# A minor gripe about upvoting things that have been downvoted

I have noticed that sometimes, when I downvote a new question or answer that is borderline, exactly one other person will upvote it and no others. I can guess at the reasoning behind this:

• This question / answer is not quite good enough to have a positive vote total, but does not deserve a negative vote total.
• Therefore, I will vote it back up so that its vote total is zero out of fairness.

This is quite a minor gripe, but I would appreciate if people did not do this. Since downvoting only gives -2 reputation while upvoting gives +5 (questions) or +10 (answers), the net result is +3 or +8 reputation for the user. This behavior is not "fair": it incentivizes borderline content and disincentivizes minor downvotes.

In other words, please only upvote such things from -1 to 0 if you would have upvoted them from 0 to 1.

• Occasionally I have done this, since I tend to vote often. I won't from now on on downvoted questions. (Unless they have been improved, or should be upvoted) – Eric Naslund May 24 '11 at 15:33
• I think the reputation system is somewhat separate from the totaling mechanism. For the reputation, it would make more sense for downvotes that keep the total non-negative to completely cancel the reputation from the corresponding upvotes, and upvotes that keep the score nonpostive to cancel the corresponding downvotes. But it seems like a minor issue to me. When I vote I'm not worried about reputation, I'm worried about giving information to others about the question or answer I'm voting on. – Carl Mummert Jun 7 '11 at 15:12
• In short, I think we should just change the point system rather than trying to change voting behavior. – Carl Mummert Jun 7 '11 at 15:16
• In my opinion, this complaint isn't just minor but also misguided. You've made a measure of community approval (rep) the target, rather than what should be the target: actual community approval of a question or answer. – AppliedSide Jun 8 '11 at 19:50
• @Nick: isn't rep supposed to roughly correspond to community approval of a user? I am just as worried about that as I am about questions and answers. – Qiaochu Yuan Jun 8 '11 at 20:17
• @Qiaochu: Roughly, sure. It also serves as a way to give users a gold star. I think unwarranted negative feedback is a sufficient reason to give a user a little boost of serotonin. – AppliedSide Jun 9 '11 at 1:02
• "This is quite a minor gripe, but I would appreciate if people did not do this." I see nothing wrong with this type of voting, I find once a question has its first downvote others come more quickly (like the first vote is a catalyst). It's my opinion you should NOT downvote "borderline" questions/answers/posts because what you may find to be "borderline" could be seen as very useful to someone else, but it may not get the attention it deserves because of your vote. – Squirtle Apr 22 '13 at 1:53

I have done this a few times for some first questions, and I don't do this out of "fairness", it is my assessment whether this first post deserves standing at $-1$ in my opinion. And I can guarantee you that "-1" has quite a different psychological impact on a new poster than "0" and a comment.

I don't see how one or two +3s for the very first post of a new user is a problem or will lead to more borderline content. It is in fact quite unlikely that a new user understands how they got +3.

I also sometimes happen to upvote first and non-first posts who stand at -1 because I just want to upvote.

I would say that this phenomenon is a consequence of the system and I actually think that is very unlikely that people are smart enough to hit the right obfuscation level on this forum and at the same time want to exploit this to gain reputation. (Except maybe one time for curiosity's sake.)

I disagree that voting "should" be independent of other votes, but this may be a topic for a different thread and "should" should be defined first, although I probably disagree for most definitions.

(Simple recent example: I happened on an older thread with a highly upvoted accepted answer where I added my vote. Then I read the second answer with significantly less votes and realized that it was better. So I removed my vote from the accepted answer and upvoted the second because this was the only way to help indicate that the second answer was better, especially as many people would not read the second less-voted unaccepted answer. Now, the accepted answer totally deserved my upvote, so my vote was based only on the voting pattern.)

Edited to add: The example shows that the votes of the people who "only vote on the content" also depend on prior votes. Since everything is sorted by vote, it is wrong to pretend that there are any "independent" votes.

• Note that the "should"-paragraph was a reaction to a comment of Bill Dubuque that he deleted in the mean-time. – Phira May 23 '11 at 12:02
• I was referring to votes on the same answer - not different answers. But I don't disagree that there may be some exceptional situations where it may make sense for votes on the same answer to be dependent on prior votes (e.g. to discourage piling on downvotes on newbies, who have yet to learn norms). As for your example, I do agree with using votes to affect the sort order of answers, as a way to get better answers exposed. I often encourage folks to do such because far too often the best answers have fewest votes (sometimes only because they weren't first). – Bill Dubuque May 23 '11 at 17:30
• In other words, while Qiaochu may want the question or answer to stand at -1 to communicate his disapproval, some other voter may not want such (harsh?) disapproval communicated to the poster (especially if it's a new user). – ShreevatsaR May 26 '11 at 5:05
• @ShreevatsaR "especially if it's a new user" It sounds like you bend the rules based on who a user is. Are you also one who would be sympathetic to female profile pictures? (No pressure to not answering the last question) – Simply Beautiful Art Dec 6 '17 at 0:31
• @SimplyBeautifulArt I was summarizing this answer. But to answer your question, yes I most certainly do on the one hand expect regular users to understand the community norms, and on the other hand consider it important to create a welcoming environment to newcomers, for the sake of the long-term health of the site. (See Please do not bite the newcomers for an essay at another website.) I do not consider this “bending” the “rules”. (And no I don't care about profile pictures.) – ShreevatsaR Dec 6 '17 at 0:38

I don't like this behavior (counter-voting) for the following reason:

I think that it is not the job of any particular user to decide what is the right total vote for a question, it should be decided by the community as a whole, the total vote should be an indicator of what a user reading the question thinks about the question on average and this is a signal to other users.

