Following on from the recent question about upvoting (or the lack of it), here is a specific example where I'm not sure what we 'should' be doing with regards to upvoting. I provided a short answer to this question, but didn't upvote because it isn't a great question for the reasons Cardinal points out in a comment. None the less I hoped my quick answer might be useful.

To me, upvoting a question says 'this is a good question worth reading' which I don't think is true in this case. However I've taken the time (admittedly a very short time) to provide an answer. Is that inconsistent? By providing an answer am I encouraging poor questions? In my view this isn't worthy of a downvote, just not an upvote. Thoughts?

  • 6
    $\begingroup$ Just to be complete, an upvote is defined as denoting that "This questions shows research effort; it is useful and clear." $\endgroup$
    – Aarthi
    Oct 8, 2012 at 23:31
  • $\begingroup$ Where is that written? I have tried to find a clear definition (i.e. in the FAQ) with no success. $\endgroup$ Oct 8, 2012 at 23:34
  • 4
    $\begingroup$ Hover over the upvote/downvote button. The hover text provides guidance on what +1/-1 really mean. :) $\endgroup$
    – Aarthi
    Oct 8, 2012 at 23:35
  • 2
    $\begingroup$ Ah ha! Thanks for that. I guess tool tips probably end up getting read more than the FAQ anyway. $\endgroup$ Oct 9, 2012 at 0:34
  • $\begingroup$ Regardless of your question, in that particular case it sounds to me this question is just a duplicate of earlier questions asking for references on mastering ML techniques (e.g., stats.stackexchange.com/q/18973/930; stats.stackexchange.com/q/26044/930). As such, I would recommend looking for the best duplicate and let the community decide. $\endgroup$
    – chl
    Oct 9, 2012 at 10:57

2 Answers 2


+1, great question. I'll admit that I often don't really know what it means for a question to be a really good question. I mean, it's clear what a really good answer is (e.g., they are, 'clear, authoritative, correct, definitive, well-illustrated, linked to appropriate references, at most lightly edited, and even..."pretty"'). You can just recognize a really good answer when you see it.

But what is a good question? At first blush it would seem that questions can't be good, because by definition, if you are asking a question, you don't understand something, and it doesn't seem clear how you go about not understanding something particularly well.

Of course, a question can "show research effort", and be "useful and clear", as @Aarthi points out. However, that's an and function--all three conditions must be met. Moreover, if you want to be strict about it, it can be difficult to meet any of those conditions. If someone really does a lot of research before asking their question, there's a decent chance they will have stumbled across the answer somewhere else and never post their question here in the first place. Furthermore, what is "useful" supposed to mean? I would guess it means the question will help many other people in the future, but what about an otherwise good question on a highly specialized, esoteric topic? We do get some of those here, and I wouldn't want to exclude them out of hand. "Clear" seems like it would be the easiest criterion to meet, but consider that many (maybe more than half, maybe less, I have no idea) of the askers on CV don't speak English as their native language. It's impressive to me that someone can ask a highly technical question in a second language in the first place; requiring that it be asked 'clearly' may be a high hurdle to jump. I don't want to take this too far, certainly there are many questions that meet all three criteria. But if someone wants to be strict about it, it's probably less than half and clearly the idea isn't to have only a few questions with upvotes. (When I was starting out on this site, I would mostly just vote on answers, and this little yellow flag would pop up saying "I notice you've been voting on a lot of answers recently, you know questions need votes, too".)

To be honest, I'm still not always sure whether a question is good (aside from those cases where the official criteria are met). My resolution is to also upvote those questions that afford good answers. Think of it this way: when someone searches the site, what's listed are question titles along with their tags and their total vote count. Having a higher vote total on the question is one of the signals that that thread contains useful information (i.e., the whole thread--the question and the answers). So if there's a really good answer on a thread, I'll usually upvote the question, even if the question may not quite meet all the official criteria. One result of this is that I usually upvote questions that I answer (albeit not always--there are cases where I think the answer might help the OP, but isn't necessarily great and won't be of much interest to anyone else). Thus, I advocate an or function here: if the question meets the cut on it's merits, or if any answer on the thread is particularly good, or if I think the thread will be helpful for people in the future; or even if there aren't any such answers yet, but I suspect there will be, I will typically upvote. (Disclaimer: I'm under the impression that I may upvote more than most, so take that fact into consideration.)

  • 1
    $\begingroup$ You know, this whole discussion reminds me of conversations I used to have with my old friend Plato, drinking wine and looking up at the stars... $\endgroup$ Oct 9, 2012 at 3:00
  • $\begingroup$ That Plato guy was a smart dude. His mentor, Socrates, though? Kind of a kooky old man. $\endgroup$
    – Aarthi
    Oct 9, 2012 at 3:02
  • 4
    $\begingroup$ +1, especially for the "...questions that afford good answers" point. I have upvoted and favorited a number of questions because the discussion it bred was interesting and informative. $\endgroup$
    – user5594
    Oct 11, 2012 at 0:31

I think I have voted up $100$ questions and $119$ answers.

I think there are plenty of questions which are worth answering which are obviously flawed. Perhaps no motivation or definitions were given for obscure terms, but I happen to recognize them anyway. Perhaps the question is needlessly ambiguous, and I think one interpretation is interesting. Sometimes it is possible to take even more time to fix the question, but I think answering without voting the question up is reasonable. That there are answers already draws attention to the question.


You must log in to answer this question.

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