5
$\begingroup$

I've noticed that when I ask a question on MathOverflow, or any StackExchange Site, and I either upvote or downvote an answer written by a high-reputation user, regardless of how old the question and answer are, I very frequently get a corresponding upvote or downvote on my question shortly thereafter. I assume this is because high-reputation users can see who cast the upvote or downvote, and they reciprocate the questioners' vote.

First of all, is it true that sufficiently high-reputation users can in fact see who casts the vote, is it true that some of them engage in this practice of "reciprocal voting", and is this an acceptable practice? Personally, as far as I can tell, the upvotes and downvotes I've received as a result of it have pretty much cancelled each other out, or maybe redounded slightly to my benefit in terms of reputation. (Assuming this is all not just in my head.) But is it fair for the vote count on a user's question to reflect how the user has voted on the answers?

$\endgroup$
  • 20
    $\begingroup$ No, they cannot see anything of the sort. $\endgroup$ – Todd Trimble Sep 21 '13 at 1:27
  • $\begingroup$ @ToddTrimble Can absolutely no one see this information? Can moderators not see it either? What else can explain how old questions get upvoted the day after an answer has been upvoted, and correspondingly with downvoting? Is it that the user sees that their answer has been upvoted or downvotes, not knowing by who, but then because of that they decide to look at the question again and vote on it? $\endgroup$ – user40280 Sep 21 '13 at 1:31
  • 2
    $\begingroup$ @ToddTrimble Could it be some psychological thing, like when a user is upvoted, they think to themselves "Wow, it looks like people really liked my answer. That's great! It was an excellent question after all, so I'm glad I answered it", and if they're downvoted they think to themselves "My answer didn't do very well. But what could I have done? The question didn't really make much sense in the first place."? $\endgroup$ – user40280 Sep 21 '13 at 1:37
  • 1
    $\begingroup$ I am only reporting what I know, speaking as a user with 24.1k rep. I think moderators have tools I don't have access to which might enable them to track voting patterns. I'm not sure what you are seeing or what you think you are seeing. I see now that Scott Carnahan has also answered, in a manner consistent with what I know. $\endgroup$ – Todd Trimble Sep 21 '13 at 1:40
  • 2
    $\begingroup$ If you want to know my opinion, though, I think your proposed explanation (the "psychological thing") is extremely unlikely for high rep users. $\endgroup$ – Todd Trimble Sep 21 '13 at 1:43
  • 1
    $\begingroup$ @ToddTrimble Well, it might be a subconscious thing. And the high-reputation users part may just be a coincidence; they're the ones who frequently answer questions, so it may just seem like it happens more often with them. Or this could all be in my head, which is likely if no one else reports this phenomenon. $\endgroup$ – user40280 Sep 21 '13 at 1:45
  • 6
    $\begingroup$ I reeeeaaaalllllyyyy doubt it's some sort of subconscious reaction on the part of high-rep users. If there's something that's bumping the question to the front page, then people who hadn't seen the question before might be voting, pro or con. $\endgroup$ – Todd Trimble Sep 21 '13 at 1:55
  • 1
    $\begingroup$ @ToddTrimble No, this is about question threads that don't get bumped up at all. I just upvote or downvote an answer, and then a little later I get the corresponding vote on my question. And when this phenomenon occurs (assuming it's not in my head), it doesn't happen in the opposite direction, where I get a downvote after giving an upvote, or vice versa. But as I said, I may be wrong about the high-reputation users thing. It may just be that they comprise a lion's share of the answers, so I'm getting a misleading impression of them. $\endgroup$ – user40280 Sep 21 '13 at 2:54
  • $\begingroup$ I've nothing more to add, and don't know what to say. I've reported what facts I know. $\endgroup$ – Todd Trimble Sep 21 '13 at 3:12
  • 1
    $\begingroup$ So ... the question about why votes are anonymous may be relevant here ... meta.mathoverflow.net/questions/828/… $\endgroup$ – Gerald Edgar Sep 21 '13 at 13:05
  • 1
    $\begingroup$ I have a reputation of slightly under 60,000 on math.stackexchange.com and I am not able to see how anyone votes, other than myself. $\endgroup$ – Michael Hardy Sep 26 '13 at 4:35
  • 1
    $\begingroup$ .....but perhaps I should add that organized bullying does exist on stackexchange and mathoverflow. $\endgroup$ – Michael Hardy Sep 26 '13 at 4:46
  • 1
    $\begingroup$ @MichaelHardy Data? $\endgroup$ – Did Sep 27 '13 at 12:47
  • 2
    $\begingroup$ @Did : I would like to adduce data. This present version of "meta" has existed only since the recent conversion to stackexchange. If I had access to a searchable form of the old "meta", than I would gather up the data and present it either here or to the moderators. (BTW, contact information for one of the six moderators is nowhere to be found, and for four of the others, it is not on stackexchange; one must search the web.) $\endgroup$ – Michael Hardy Sep 27 '13 at 22:20
  • 2
    $\begingroup$ @MichaelHardy, as we've mentioned a number of times, e.g. meta.mathoverflow.net/questions/833/… anyone can reach the moderators via moderators@mathoverflow.net. $\endgroup$ – Scott Morrison Sep 30 '13 at 3:49
17
$\begingroup$

