54

As my friends and family have discovered to my chagrin, I'm pretty good at solving general computer problems. When I turn to Super User for help solving a problem of my own, therefore, it usually means I've ruled out all the obvious possible solutions as well as several of the more non-intuitive ones. In short, any question I ask is likely to be hard, and may not even be answerable: as complex as computers are, there's a halfway decent chance that the problem I'm having is due to a confluence of factors unique to my system that I couldn't even begin to guess at--strange registry entries, a hardware component with an otherwise-undetectable flaw, the phases of the moon, etc. So a lot of the time, the chance that someone else is going to be able to answer my question correctly may be pretty slim. Fine. I accept that.

What's not cool is that by asking such questions, I get a low accept rate that I have to wear as a scarlet letter and which may discourage people from even trying to help me at all. I just accepted an answer that, while informative and well-written, isn't really what I was looking for, and I probably won't implement it. Why? Because my accept rate is appallingly low and I have to raise it if I want anyone to answer my questions in the future. Surely this isn't the kind of behavior we want to incentivize?

Improve this question
3
  • 26
    +1 I agree wholeheartedly. I'm of the opinion that, if I don't like any of the answers that were given for one of my questions, why should I be forced to accept one of them anyway? That's the main reason why I'll never put a bounty on any of my questions.
    – RobH
    Aug 26, 2009 at 2:02
  • 2
    +1 for "As my friends and family have discovered to my chagrin" - yep, know all about that one :)
    – Tom Zych
    Sep 7, 2011 at 14:52
  • 1
    +1 I'm not smart enough to have this problem, but I really hope something like MPelletier's answer gets implemented
    – Adam Rackis
    Sep 12, 2011 at 15:42

10 Answers 10

Reset to default
14

Perhaps there should be a way to escape this. Could an "unresolved" status be provided? If after n days, the question has answers, but none satisfy the asker, the asker can then set the question to "unresolved", meaning that the question is still open to new answers, but at the same time is not "well answered but abandoned".

That way, we would distinguish between your two origins for unaccepted questions:

  1. give me teh codez types, where acceptance is accessory (and those are the one we want to discourage)
  2. genuine "no good answer yet", by a caring, contributing community member.

The accept rate would then be calculated as:

(accepted + unresolved) / asked

accepted / (asked - unresolved)

instead of

accepted / asked
Improve this answer
4
  • The problem I see here is that the question of whether an OP is a “caring, contributing community member” (and thus the question of whether they should have access to such an “unresolved” state) is something that should be decided by the community (is reputation a sufficient proxy?). Or maybe it is the viewers (potential question answerers) that should be able to nominate a question for potential “unresolved”-ness. Enough nominations by qualified viewers and the OP is able to mark a question as “unresolved”. Do new answers reset this? Raise the bar for nominations?
    – Chris Johnsen
    Apr 12, 2010 at 20:08
  • This should be discussed properly. I'd start a thread on that specifically, but the dupe police will get me once again and point here. SIGH My premise though, is that a genuine asker will take the time to check "unresolve". The careless one will not. Maybe.
    – MPelletier
    Apr 12, 2010 at 21:40
  • @Chris. Furthermore, the idea of "unresolved" is not to deny any one of the ever so important acceptance of their answer (and I cannot fathom it being used for that, although this is the Internet...). The goal of "unresolved" would be to guard oneself against a low accept rate due to asking tough questions.
    – MPelletier
    Apr 13, 2010 at 0:29
  • @Charles I can't believe I did not see that. Good catch!
    – MPelletier
    Sep 12, 2011 at 14:54
7

If you feel that none of the answers solve your problem, then you should go back and answer it yourself with what you did to move past the problem, and then accept your own answer.

If someone came close to a solution, then accept their answer and edit it or comment on it with why it wasn't perfect, but how it helped resolve your issue.

If you are asking subjective questions (ie, you are looking for an "ideal" or "best" solution) that bring a lot of work arounds, but no answer is 'elegant enough' for you, then yes - people are justified in looking at your future questions and moving past them because they can see they have a low chance of meeting your needs.

If you truly have not moved past the problem (ie, no workaround), no one comes close to solving it or suggesting a reasonable alternative, and it's a concrete, objective question, then I suggest the following if you absolutely insist on getting a higher accept rate:

Post your own answer:

So far I've found out the following, and here are a few leads, but the main point of the problem is not resolved. If you have better luck than I, please post more answers.

And accept it. It is the best answer available so far, so there's no deception or ethical question about doing so. Keep in mind that by doing so, though, you no longer show up in any unanswered lists, and it's unlikely your question will receive an answer later on.

It's important to note that if the question has no long term value (ie, too localized) then you might seriously consider deleting it.

