# mysterious immediate upvotes not based on content

I had noticed before that answers sometimes receive upvotes almost immediately, and had been wondering how the upvoter had been able to read and assess the answer so quickly, but so far there had always been at least something like $10$ seconds between the posting and the upvote.

However, today I posted this answer (now deleted), and I never got to see it without upvotes – by the time the page had loaded, the upvote counter already said "$1$". The page didn't take unusually long to load in my browser, so the upvoter must have loaded the page just when the answer appeared and then almost immediately clicked to upvote.

Now usually one might perhaps have thought that they had also been working on an answer and thus knew the answer, saw the correct result at the bottom and immediately clicked. That would still be somewhat negligent, since the rest of the answer might be wrong or misleading or badly written, but at least it would be acknowledging a correct result.

It just so happens, however, that the answer was completely wrong and didn't answer the question, since the question asked for the first moment and I (being a physicist) associated "moment of inertia" and calculated the second moment instead.

So this upvote was almost certainly not based on an appreciation of the answer; and that makes me think that some of those other surprisingly fast upvotes that I've seen in the past weren't, either.

Which raises the question: Who's doing this, and why, and how can we prevent it?

• Isn't it possible that whoever upvoted your answer had the same misinterpretation of the question that you had? – Rahul Mar 3 '13 at 21:05
• @Rahul: It's certainly possible, but that was just an additional aspect that makes it even less likely -- I'd already started writing this question before the OP pointed out my mistake, because even without that it just seemed too fast. I literally never saw the $0$ on the upvote counter. I'm surprised even that someone could click that fast, even without glancing at the answer. And even if they did have the same misinterpretation and did somehow manage to load, glance and click all in a second or two, it would still be unfortunate that answers are upvoted without being read even cursorily. – joriki Mar 3 '13 at 21:18
• It'd be natural to suspect me based on my answer here. But rest assured that I (a) neither read nor upvote calculus-type answers; (b) do read the answers that I upvote. – user53153 Mar 3 '13 at 21:35
• I wonder whether the server end of the SE software allows a rogue client to vote for a question number before a question to go with that number even exists. That would seen to be necessary to explain your upvote-on-the-first-pageload observation. – hmakholm left over Monica Mar 4 '13 at 5:44
• (Well, there's a window of time between when the server replies to your post with a redirect and when your browser requests a new page load, during which an upvote could theoretically arrive. That would be at most a few hundred milliseconds, during which someone else would need to have received a fresh pageload and sent a vote back to the servers.). – hmakholm left over Monica Mar 4 '13 at 5:51
• #highrepproblems – Alexander Gruber Mar 4 '13 at 6:16
• Essentially the problem is "how do we prevent people from upvoting before we think they have properly evaluated the question," which seems to be an intractable problem fraught with subjectivism, to me. (Unless, of course, there really are some voting shenanigans going on...) – rschwieb Mar 4 '13 at 19:59
• This is what I have seen many times at this site. – mrs Mar 11 '13 at 16:08

## 3 Answers

Voting on the site is not necessarily based on mathematical content. Many times people vote out of "loyalty" or "appreciation" for a user. I know that sometimes I do that as well, when I see a user which consistently writes good answers I will usually vote his/her answer before reading it fully. Why? because I know that the probability that the answer is going to be great is very high. Rarely I am mistaken, in which case I may decide to undo my vote.

Certainly it sounds very reassuring that people will tell you that "they only vote answers they read", but in a site this big it's not true anymore. Voting has to do with the mathematical content, but it's still not 100% correlated with it.

I do believe that this happens only with well-established users that wouldn't get there without writing a lot of good answers to begin with. But I don't know about that, and we'll probably never know until voting habits are made public.

• (+1) but myself, I'd change "until voting habits are made public" to "unless voting habits are made public". – robjohn Mar 3 '13 at 23:41
• Along these lines, here is a strange case: math.stackexchange.com/questions/317110/… In this case, there are 2 answers, one of which I provided. I believe my answer addressed the OP's concerns and correctly provided a path to a solution - as well as the solution for good measure. The OP accepted the other answer, however, despite the fact that it provided a poorly-thought out hint that led nowhere. My requests to show how this solution would work went unanswered. So, yes, I understand the concerns here. – Ron Gordon Mar 4 '13 at 1:16
• @rlgordonma Note, however, that your answer got upvoted more than the accepted one, which means the voters exercised better judgement than the OP. I had a similar case here. This seems to be a trend: when voters see that the best answer was not the one accepted, they "compensate" with extra votes for the non-accepted answer. – user53153 Mar 4 '13 at 1:39
• @5pm: Yes, over the long haul, that is true, for which I am grateful. But in the short term, it was not the case. Thanks for posting your case, I see what you mean. – Ron Gordon Mar 4 '13 at 1:44
• +1 ... now I will read the answer. – Neal Mar 5 '13 at 16:41

It is hard to impose any sort of conditions on the way people vote. Some votes seem very well thought out and others seem random and even counter-intuitive.

