What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 131 Stack Exchange communities.

I'm looking at this from the standpoint that StackOverflow is not just a means to get an answer to a question (i.e. a forum), but so much more. Isn't becoming a database of information an additional goal of SO?

I've noticed being a regular user for a few months now is that questions that have received answers are buried underneath others that haven't received enough attention. I feel that some of my questions haven't ever been looked at once they were answered, even though there might have been scope for improvement. Users who answer tend to look for unanswered questions since they have more incentive to get reputation this way. Many questions end up with one or two answers and are then shelved with no further improvements.

Obviously, all questions on SO can't cover revolutionary, ground-breaking areas, and most questions tend to be specific to the asker. For questions that deal with instructions/steps on how to perform a task, such as this one, I'd say this blog did a much better job of getting a complete, refined answer than SO (or any stack exchange site) would have ever have done.

My idea is to provide some means for the asker to request improvements to existing answers, and users get alerts to these kinds of questions with existing answers. The feature would be something like a bounty, except that it shouldn't cost the asker reputation. The main aim should be to provide the community with a fine-tuned, clear answer that has been built with multiple improvements.

Does something like this already exist?

EDIT: Bounties are fine when the asker knows that his answer can be improved. What about when the answer is posted, the asker is happy, and neither of them realizes that a better solution exists? Everyone loses out here.

share|improve this question
    
If it wouldn't cost reputation, everybody would do it. And once again, all posts would be buried in the masses... –  sudo May 1 '12 at 14:55
    
@Dennis I agree, costing reputation is a way to ensure that not everyone does it. But the main problem is not how much reputation is lost, it's that existing answers are considered done and dusted in many cases. Only the really popular, broad-area questions are really improved upon. The specific ones tend to be neglected –  prrao May 1 '12 at 14:58
    
I tried to answer, then I realised you said like bounties, but different. Sorry! –  user142852 May 1 '12 at 15:01
    
@Ninefingers Thanks, I just saw your answer :). Bounties are fine when the asker knows that his answer can be improved. What about when the answer is posted, the asker is happy, and neither of them realizes that a better solution exists? How would bounties help in this case? –  prrao May 1 '12 at 15:02
1  
@prrao they wouldn't, hence me removing the answer :) –  user142852 May 1 '12 at 15:03
    
I'm wondering if my question can be made clearer. If it can, please let me know or feel free to edit! –  prrao May 1 '12 at 15:03
6  
"" What about when the answer is posted, the asker is happy, and neither of them realizes that a better solution exists Go to Philosophy? –  Won't May 1 '12 at 15:38
1  
This entire question is based on the false premise that the answerer knows best. If your answer didn't get accepted, it wasn't the solution the asker liked the best. If your answer didn't get upvoted, it didn't help the community. Accepting and voting are different concepts which drive different behaviors and connote different meanings. –  user7116 May 1 '12 at 16:35

2 Answers 2

up vote 7 down vote accepted

One problem I have is that users accept answers way too quickly, and this is a disincentive for folks that are willing to come up with a better answer (which usually takes more time). When folks accept answers too early, the rest of the community assumes that they've solved their problem and won't bother to come back and look at new answers.

At odds with this are the people who beg and plead with users to up-vote and accept their answer within minutes of posting it.

These two things combine to make it much more likely that the first decent answer is accepted and that, even if better answers come along later, they are largely ignored. I think that's okay in general, even in the case where the OP ignores later answers, because hopefully the general public consuming the site as a dictionary of knowledge rather than a help desk will be able to glean information from all of the answers, regardless of whether they were first, accepted, have many up-votes, etc. I still get up-votes on questions that are years old, so I know they still benefit other users even though they weren't always the OP's accepted answer.

As a very recent example of this, this morning an answer to this question was accepted within 10 minutes (sorry, must have been 15 minutes, the "1 hour ago" always throws me off), even though the selected answer didn't actually answer the question completely. I've added a more complete answer (and commented on the accepted answer that it missed part of the question) but since the answer has been accepted I'm not sure how many of the original stakeholders will even notice.

I'm wondering if the threshold for allowing a user to accept an answer is too low, and if we should do a better job of balancing that between not having tons of unanswered questions and allowing more time for quality questions to be submitted. I'm sure this has been discussed at length here before, but I'm far too lazy to search for the threads. The one about time zones is relevant too because the person with the best knowledge about your problem may be 18 hours behind you, and you're missing out on a great answer because you've accepted a mediocre one (and they're not going to bother now).

share|improve this answer
    
So are you talking about a 'latency period' before a user's answer can be accepted? Has such a thing been proposed? –  prrao May 1 '12 at 15:18
    
@prrao yes, meta.stackexchange.com/questions/105695/… - wasn't met with very warm reception. –  Aaron Bertrand May 1 '12 at 15:21
    
Well, let me re-phrase - the powers that be believe that 15 minutes is long enough. Which means that the accepted answer is more likely to be someone in your time zone (or at least that keeps the same hours as you) and that types quickly. :-) Personally, I don't think 15 minutes is long enough, and I don't buy that making someone wait before accepting an answer is such an inconvenience to them. JMHO. –  Aaron Bertrand May 1 '12 at 15:24
    