Improve this answer
7
  • 1
    The self-answer is a valid approach, but very often the problem will be circumvented, so much that it goes out and beyond the original problem, thus making the solution not always interesting (i.e. not general enough) for future reference.
    – MPelletier
    Apr 13, 2010 at 1:01
  • @MPelletier - Yes, but it's still a workaround, and was, in fact, the implemented answer to the problem, even if it's not generally applicable.
    – Pollyanna
    Apr 13, 2010 at 1:19
  • Come to think of it, can you confirm that deleting a question (answered or not) will impact on one's accept rate?
    – MPelletier
    Apr 13, 2010 at 2:05
  • @MPelletier - I just deleted 3 unanswered questions of my 54 questions. My accept rate is still 61%, so if it changes, it does so slowly. I expect that it's recalculated and cached periodically, so probably tomorrow we'll see the results of my deletes, which should bring me up to about 64% (give or take 1% depending on whether they round or truncate - I'm guessing they truncate)
    – Pollyanna
    Apr 13, 2010 at 3:24
  • Yeah, it is a batch job. I had an unanswered question a while ago, and accepting took a while for my rate to jump back to 100%.
    – MPelletier
    Apr 13, 2010 at 13:37
  • 1
    When I have a question I cannot answer, I turn to the Stack Overflow community for assistance. I've asked 5 questions so far, only one of which got any answers. So I'm starting to answer my own questions with "I don't know" and accept it so users won't think that I have a low accept rate -- when in reality for all answered questions I should have a 100% accept rate. Which raises the obvious question: shouldn't accept rates be calculated based on answers (or answered questions), not based on questions?
    – Daniel
    Mar 20, 2011 at 15:29
  • @Daniel That's a good point, but it appears that unanswered and closed questions do not count against accept rate: meta.stackexchange.com/questions/25220/…
    – Pollyanna
    Mar 20, 2011 at 16:48
6

Just don't accept an answer if it doesn't solve your problem. 70% and up is considered green good. You don't have to shoot for 100%. Just stay out of the red and you'll be sweet.

People answering will also be able to see the answers you've chosen on previous questions and the quality or how it met your needs. If it doesn't, that could be stopping them from taking you seriously if you go on just accepting anything just for a high accept rate.

Looks like on SU you only have 3 questions to your name, in which case the accept rate won't show as yet for you.

Refer to the blog on, New Question Asker Features for the questions that are counted toward your accept rate:

  • Questions must not be community wiki
  • Questions must not be closed
  • Questions must be more than 3 days old
  • Questions must have at least 1 answer
  • There must be at least four eligible questions as determined by the above rules, otherwise the statistic will not appear.
Improve this answer
7
  • 7
    Hell, that just discourages me from ever asking a fourth question...
    – phenry
    Aug 26, 2009 at 2:07
  • 9
    I feel like all the people complaining about this feature don't understand it, so their complaints make no sense. "I dislike the way I misunderstand this thing!" Uh, ok.
    – Jeff Atwood
    Aug 26, 2009 at 2:15
  • 20
    Well, Jeff, I don't see how my question betrays a lack of understanding of how the feature works, or of what it's meant to accomplish. (I did in fact read your blog post before I wrote the question and understood why the accept rate wasn't showing up for me on SU; I just didn't mention it in the question.) I just think that there will be unintended consequences. The fact remains that if I ask a couple more hard questions on SU I'm going to have to wear a radioactive accept rate for the near term unless I start accepting answers arbitrarily, neither of which sounds like a great option.
    – phenry
    Aug 26, 2009 at 2:36
  • 11
    @Jeff, I don't understand your comment!
    – jtimberman
    Aug 26, 2009 at 7:02
  • 14
    @Jeff: If lots of people misunderstand something, it's probably easy to misunderstand. Sticking something like accept rate on the screen means that people will see it, and if it isn't clear what it is, or means, people will misunderstand it.
    – David Thornley
    Sep 28, 2009 at 16:21
  • 3
    It seems a simple solution to this problem is to make the "Questions must have at least 1 answer" rule to be "Questions must have at least 1 upvoted answer"...
    – Billy ONeal
    Apr 12, 2010 at 23:41
  • @Billy ONeal: Seems like you're on the right track, but the accept rate should probably be higher, maybe proportional to total views or using some other heuristic.
    – intuited
    Jun 15, 2010 at 16:07
3

What you have to remember when you answer a question is that there are lots of reasons why people accept particular answers including, but not limited to:

  • Most upvotes
  • Longest
  • Shortest
  • Lower rep among eligible answers
  • Higher rep among eligible answers
  • First
  • Has a code sample
  • Has a link, image or other formatting to make it stand out
  • A random comment that peaks the OP's interest or just happens to be the real problem

You can't count on getting the accepted answer even if you're right and first. Of course, that gives you a greater probability but you never know what motivates the OP.

Your best approach is simply to answer the question as best you can and move on. You'll get some accepted answers you deserve, you'll get some you don't and you'll miss out on some that you (think you) deserve.

