23

Possible Duplicate:
Limiting Self-Answers

It appears to be a common mistake of the new-user to treat Stack Overflow as if it were a forum, posting the initial question, and then adding follow-up posts below as "answers." To my knowledge, I've not seen this amongst more seasoned users, which makes me wonder if a user should have to meet a particular rep-requirement before being permitted to "answer" his/her own question.

While it's true that the user may have the genuine capacity to answer their own questions properly, it seems safer to require the rep, just as sufficient rep is required for doing many other things new users may genuinely have the capacity to do, downvote, edit, etc.

It is my sincere opinion that this would cut down on innapropriate noise in the provided answers.

11
  • 1
    Duplicate - meta.stackexchange.com/questions/6593/limiting-self-answers
    – ChrisF Mod
    Jan 19, 2010 at 23:27
  • 3
    ChrisF, I don't see how that is a dupe... I'm asking that you have a rep-requirement to even answer in the first place. Whether there's a limit is a different issue altogether.
    – Sampson
    Jan 19, 2010 at 23:37
  • Good idea, Jonathan
    – Vinko Vrsalovic StaffMod
    Jan 19, 2010 at 23:39
  • 2
    Yah, I agree with Jonathan on this one. 6593 does not mention anything about a rep limit, so I think this request is ok.
    – Troggy
    Jan 19, 2010 at 23:42
  • It's not a real dupe, but both requests do solve the same problem.
    – fretje
    Jan 19, 2010 at 23:43
  • 2
    @fretje: How does the other question solve my problem? My problem is not how many times a new user can post a non-answer as an "answer." My problem is that they can do it just once :)
    – Sampson
    Jan 19, 2010 at 23:54
  • 1
    @Jonathan: No, the problem is that noise gets posted all together. It's not really about the amount. Btw, I find your solution better than the other one.
    – fretje
    Jan 20, 2010 at 0:03
  • @fretje: It seems more likely that noise will be posted by new users as opposed to those who understand the system. I don't recall ever seeing anybody with a couple thousand rep post non-answers as "answers."
    – Sampson
    Jan 20, 2010 at 0:07
  • @Jonathan: Exactly. That's why I said your solution is better.
    – fretje
    Jan 20, 2010 at 0:44
  • 5
    Personally I feel more should simply be done to educate the user. Why search for programmatic solutions to human problems? Maybe a bright pink flashing unicorn that says "Are you ANSWERING your question, or PROVIDING AN UPDATE?" and if you click on the waffle it takes you to the Update your Question page when you focus on the "Answer your Question" textbox and you're < 1k rep Jan 20, 2010 at 1:20
  • 1
    Whether this is a dupe hinges on what the previous poster intended by "limit". If "limit" == "prevent", then they are very similar.
    – Ether
    Jan 20, 2010 at 1:29

7 Answers 7

8

A tempting proposal

On the one hand, this is very appealing - those answers are annoying. Of course, so are non-answer answers posted by such users on questions they don't own, which is at least as common and something this will do nothing about.

On the other hand, this locks new users out of a valid and accepted means of using the site. It breaks a SO feature for these users simply because other users have misused it. As Pollyanna also notes, SO already tries to discourage users from answering their own questions, suggesting the use of comments instead - of course, some ignore this...

Proof of the problem?

It would be nice to see some evidence that this solution has at least a chance at working; some data as to which rep levels have the highest occurrence of self-answering would go a long way toward demonstrating the scope of the problem and provide a good tool for choosing the right rep-level for the cutoff (my gut feeling is that even 100 points is unnecessarily high).

Encouraging desired behavior

Normally on SO, comments are intentionally discouraged in favor of posting answers: when viewing questions from other users, a SO user will see a great big big "Your Answer" entry field and only tiny gray "add comment" links. And new users don't see the "add comment" links at all on answers to questions they don't own...

So perhaps a better solution to self-answer non-answers might be to similarly encourage new users to post comments on the answers to their own questions by showing the comment entry fields instead of tiny gray links, while keeping the self-answer entry field hidden:

Post a comment to respond to an answer

6
  • @Shog9 I don't think this "locks new users out of a valid and accepted means of using the site". I think it postpones it, which is the case for voting, editing, etc. Sure, it's a valid use, but new users shouldn't be able to do it carte blanche.
    – Sampson
    Jan 21, 2010 at 0:28
  • They cannot today do it carte blanche - there are rate limits built in to the system for new users. They can, however, do it - you're talking about shutting that door. So it deserves some consideration... Have you given any thought to a means of measuring the rep-level at which an average user might be reasonably expected to understand the proper use of answers?
    – Shog9
    Jan 21, 2010 at 1:03
  • Shog9, you're correct, I'm not interested in limiting this activity, but barring it altogether, until they've got sufficient rep. As for how much rep would be reasonable, I don't know, that would be a different question altogether IMHO.
    – Sampson
    Jan 21, 2010 at 5:30
  • we already kind of do this; see Pollyana's answer. Your idea of expanding comments by default for new users on their own questions is a good one, but might be a PITA to implement. Jan 21, 2010 at 9:10
  • 4
    update -- we now show the comment fields expanded by default for new users on answers to questions they own. Will be deployed Jan 21 around 6 PM Pacific. Jan 21, 2010 at 12:51
  • @JeffAtwood That seems like a pretty good step towards eliminating this problem.
    – Sampson
    Jan 21, 2010 at 22:13
4

No, we already have an information box that asks if they're sure they want to add and answer, and suggests that they edit the question if they have more information on the question.

