upvoted answer is the one the OP would have selected. Not always so, IME.
    – Bo Persson
    Commented Aug 14, 2011 at 11:10
  • 3
    possible duplicate of Would it be possible to have a "community accepted" feature?, Moderators accepting answers on user's behalf after a certain time period, and tons of other duplicates. (I can't find as many because I can no longer use the obnoxious inflatable search box.) Commented Aug 14, 2011 at 11:15
  • As to what is flawed about this: it's missing out on the notion that accepted answers can only be marked by the person who asked the question. No exceptions. Commented Aug 14, 2011 at 11:16
  • @Bo Persson please read my suggestion more carefully. 1) This is addressing users who either don't know how to accept or don't care. If they have accepted one answer, that removes them from my target population because they must consciously know about accepting. 2) this isn't about pleasure, this is about administration of a quality wiki-style site. 3)It does not assume anything because of my final requirement: the OP him/herself made comments indicating they would have accepted an answer had they known/cared to. Commented Aug 14, 2011 at 11:18
  • @Cody I disagree that these are duplicates. The first one isn't even addressing the same issue. What the second one is missing is my final requirement - that the OP clearly indicated the answer was correct before leaving. Commented Aug 14, 2011 at 11:23
  • 2
    If the asker leaves a comment indicating that it is the correct answer, then just ask them to accept it. Problem solved, no feature request needed. Commented Aug 14, 2011 at 11:24
  • Similar to meta.stackexchange.com/questions/26639/… which covers the point of users indicating an answer helped them too. Commented Aug 14, 2011 at 11:49
  • @Cody see my edit for examples. If you're still not convinced I'll hand it to you. Commented Aug 14, 2011 at 12:51
  • Just because people say "thanks" doesn't necessarily mean that they think the answer is worth accepting. I've thanked people for taking the time to answer my questions, but I'm still holding out for a better answer, one that addresses the rest of my concerns. Your first example is particularly telling here. The first comment says "thanks 4 responce but i am calling static functin from non-class member function." The "but" indicates to me that it doesn't solve his problem yet. Commented Aug 14, 2011 at 13:41
  • And then later, he says "i will mark after accepted ans.". I'm not really sure what this says, but I think it says that he'll mark an answer as accepted when he gets one that solves the problem discussed in the previous comment. Either way, the difficulty in assessing people's intentions is why this feature request doesn't work. Accepted answers are reserved for the asker of a question only. If you think those answers are good and should be accepted, then you should just upvote them. That's how the community "accepts" answers, anyway. Commented Aug 14, 2011 at 13:41
  • Thanks to everyone for the reasonable responses. I know people have tried to address this issue endlessly on meta - I thought my suggestion to use flagging and strict requirements would be realistic enough to gain favor. However what you've taught me is that the green check on an answer is not as essential to the site as I thought it was, and that in their absence upvotes are enough to make a post helpful to future readers. Commented Aug 14, 2011 at 17:44
  • Several years later, dozens of other users (including me) are still trying to debate this issue. I really hope SO staff will consider the idea eventually. meta.stackoverflow.com/q/354584/534406 Commented Aug 6, 2017 at 22:01

3 Answers 3


I'm against this and for whatever reason anything to do with the accepted answer status because I think in the end the accepted answer status is not that important in the way stack exchange is designed to work.

the whole point of SE is to make the internet better and they've set out to do that with a system that allows easy access to questions and answers not just for now, but for future visitors who can equally contribute and improve and that is the most important part.

whether "timmy" was helped is less important than whether future visits to that question are helpful (imho). so my point is that the accepted answer in a way is very localised...and will regularly be useless to future readers. and this is where votes and editing are so powerful...to help in improving and keeping useful information through time.

is this issue any different to accepted answers that are wrong but still accepted? if anything, these are more harmful and more of a problem...are we going to create a feature of force unaccepting?

I'm happy to be convinced otherwise...I just see this area as not really being a problem.

  • Don't worry, I've been convinced otherwise. Thanks for the response. Commented Aug 14, 2011 at 18:21

The answer the users who asked the question would accept is only know from those users, who would accept the answer that would help them more; the fact the answer helped the users who asked the question doesn't automatically mean the same answer would be the most helpful answer for another user.

When the OP doesn't select an accepted answer, the community give an indication of its accepted answer: It is the one that received more votes than other answers.


As a comment to the new links in the question I can add a few counter examples from personal experience during the last week.

Here is one example

using compile time constant throws error

where the poster selected one answer (mine :-) even though it didn't get any upvotes at all, and is technically inferior to the other answers. However, the person asking the question obviously thought that - this time - a simple explanation was what he wanted.

Another example is

Why declare a struct that only contains an array in C?

where a similarly quick answer got lots of upvotes, and a personal "Thank you!" comment from the person asking the question. I even got an entirely undeserved "Nice Answer" badge for that.

However, some of the other answers have an even more hilarious number of upvotes, and a "Thank you!" comment from the questioner. Every single one of them!

But no selected answer.

So, now that we haven't heard from that person in a week, how do we select the forced answer?

We just cannot!

  • Nice counter-examples, especially the second one. Thanks! Commented Aug 14, 2011 at 18:19

Not the answer you're looking for? Browse other questions tagged .