0
$\begingroup$

It appears to me that questions/answers receive votes at a rate quite positively correlated with the votes they already have (of course, this only applies for a certain period of time, I am not implying a runaway behavior by this hypothesis :P). A highly upvoted answer seems likely to receive further upvotes more quickly compared to the rate at which it received the same number of votes before it was highly upvoted. Admittedly, this is purely based on personal observation and I have no proper analysis to support the conjecture. But it does seem to fit well with my informal observation as well as intuition. It might also be that once a post gets more votes, it simply becomes more visible and that causes the accelerated upvotes but I am not sure if this is the only factor (again, I have no actual analysis).

I was wondering if it would be a good idea to keep votes invisible by default and make it so that they can be viewed only by clicking between the arrows (the way the visibility of the distribution of upvotes vs. downvotes works). This would be a safeguard against premature involuntary positive feelings towards a post and it would still be possible for everyone to access how much support each post has gotten simply by a click. Answers can still be arranged according to the vote counts, of course.

This is most certainly an SE wide thing but just wanted to float the idea first here.

$\endgroup$
7
  • 1
    $\begingroup$ Some runaway effect, aka snowballing, does occur though, especially when a question gets Tweeted or hits the HNQ. $\endgroup$
    – PM 2Ring
    Commented May 11, 2020 at 11:54
  • $\begingroup$ @PM2Ring Yeah, agree, I just meant that it'd stop after a time :P Nevermind, in any case, what's HNQ? $\endgroup$
    – user87745
    Commented May 11, 2020 at 11:59
  • 1
    $\begingroup$ HNQ = the Hot Network Questions list. Here's a discussion from late last year about the impact of the HNQ on Physics. physics.meta.stackexchange.com/q/12515/123208 $\endgroup$
    – PM 2Ring
    Commented May 11, 2020 at 12:07
  • $\begingroup$ For example, this question got a boost from being on the HNQ. ;) $\endgroup$
    – PM 2Ring
    Commented May 11, 2020 at 12:17
  • $\begingroup$ @PM2Ring Ooh, that explains a lot of questions I thought I would never get an answer for ;) $\endgroup$
    – user87745
    Commented May 11, 2020 at 12:21
  • $\begingroup$ I think you need to give a good argument as to why this "voting feedback" is not good for the site. Do you think that mediocre answers that get a couple of up votes then get many up votes. I agree with your observation, but most of the time when I see an answer that has this feedback effect I don't think "this answer is horrible" (it does happen sometimes though). $\endgroup$
    – BioPhysicist
    Commented May 11, 2020 at 12:27
  • 1
    $\begingroup$ @AaronStevens No, I don't think horrible answers or even mediocre answers are snowballing their way to high votes. I just think it reflects a kind of issue nonetheless. In the sense that voting is supposed to be based on an individual's assessment, fairly independent of other people's assessment. $\endgroup$
    – user87745
    Commented May 11, 2020 at 12:33

1 Answer 1

Reset to default
5
$\begingroup$

Ideally, questions and answers should be evaluated on their merits. I agree that the displayed score can bias subsequent votes to some degree, but that's actually inherent in the design of the Stack Exchange model.

Votes are intended to send a signal to the post author (and to affect their rep), but their primary purpose is to indicate to other readers whether a post is useful or not. It's only natural that if the early readers find a post useful (or vice versa) that many of the later readers will agree, and vote the same way.

There are other effects at play, too. For example, if I see an answer with zero score that looks reasonable I may not vote on it if I don't have sufficient expertise in that domain. However, if the answer has a positive score, and I see a favourable comment or two from people who I recognise to have some knowledge in that topic, then I'm more likely to trust the answer and add my upvote to it. And conversely, if comments from those experts criticize an answer (and I understand that criticism) I'm more likely to add my downvote.

I don't think it would be useful to hide scores. We want people to be able to quickly see which posts the community has high confidence in, and which one's are probably not worth looking at. Your proposal doesn't make scores invisible, but it does slow down that evaluation process.

FWIW, mid last year, a score hiding experiment was conducted on Stack Overflow. The idea was to prevent the pile-on effect of downvotes on low quality questions and answers from new members. The system correctly tracked the question's score, but the displayed score never went below a minimum threshold, both 0 and -1 were tested. This experiment was rather unpopular with the long-term residents of Stack Overflow meta. I can't find a clear statement regarding the outcome of that experiment, but you can read about it here: Why can I no longer see that a post has a negative score?

$\endgroup$
2
  • $\begingroup$ Well, FWIW, it looks like they really were trying to not hurt new users' feelings and it wasn't really an experiment designed to see how to go about preventing pileup (why focus on only negative pileups otherwise?). In any case, the response of the community seems to be heavily influenced by this. Also, the score was completely hidden--as in it couldn't be seen by clicking somewhere, etc. as far as I understood. $\endgroup$
    – user87745
    Commented May 12, 2020 at 3:25
  • 1
    $\begingroup$ @DvijD.C. The score wasn't completely hidden. Similar to your proposal, "You can see the real up/down vote counts at any time by clicking the score while viewing a post - even if you haven't earned the Established User privilege". See meta.stackoverflow.com/a/390179/4014959 $\endgroup$
    – PM 2Ring
    Commented May 12, 2020 at 4:23

You must log in to answer this question.