Are you sure you want to answer your own question?

Note:

  1. The standard answer editor does NOT appear by default for post owners. Instead there is a single "Answer Your Question" button.

  2. Clicking the "Answer Your Question" button nags you:

    Are you sure you want to answer your own question? If you're responding to answers left on your question, use the comments link under each answer

Quite frankly, I'd like to encourage new users to answer their own questions when they've solved their problem, if no other answer is appropriate.

Locking them out of a basic feature of the site will only result in more questions edited with "Solved: I did x, y, and z" and them never accepting another answer.

1
2

Personally, I've asked many and answered many of my own questions. I asked initially looking for a quick response, but then as I awaiting the response either had one of those coming to the code moments or a colleague was able to help out (or google...).

Adding the answer at that time not only shows the correct way to resolve the issue, but also prevents folks from wasting time. I'm concerned that not allowing the answer will actually keep the noise high as people try to answer questions that folks are really no longer needing an answer to.

We will then see even a larger number of questions here on meta about acceptance rates like they actually are worthy of discussing on meta in the first place. Don't get me wrong, I understand the sentiment, but the unintended consequences will probably be a larger issue...

7
  • 3
    Yes, but you probably used the site for awhile before you began answering your own questions.
    – user102937
    Jan 19, 2010 at 23:24
  • 2
    If it's a small enough rep needed, then any user will be able to get to it quick enough if they feel the need. It'll just keep the anonymous 1s from trying to use the site like a forum. Jan 19, 2010 at 23:28
  • What difference does it make? New or old the results are still the same...
    – RSolberg
    Jan 19, 2010 at 23:53
  • 3
    @Chester: I'm talking about users who post "answers" which say, "Thanks guys! I finally got it working. Bye!" or "What do you mean? (referring to some non-stated previous answer) I don't have css on this. What can I do?"
    – Sampson
    Jan 19, 2010 at 23:55
  • 2
    I've done self-answers also, and I think it's a good thing. We're just trying to cut down on the noise from users who don't understand how SO works yet. Jan 19, 2010 at 23:58
  • 3
    @LanceRoberts: Correct, self-answers aren't the problem. Noise, posted as a self-answer is the problem.
    – Sampson
    Jan 20, 2010 at 0:07
  • @Jonathan I think that was what @Lance was saying, too. Jan 20, 2010 at 10:21
1

I agree, great idea.

I'd propose 100 rep, this would give them time to figure the site mechanics out.

1

This conflicts a little with the answers of Should I answer my own question, or not?

I think it would be better to just show an advice to users with less than a given amount of rep if they try to post an answer to they questions.
Maybe the advice should be presented for the first 3 answers, regardless which question. I've seen a couple of "follow-up posts" this weeks, like "Thank's, that helped at lot" or "Just what I was looking for".

EDIT:
as I just learned, the warning is already there :--)

4
  • See my answer to the "duplicate" question meta.stackexchange.com/questions/6593/limiting-self-answers. You already have to go through an "Are you sure?" dialog when posting an answer to your own question.
    – ChrisF Mod
    Jan 20, 2010 at 10:57
  • @ChrisF - haven't seen it, not used to answer my (only) question :-)
    – user141148
    Jan 20, 2010 at 15:43
  • See the screen shot in @Pollyanna's answer for what I'm talking about in that answer.
    – ChrisF Mod
    Jan 21, 2010 at 11:15
  • @ChrisF: I actually have tested it myself :-) thanks...
    – user141148
    Jan 21, 2010 at 13:35
0

Agree. Suggest rep of 300 required to demonstrate that user has sufficient understanding of SO to answer own question.

My thinking is that it is unlikely that user will obtain rep of 300 on the same day as they join. -- if they do achieve such a feat then perhaps they ought to get a badge as well...

2
  • 7
    They will get a badge - the [Mortarboard] for hitting the rep cap.
    – ChrisF Mod
    Jan 19, 2010 at 23:30
  • 3
    300 is too much
    – sth
    Jan 20, 2010 at 3:52
0

I think the problem behaviour comes from the number of forums out there that use the answer section as a discussion. The MSDN forums do it this way, for example.

So the asker replies to an answer by posting another answer, and the discussion goes down the page, and then disrupted by an upvote, and so on.

When the asker actually discovers the answer themselves and posts the answer, that's not really a problem. So perhaps people with sufficient reputation should be able to move an answer from being an answer into being a comment on a different answer, or being able to move an answer into the question itself?

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