You wrote

when I ask a question on MathOverflow, or any StackExchange Site, and I either upvote or downvote an answer written by a high-reputation user ... I very frequently get a corresponding upvote or downvote on my question shortly thereafter.

It's not clear from that phrasing whether the answer you upvoted is to the question on which you receive the upvote. If so, then a plausible mechanism is that the high-rep user get's notification of the upvote (without your name attached to it). This has nothing to do with being high-rep; if you mouseover your name at the top of the screen, you should see recent votes on your questions. The high rep user clicks on it out of curiosity ("when did I answer a question about large cardinals?"), sees your question and upvotes it. I have definitely done such things. (In my case, I feel fairly confident that I have only done this in the up direction; 99.5% of my votes are upvotes.)

Nothing here logically depends on the user being high rep, but high rep users are probably more likely to check the vote notification at the top of the site, to have forgotten one of their own answers and to remember to upvote questions.

$\endgroup$
10
$\begingroup$

There is only one group of users that can see who voted on specific posts and that is SE employees with developer access. Not all SE employees have that kind of access, only those that need it for their work (developers and the community team, mostly). And even then they can't easily see this information on their own posts (though devs could check in the database directly).

Moderators can't see who voted on specific posts, but they have certain tools that show voting information for the purpose of detecting vote fraud, but those don't display individual votes.

But all users can see their recent votes if they hover over their user name at the top of the site. If a user checks the post that was recently upvoted, they might also vote on some other posts on that question as well.

$\endgroup$
  • $\begingroup$ But then what explains the correlation of upvotes with upvotes, and downvotes with downvotes? Are users more likely to upvote a question if they see their answer has been upvoted, and correspondingly with downvotes? If so, isn't that somewhat problematic? $\endgroup$ – user40280 Sep 21 '13 at 16:35
  • 1
    $\begingroup$ By the way, what is vote fraud? Is that where users use sockpuppets to vote, or are there other forms of vote fraud? $\endgroup$ – user40280 Sep 21 '13 at 16:40
  • 10
    $\begingroup$ @user40280 I don't think this discussion can go anywhere unless there is first hard evidence presented for this alleged correlation. $\endgroup$ – Todd Trimble Sep 21 '13 at 17:36
8
$\begingroup$

Users cannot see who voted on things. It is physically possible but extraordinarily unlikely that someone is monitoring your vote count and correlating that with her/his score. Moderators can see some voting information, but I don't think it would be proper to spell out the details.

You may be seeing correlations for the following reason: When an old question gets bumped by an edit or new answer, the question and answers tend to attract new votes. Alternatively (and very speculatively), someone who answers a question might be more inclined to vote the question up after seeing that the answer is appreciated by at least someone.

$\endgroup$
  • $\begingroup$ Aren't the capabilities of moderators publicly available information? mathoverflow.net/help/privileges?tab=all mathoverflow.net/help/privileges/moderator-tools $\endgroup$ – user40280 Sep 21 '13 at 1:47
  • 7
    $\begingroup$ The moderator tools accessible to 10k users form a strict subset of the full set of moderator tools. For example, 10k users cannot suspend other accounts. $\endgroup$ – S. Carnahan Sep 21 '13 at 1:50
  • $\begingroup$ This is about question threads that don't get bumped at all. Now your more speculative explanation may be right, but if it is isn't that somewhat problematic? Users rewarding a question because they liked that their answer was appreciated by someone, and downvoting a question because people didn't like they answer?. On the other hand, it could just be that the users just take a closer look at a question for which they recently received a vote, which would be a good thing. But then why the correlation between upvotes and upvotes and downvotes and downvotes? $\endgroup$ – user40280 Sep 21 '13 at 3:01
  • 6
    $\begingroup$ @user40280 I don't really see a point in speculating further in the absence of hard data. See en.wikipedia.org/wiki/Confirmation_bias $\endgroup$ – S. Carnahan Sep 21 '13 at 12:18
  • 1
    $\begingroup$ You're right, I might be suffering from confirmation bias, since I may not notice all the occasions in which this phenomenon did not occur. $\endgroup$ – user40280 Sep 21 '13 at 16:36

You must log in to answer this question.