It's really sad that people believe the answer they first see is the best one and that they can figure out the rest. I came across a blog that talked about a Windows batch file. People continually looked at it and suggested improvements –  prrao May 1 '12 at 15:27
1  
I still update blog posts I wrote 5 years ago, so I am the last person you'll ever convince that the first thing you bang out within 15 minutes is going to be top-notch. Unless you're Jon Skeet, I guess. –  Aaron Bertrand May 1 '12 at 15:29
1  
@sixlettervariables -1 What about people who don't happen to see a question the moment it's posted, but have a better solution? The best way to have community votes decide which answer is best would be to have a latency period before an answer is accepted. Currently, what tends to happen is that new answers are upvoted by users who happen to be on SO at that time, and users in another timezone have their answer upvotes pushed back –  prrao May 1 '12 at 16:13
2  
@sixlettervariables I don't think that's the point. The point is that requiring more than 15 minutes can't possibly yield worse answers. And while you might go ahead and answer a question that already has an accepted answer, a lot don't even get looked at, never mind further answers. –  Aaron Bertrand May 1 '12 at 16:16
1  
@sixlettervariables Why lose out on a better way to do something, if it exists?. Currently I believe the main problem with answers being buried in the masses and not improved is that users accept an answer, and others who might have answered with improvements don't even bother because they think the user accepted an answer and its not worth their time –  prrao May 1 '12 at 16:19
1  
While delaying their ability to accept immediately doesn't really inconvenience them, it can help all of the other users who might benefit from a variety of answers to the same question. If more users have incentive to answer that question, they may not be any more appreciated by the OP, but the content will certainly be richer for all other/future readers. –  Aaron Bertrand May 1 '12 at 16:23
1  
@sixlettervariables I don't know about you, but if I asked a question and received an answer, I'd be delighted if someone voluntarily came forward and gave me a better answer. There's just not enough incentive to get this kind of response with the current setup. Isn't that the whole point of SE? To be much more than just some online forum where I ask a question, get an answer and move on? –  prrao May 1 '12 at 16:25
1  
@AaronBertrand: it actually inconveniences answerers. Accept rates for drive-by askers will go down as they will leave with only a "thanks this helped" comment. Besides, the system already alerts users when new answers are posted and they are free to accept those when they appear. I also believe it is in the asker's best interest to accept the quickest complete solution and in the community's best interest to upvote the most complete solution. –  user7116 May 1 '12 at 16:26
1  
I found out about SO through a million google searches. That's where I'm assuming most of SO's traffic comes from. So that's the major point here. Others benefiting, not just the guy who wants his problem solved –  prrao May 1 '12 at 16:28
1  
@sixlettervariables the challenge is, like I said, that I am far less likely to even open a question that already has an accepted answer. If I took the time to open every question with an accepted answer to validate that the answer was in fact good, I'd never answer any questions, sorry. –  Aaron Bertrand May 1 '12 at 16:44
1  
@prrao: I don't ask a lot of questions, and I accept the first answer which solves my problem. I upvote any answer that helps or is also right. I don't accept an answer just to improve my accept rate and I've applied bounties when I want a better or different answer. If it takes more than 15 minutes to solve my problem, I wait for answers that solve my problem. –  user7116 May 1 '12 at 16:57
1  
Why are you so confident that the person who knows how to solve your problem will be able to do it in 15 minutes? Are they even online right now? Are they monitoring your tag? Are they busy doing something else, maybe answering another question? Maybe it's true for you that you always get a perfectly acceptable answer worthy of acceptance in the first 15 minutes, but I really find it hard to believe that on average that happens for all questions. I'm not trying to convince you, because obviously you think the current limit is fine. Just expressing my viewpoint (which, by def, isn't wrong). –  Aaron Bertrand May 1 '12 at 17:02

I'm looking at this from the standpoint that StackOverflow is not just a means to get an answer to a question (i.e. a forum), but so much more. Isn't becoming a database of information an additional goal of SO?

Yes. Ideally, every question should have a perfect answer. Quite impossible, but ideal.

I've noticed being a regular user for a few months now is that questions that have received answers are buried underneath others that haven't received enough attention. I feel that some of my questions haven't ever been looked at once they were answered, even though there might have been scope for improvement.

As long as none of the answers received any upvotes, your question will remain in the Unanswered tab. Also, the Community user will bump it from time to time to the front page.

Users who answer tend to look for unanswered questions since they have more incentive to get reputation this way. Many questions end up with one or two answers and are then shelved with no further improvements.

That's natural. A question with an upvoted answer is considered answered. Gaining reputation shouldn't be the ultimate goal, but it's nice to see those little numbers rise...

But even without considering the rep, time is just better spent answering questions that haven't been answered yet. The other extreme would be having the community continuously improving the answers of the same questions, neglecting new (unanswered) question. That would be much worse!