If you go in with the attitude that you need to get the accepted answer, it's simply a recipe for disappointment.

Improve this answer
1
2

Strictly speaking, the acceptance rate isn't primarily an incentive for question askers, but a piece of information for answerers. The incentive function is secondary, and the intended incentive is for question askers to take the time to properly accept the most useful answers.

It so happens that sometimes the most useful answer isn't terribly useful. Whether or not an "informative and well-written" answer is the most useful response to other people with the same question is difficult to judge. The cases where it is not are probably a very small minority compared with the problem this system addresses directly: Question askers who ask run-of-the-mill questions which are answered in good faith, correctly and concisely, who nonetheless fail to accept those answers.

This byproduct incentive of tough questions having "inappropriate" accepted answers is probably an edge case.

Improve this answer
2

Actually, to boost your accept rate, you could also ask a few simple questions that haven't been asked before. Just to satisfy your curiosity. Those questions would be easily answered and an answer would be easy to accept, which in return will boost your acceptance rate.

Is this cheating? Maybe, maybe not. But SO is supposed to be a kind of wiki for all related questions and all you're doing is making sure more questions are asked. Just make sure your "simple" questions aren't duplicates...

Actually, if everyone continues to do this, sooner or later all easy questions will have been asked and the global acceptance rate of everyone will drop. When I look back at Experts Exchange, which I joined back in 2000, I cannot help but notice that most questions about Delphi were reasonable simple to solve. But during the years, all the simple questions ended up being asked already and new questions started to become more difficult. Nowadays, several sections of EE are saturated by the simpler questions, thus there's not much to gain from those sections! It's one reason why EE is losing it's value. (Not the main reason, though.)

Improve this answer
2
  • 1
    Too bad Joel already asked about moving the turtle in LOGO.
    – phenry
    Aug 26, 2009 at 13:24
  • 2
    This isn't cheating so much as gaming the system. There are serious problems if one needs to ask useless questions in order to rebuild one's reputation.
    – MPelletier
    Apr 12, 2010 at 17:48
2

If you are asking well though out questions and make it known in the question that you have tried the obvious (you don't have to state everything you've tried but make it clear this is not a newbie asking a silly question) then you will get two kinds of answers: trolls (inevitable) and well thought out and (hopefully) helpful answers from people with the right knowledge and who are trying to help.

If all you get is trolls then don't accept an answer. Don't worry about your acceptance rate, worry about the quality of your questions and you will get good answers if they exist. Then you can accept them.

Improve this answer
2

I'd say that the acceptance rate is more of an issue if your question and earlier ones appear simple to answer. If your question looks challenging or interesting, i don't think many would care about your acceptance rate. After all, if you accept an answer, they get just 15 rep. If they answer well, they can get far more from others reading their answer and upvoting.

Improve this answer
1

I just want to add: Some relatively new users don't actually know what the accept rate numbers are until you tell them. So for new users, I guess it doesn't really discourage them to ask relatively difficult questions.

Improve this answer
1

Consider the extreme case, where none of the questions someone asks receive acceptable answers. Aside from the implications about the community, a low accept rate then theoretically implies about that person that:

  1. they poorly describe their questions or fail to provide enough information to yield answers that they find acceptable
  2. they received good answers but they failed to recognize them
  3. they describe questions well and provide adequate information but every question is legitimately unanswerable (they are the least lucky person in the world when it comes to computer problems)

In the final case, the community deserves to know who they're dealing with, given that statistically, future problems that unlucky person encounters are likely also to be unanswerable.

Note: Tom W makes a good point in the comments.

Improve this answer
3
  • 2
    I'd like to express point 3 in a stronger form and simultaneously rebut your conclusion. 3 implies that the poster asks very good questions, because very few (if any) people know the answer. I don't know how you'd better define a 'good question'. That person is more deserving of expert attention than the average community member, not less. Their (low) accept rate does the opposite.
    – Tom W
    Sep 26, 2010 at 9:18
  • @Tom W, yes, I see and agree with your point. Perhaps the "Accept Rate" statistic needs to be qualified by the average vote count of the poster's questions. So the poster in point 3 might have a badge that reads "10% accept rate, 5.6 avg question score". Or the question score can be Bayesian ranked and read "89%tile avg question score". Alternatively, both numbers can be expressed in a single (somewhat cryptic) value (10+89/200): "50% avg question value".
    – Mike
    Oct 4, 2010 at 15:36
  • 1
    I think the system is broken. It encourages people to ask easy questions that CAN be answered. After noticing all the snide comments about accept rate, I went back and reviewed my own questions. They were mostly things that were hard to impossible (as it turned out), and thus had no correct answers.
    – David Dunham
    Dec 1, 2011 at 18:16

You must log in to answer this question.

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