The counter-voting biases the average towards the opinion of those who are counter-voting, and makes the signaling to other users about the quality of the question less useful.

I think users should vote independent of the total vote a question has at the any time and the votes of other users and solely base their votes on the quality of the question/answer in their opinion.

Counter-voting is not just voting about the question but it is voting about the votes and opinions of other users about the question and deprives other users from their rights to their votes. Would you like it if other users used their votes to counter-vote not because they think the quality of the question deserves their vote but just to cancel your vote?

• I think I disagree. My opinion is that the total for each question is determined by the community as a whole as a sort of simulated annealing: if someone thinks that the total is lower than it should be, she can vote up, and similarly for down votes. The end result is that each question arrives at the total that the community thinks it should have. At least when I vote, the question I have in mind is "does the total for this question match what I think the total should be". I have no objections if others vote the same way, of course. – Carl Mummert Jun 7 '11 at 15:07
• @Carl: +1, you have a very interesting point that I didn't think about. It looks plausible that a simulated annealing like process is at work in practice. But I still think that this view complicates the process a lot, the total vote a question deserves is quite subjective and also relative to other questions, I can try to think of situations where the outcome of the process is not a satisfying one, and may even find situations where the mechanism is not truthfull if the user's utility function is to make the total vote as close as possible to the amount they think the question should have. – Kaveh Jun 8 '11 at 2:33
• In comparison, the view that a vote represent the opinion of the user w.r.t. the quality of question independent of other users makes the voting process much simpler (and probably truthfull). So if we can convince the users that their votes should be independent from other users, i.e. make the utility function of users independent from each other, then I think the result will be a better indicator of the quality. But maybe this is just not a realistic view. – Kaveh Jun 8 '11 at 2:44
• I do see your point. – Carl Mummert Jun 8 '11 at 12:50

I don't see any problem at all.

I think the appropriate comparison (for a question) is between net +3 (a downvote and an upvote) vs. 0 (no vote either way). If you explain the downvote (you do that, right?), the OP will have something to learn.

This is not about 10 downvotes followed by 10 upvotes = net +30...that does seem wrong. But just the smidgen of +3 (if that is all it ever amounts to), is not really enough to care about.

If upvotes and downvotes were more psychologically similar, then they'd be designed to be more equitable in value (or the other way around).

I stumbled across a related question and commentary at meta.SO about sympathy voting. The end result is that (with numbers computed on those sites)...well the opinions go both ways even with objective numbers.

And here's a another relevant meta.SO thread about preventing pity voting. Mostly contra.

• The absolute value of the reputation difference is not what I care about: it's the sign. Marginal changes are what drive incentives, and if I'm going to end up awarding reputation to someone because I think they wrote something unhelpful, then downvotes become useless: the only reason I'd downvote something is if everyone was more or less equally willing to agree that it's useless, and in that case I don't need to downvote at all. – Qiaochu Yuan May 23 '11 at 19:01
• A '0' net vote tally (not rep points) is also a disincentive. And even if the new user doesn't realize that they don't have the marginal -1 vote, they'll learn from the downvote explanation. – Mitch May 23 '11 at 20:16
• :) I just noticed the downvote on my answer, and if your comment was an explanation of that, it didn't register as such with me. So would you have no problem with downvoting in this situation if the rep value of a downvote equalled that of an upvote? – Mitch May 23 '11 at 20:20
• @Mitch: I did not downvote this answer. I don't think I was sufficiently clear: I am talking both about incentives for the person I'm downvoting and for my incentives. If people are just going to upvote whatever I downvote that's borderline, then I am disincentivized from downvoting borderline things, and I don't like that because it prevents me from communicating my disapproval quickly. Making the rep value of a downvote equal to that of an upvote would make it too easy to abuse downvoting, so that's not a solution either. Perhaps a slightly better solution would be to display both the... – Qiaochu Yuan May 23 '11 at 20:25
• ...upvote and downvote totals by default. – Qiaochu Yuan May 23 '11 at 20:26
• OK, I see. I think you were clear, and maybe I was emphasizing the view from the votee. I think the effect on the votee does help with -your- incentive though, with an explanation about the downvote an annihilating upvote won't remove the incentive of telling the votee that their things needs improvement. +1 on displaying both up and down votes especially since they are of different rep-value. – Mitch May 23 '11 at 21:17
• There might also be some other psychology at work. Posts with positive or negative scores probably get read more often, because the person reading sees the score and wonders what the reason for the vote(s) was. This definitely happens when there's a ton of up- or down-votes. Zero-score questions then look the least interesting: they look like people don't even care or didn't even read it. This suggests that borderline posts get "stuck" at zero, due to people not reading them, and quickly return to to zero if they get a vote in either direction, when other people notice them and vote. – Douglas B. Staple Mar 27 '13 at 19:55