My idea is to provide some means for the asker to request improvements to existing answers, and users get alerts to these kinds of questions with existing answers. The feature would be something like a bounty, except that it shouldn't cost the asker reputation. The main aim should be to provide the community with a fine-tuned, clear answer that has been built with multiple improvements.

As I said in the comments, I don't see this happening. If it wouldn't cost reputation, everybody would do it. And once again, all posts would be buried in the masses...

I agree, costing reputation is a way to ensure that not everyone does it. But the main problem is not how much reputation is lost, it's that existing answers are considered done and dusted in many cases. Only the really popular, broad-area questions are really improved upon. The specific ones tend to be neglected.

Questions that are too specific should get closed as too localized.

For all other questions, well... there are currently around 3,000,000 questions on SO. It's impossible to dedicate time to all questions. You have to draw the line somewhere.

Also, we have to consider the rate at which questions are asked. Taking the 500,000 newest question as a sample, SO receives 5,000 question per day. Even without an answer, most users won't even get to see a question after it's a few hours old...

Bounties are fine when the asker knows that his answer can be improved. What about when the answer is posted, the asker is happy, and neither of them realizes that a better solution exists? Everyone loses out here.

Not true. You can put a bounty on any question on SO, not just your own.

share|improve this answer
1  
Sure, there are 3 million questions, but how many of those are in tags/areas where you are comfortable answering? Even the volume in the sql-server* tags is overwhelming, but I don't focus on only unanswered questions in the strict sense (meaning no answer with an up-vote). Depending on volume, time of day and schedule, I often look at questions that have answers (with or without up-votes). But I very rarely look at questions with accepted answers, unless the title is very intriguing. –  Aaron Bertrand May 1 '12 at 15:18
1  
Exactly my point. Depending on how busy I am, I end up looking at answered questions (yet unaccepted), but would rarely end up trying to improve an accepted solution that I'm not sure will ever be looked at. Ten months down the line, some other poor soul facing a similar problem might end up missing out on what could have been a vastly improved, complete answer –  prrao May 1 '12 at 15:22
1  
@AaronBertrand: There are almost 500,000 questions in my three favorite tags, 300 of which were posted in the last 6 hours. If without an answer, the chances that I even see a slightly older question are small. –  sudo May 1 '12 at 15:24
    
Maybe you need to narrow your filtering. :-) –  Aaron Bertrand May 1 '12 at 15:26
    
@AaronBertrand: That's just [php], [javascript] and [c]. There isn't much activity in [c], and the first two aren't even the most popular tags. –  sudo May 1 '12 at 15:30
    
@Dennis I understand there are a lot of questions, but that doesn't make quantity > quality. So what if there are 10,000,000 questions here with 99% of them having crappy answers? Does that meld with the SE goal? –  Aaron Bertrand May 1 '12 at 15:30
    
@AaronBertrand: I'm not saying the current situation is ideal. But the solution the OP proposes (bounties that don't cost any rep) won't make it better. And I can't think of any change myself that could make question with existing but less then perfect answers stick out. –  sudo May 1 '12 at 15:35
1  
Agreed, I don't think a "free bounty" is better. But I think there could be improvements in (a) the time required before I can accept an answer, and (b) the criteria that make an answer show up under "unanswered." My interpretation of "unanswered" is not that a question hasn't received at least one answer with an upvote, it's that the question does not have an accepted answer. Maybe there needs to be two ways to express those. I'm far more interested in the union than just the former. –  Aaron Bertrand May 1 '12 at 15:43
    
@AaronBertrand I think my question's been answered. I agree, the free bounty idea is not the best. Is there an existing question similar to what you suggested, and if not, is there a way to open this up as a separate question? Thanks to you both! –  prrao May 1 '12 at 15:46
    
    
@Dennis Sorry I meant the second option: Interpretation of 'unanswered'. My searches are giving ambiguous results –  prrao May 1 '12 at 15:47
    
@Dennis I pointed to that question too, unfortunately it's long been forgotten because the "answer" was that there was a 15-minute limit instituted, the OP was just not looking at recent questions. I think a new question would be, "Can we increase the 15-minute limit?" –  Aaron Bertrand May 1 '12 at 15:50
    
@AaronBertrand: I think a distinction by accepted answer would have too many false positives. There are thousands of questions with a perfectly valid answer that wasn't accepted. –  sudo May 1 '12 at 15:52
    
@AaronBertrand I was also interested in the second point you made. A large part of the problem is that questions without accepted answers (but with answers nonetheless) don't show up in 'unanswered' searches. This must surely be having an impact on the quality of answers in general –  prrao May 1 '12 at 15:52
1  
@Dennis there are also thousands of questions with poor answers that happen to be accepted, and thousands of crappy questions with no answers because they're crappy questions and haven't yet been closed - this was discussed recently here because on some low-volume tags crappy questions stay open a lot longer. What filtering makes sense to you depends on what type of questions you are trying to answer. Note: our filtering preferences may not be the same. –  Aaron Bertrand May 1 '12 at 15:53

You must log in to answer this question.

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