In the end, all one can do if votes seem questionable, is to ask the voters for their reasoning. This may get replies and it may not, but I don't see that much else can be done.

• I suppose one could jiggle the software so that no upvotes are accepted until 10 seconds after the page is opened (or $n$ seconds, where $n$ is the length of the answer, in lines). I don't know that I'd be in favor of doing that; I'm just saying that I suppose it could be done. – Gerry Myerson Mar 3 '13 at 23:09
• @GerryMyerson Sounds like a good idea. The comment votes are already timed (must be 5 seconds apart), and the buttons in review queues do not become available at once on page load. Having the same behavior of vote buttons would make sense to me. – user53153 Mar 4 '13 at 6:47
• @5pm, feel free to make a feature request on metaStackOverflow or wherever it is that these things are discussed. – Gerry Myerson Mar 4 '13 at 12:47
• @GerryMyerson It seems that there already was a feature request along those lines (although the poster was more concerned about downvotes). As you can see, it did not receive much support (the score is -24). – Martin Sleziak Mar 6 '13 at 21:44
• @Martin, some of that $-24$ may have had to do with the poster's attitude and the specifics of the downvote about which the poster complained, but I take your point. – Gerry Myerson Mar 6 '13 at 23:09
• (I cooled down to the idea of software-imposed delay. If someone wants to vote without thinking, they will do exactly that, only 10 seconds later.) – user53153 Mar 8 '13 at 4:25

The style in which an answer is written and the author of the answer can both be used to make inferences about the extent to which an answer is correct and/or useful without thoroughly understanding it. Obviously neither of these is foolproof, but then again neither is thoroughly understanding an answer. Imperfect feedback is still probably preferable to very little feedback (which to me is the most likely consequence of requiring a more rigorous standard for upvotes, even assuming that we could implement such a standard).

• There was no time to assess the style in which the answer was written. I'd actually be surprised if there was even time to glance at the author or the final result for a second, but certainly nothing less trivial than that. Since I never saw the zero upvote count, the time the upvoter had was at most the difference between the times it took our browsers to load the page, and mine wasn't being particularly slow, so it was a matter of fractions of a second, at most a few seconds. – joriki Mar 3 '13 at 23:47
• @joriki: a few seconds is usually how long I take to decide whether to upvote or downvote an answer. – Qiaochu Yuan Mar 4 '13 at 0:10
• But that doesn't really explain Joriki's observation. When he presses the "Submit" key, the servers more or less immediately start serving his browser a freshly rendered question page, complete with the vote total for his new answer. Why would the server wait until it had shown the answer to some other user, and that user sent back a vote? Even if somone used a bot to upvote the answer without human interaction, his vote would not reach the SE server over the network until the initial render was already sent to Joriki. There should be no way he'd see an upvote_on the first pageload. – hmakholm left over Monica Mar 4 '13 at 5:41
• @Henning I think vote totals on the question page are fetched separately from the static page elements, they change without a page reload. So it's possible that during the time it took to render the page in joriki's browser the script responsible for the vote count fetched updated information. – user53153 Mar 4 '13 at 6:53
• @5pm: I've never seen them change spontaneously, and they are present in the raw HTML response from the original page load (checked with wget). – hmakholm left over Monica Mar 4 '13 at 7:05
• @HenningMakholm The count is present in HTML, but it also gets live updates via WebSocket, if your browser supports it. I saw it happening more than once, and some quirks of this feature were discussed on meta.SO here and here. – user53153 Mar 5 '13 at 3:34
• I am totally ignorant about the inner workings of the SW, but I have seen the upvote count change spontaneously. – Jyrki Lahtonen Mar 5 '13 at 10:11
• @joriki: I'd be curious, whether there is some option, that the half(?) automated "mirrors" at some foreign hosts could manage to upvote automatically based on some contents or user or some other fancy stuff... – Gottfried Helms Mar 5 '13 at 13:00