It has probably been already raised in some meta over SE. The question is, should we vote "relatively", i.e. based on if the question/answer is useful/well-written/etc., or "absolutely", i.e. by assigning a reasonable value to the question and voting in that way.

For example, if I see a good question, but not that good, that already has 200 upvotes, should I upvote? If I see a question that I find just OK, and has 200 upvotes, should I downvote?

  • Excellent question. For the most part, I vote "relatively".
    – bbgodfrey
    Jan 17, 2018 at 0:14
  • @bbgodfrey Same for me. But I didn't find a robust strategy to deal with what is not in "the most part", hence my question :-)
    – anderstood
    Jan 17, 2018 at 1:35
  • 8
    With currently 9 Upvotes, your question comes pretty close to my personal upper limit. Expect downvotes soon :)
    – halirutan Mod
    Jan 17, 2018 at 18:01
  • I would hesitate to single-handedly open this can of worms, but thanks for saving me the trouble: I read where experts are thoughtful before voting for a different expert's answer, whereas, on comprehensible q's and a's, all are "useful" to a neophyte and I can see that a Luminous member might question my profligate upvoting. But to me the answers are useful. All sorts of repu. based weightings come to mind, but what to do? Recommendations/ideas encouraged.
    – Rabbit
    Jan 18, 2018 at 2:43
  • 1
    I have been wondering this myself for a long time. I also wonder about upvoting not-so-great (but not bad) answers. The upvote tooltip says "this answer is useful". But "useful" -- having ϵ > 0 utility -- isn't neccessarily a very high bar. And if there are other great answers to the same question, I feel my voting should reflect that. In the end, I decided that everyone has their own voting system -- from stochastic to absolutist to rich-get-richer to my own agonized and overthought method -- and while many of those are kind of silly, the result seems to actually make some kind of sense. Jan 19, 2018 at 10:48
  • This question and its answers are related and relevant to this discussion
    – m_goldberg
    Jan 20, 2018 at 5:44

2 Answers 2


You should vote as if you would not see the absolute votes already given. This is my strong belief. The final result in votes should be a sum of the single decisions of the community.

If you look at it closely, everything else does not make sense. If you see a question that is written well, intriguing and probably a thing you always wanted to know, why on earth should you downvote it if it already reached the upper limit of upvotes? Don't do that. Even if it means we have to live with a buttocks-question that has close to 300 upvotes.

Far more interesting is to try to embrace a consistent voting style. Especially for questions, because it seems there is a tendency that questions sometimes don't receive the voting attention they should get. If I see 4 answers to a question that is at least OK, and there are no votes for the question itself but upvotes for the answers, something is off.

In my opinion, votes should reflect how many people find this question/answer interesting/helpful. The worst thing that can happen is if no one cares anymore and no votes at all are placed. Then, our site will go down the drain. You might have seen my little analysis of votes in our chat recently.


This is the mean of votes per week on c++ questions on stackoverflow. As you can see, the c++-tagged questions don't receive any attention anymore. If you visit stackoverflow, you will find that it turned into a rather hostile environment that is unfriendly to newcomers and will push users away.

Our site has developed a similar tendency. Here you see the percentage of questions that have no answer and no vote. There are things to consider as apparently dead questions are deleted by the system after a while, but the tendency is clear.


If no one votes, we will become an ignorant community. Therefore, as soon as I have read a question or answer, I try to place a vote. My votes are relative. If it is good, it gets an upvote. If it is a bad answer, I downvote. Bad questions rather get a close vote and a comment.

  • If I see 4 answers to a question that is at least OK, and there are no votes for the question itself but upvotes for the answers, something is off. Yes that's what I mean; then would you upvote even if you don't find the question personally interesting? Apart from that, I fully agree that people should vote (up or down). We must be careful when interpreting the curve though: maybe a lot of unanswered questions are from a class of beginners that did not know about the site 5 years ago (a bit like math.SE is mostly filled with homework questions now).
    – anderstood
    Jan 17, 2018 at 1:28
  • 8
    About the first point: I cannot speak for the upvoters of the answers, but the posters of the answers have to admit that they find the question reasonable enough to spend their time answering it. That deserves an upvote imo. (2) You can take the range between 2016 and 2017 as an indicator. The last steep part is because the system did delete the abandoned questions yet.
    – halirutan Mod
    Jan 17, 2018 at 1:38
  • 1
    "You should vote as if you would not see the absolute votes already given." Generally, I agree, but I don't apply this to downvoting. If a post already has a significant negative score, I would rarely give it yet another downvote. (At -3 closing and deletion is already a possibility.)
    – Szabolcs
    Jan 17, 2018 at 9:32
  • @Szabolcs Yes, you are right. That is one specific case where I also consider not downvoting to oblivion except the post spam or insulting. However, in most cases, I apply this rule only to questions and as my final paragraph says, questions rather get a close vote and a comment anyway.
    – halirutan Mod
    Jan 17, 2018 at 14:04
  • Hmm, these plots without AxesLabels are close to useless. I do guess that X axis is time stamp in years, but what is on the Y axis? Especially in the first plot.
    – Ruslan
    Jan 26, 2018 at 18:25
  • @Ruslan The graphics were already uploaded to imgur and I didn't want to recreate everything. Nevertheless, the plots were helpful to show the general situation. As I have written in the text, the y-axis of the first plot is "mean of votes per week on c++". To be more precise: Collect all questions within one week on SO that have the c++ tag. Calculate the Mean of the vote-count and you get the value for this specific week. The x-axis is time in years.
    – halirutan Mod
    Jan 26, 2018 at 18:28

I vote with users in mind. Here are some voting patterns which are in my past, even though it is hard to say how often or frequently I have used them.

If there is a question with zero votes that isn't inherently a bad question then I'll vote on it just to signal to the OP that the question is well written. If I ask OP to change something and he does it then I will vote for the question even though it is still not that good. If a question is undeservedly downvoted then I'll vote for it to show OP that not all of us agree with that. If the best answer has fewer votes than other answers then I'll vote for just that answer even though the other answers are also good, for future vistors' sake and for the answerer's sake. As a corollary, if I have voted on one answer then I will also want to consider voting for the other answers even though it may not have been those answers that caught my attention. This way I am not party to creating an imbalance where there shouldn't be one.

Note, however, that: Just because I sometimes do these things it does not mean that it happens very often. Usually I end up voting based only on the quality of the post, simply because none of the scenarios above are applicable.

I have a lower bar for voting on posts with few votes or no vote at all and it may equally be perceived that I have a higher bar for posts with more than a few votes. But I want to make it very clear that I support voting liberally. We absolutely want to be a community that votes a lot. The problem is that if we try to be consistent and create a baseline for what is worth an upvote, then some opportunities will be missed. I prefer to use my own judgement to e.g. encourage new users rather than to be rigid about it.

Downvotes is an interesting type of vote that has to be considered separately, because its effect on users is different. I have cast 3966 upvotes but only 22 downvotes at the time of writing. I don't see the point of downvoting a post, it just puts the user down. If I a have to criticize a post then I would rather leave a comment so that I can at least make it constructive. But, as Szabolcs commented to another answer here, it would be particularly bad to downvote a question that already has several downvotes. There is no need to pour salt in somebody's wound.

I would never use a downvote to even out a perceived imbalance of votes between answers. The significance of the downvote as I think of it makes that impossible.

It is obvious from what I have written that I vote "relatively," and yet I don't think anything that I've written here suggests that I am creating the problem of apathy described in halirutan's post, i.e. that questions and answers don't accrue as many votes in total anymore. But I feel like the analysis of why that trend exists is best left for another answer to another question.

Finally, I don't care how others vote, if they vote relatively or not. The one thing I would strongly recommend is that everyone keep their bar for what is worthy a vote low and vote frequently.

  • 9
    +1 for "If there is a question with zero votes that isn't inherently a bad question then I'll vote on it just to signal to the OP that the question is well written." On that note, a pet peeve of mine is when a question has one or more answers but no votes for the question itself. If someone thought a question interesting enough to answer, surely they could bother to upvote the question itself. edit - I just read Halirutan's answer here and see he covered my point
    – Jason B.
    Jan 17, 2018 at 14:30
  • @JasonB. I definitely agree with that!
    – C. E.
    Jan 17, 2018 at 16:48

You must log in to answer this question.

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