On Stack Overflow, I have tried to consistently vote for some of my friends' answers, since he generally provides good answers to somewhat obscure questions but doesn't get a lot of credit, since they're not terribly popular. One day I noticed that his reputation dropped a ton; upon further inspection, I noticed that all the votes I'd given him in the past had disappeared.

Since I am a long-time reader of Coding Horror and I know Jeff has weird ideas like "you can sanitize HTML with regular expressions", I just assumed that Stack Overflow had randomly lost a ton of data because it's full of bugs ;-). So I dutifully upvoted everything again, burning several days' worth of votes on this rather than voting for other worthy things, hoping to work around a bug or data migration issue.

After doing this for a couple of weeks in a row, I eventually realized that it must be intentional and did some poking around, and discovered Vote Fraud and You and More Voting Anomalies.

Personally, I don't regard this as "fraud" at all. I'm voting for my colleague's answers because I think they're good answers; I discover them through his list of answers, but I vote for them because they're good. I'm also not the only one upvoting these answers, either.

I don't think I do anything that is fraudlent. I don't vote exclusively for my friends. I legitimately participate in the site: I ask questions and I answer questions; I upvote good stuff and I downvote bad stuff.

So, while I respectfully submit that the administrators of this site and I could agree to disagree that this is legitimate voting behavior, as a legitimate contributor in other respects, I would have at least appreciated an error message telling me that my votes were being denied because I'd voted for this person too many times. Maybe a nice hint like "You've voted for so-and-so too many times. Give somebody else a chance!", with a link to the main questions page. If those are the rules, then okay, I'll play by the rules; but this is more like getting a foul in a competitive sport, and instead of a referee calling out a red card or whatever, the scoreboard just changes silently with no word from the announcer, and sometimes the other team leaves without a word.

(PS: please don't quote joel's "Not Just Usability" at me as an answer here. I am not saying that the site should disclose all information to all users at all times; I can understand that sometimes information should be hidden from attackers. But a real abuser here would have much more quickly resorted to a fleet of sock-puppets spamming from different IPs, rather than assuming that the system's algorithms were otherwise reasonable and would eventually recognize that my interest / votes were genuine.)

2 Answers 2


The question here isn't whether you are doing something wrong, it is whether or not your behaviour can be distinguished from someone who is.

The motivations you described are perfectly acceptable. Presuming that you have, in fact, verified that your friend's answers are solid. However, your actions would be identical to the actions of a sock-puppet. Since it is impossible to determine motivations from actions, (and you could lie about them if you felt like it) we have to identify the actions that a malicious person would use, and react to that.

StackOverflow reacts very benignly to such "illicit" actions, by silently reverting them and not informing the user. This is the best method, because it protects the exact mechanism (so that the malicious people won't know the parameters that tipped the authorities) and it protects the accidental upvoters, because they'll just move on with their lives not notice that their votes were removed.

The only people who will notice are people who are consistently attempting to do something which StackOverflow doesn't condone.

Or in other words, if your actions are identical to the actions of someone malicious, then maybe you should stop.

  • I think that my comments on the other answer here are a good response to this answer as well. It is possible to distinguish my actions from those of a sock-puppeteer.
    – Glyph
    Dec 11, 2010 at 9:33

Sorry, but I have to disagree with you here. Mass upvoting that is targeted at a person instead of their contributions is wrong. No matter whether your colleague's answers are all good: Your focus is on the person, and that is a) not okay and b) impossible to tell apart from malicious upvoting (i.e. upvoting a sock-puppet account).

As for warning when the fraud detection mechanism triggers, this has been discussed here among other questions, with the conclusion that it's not going to change (If I understood it correctly.)

  • I disagree with (a), but I'm perfectly happy to say that my opinion is irrelevant there. I'm not running the site and I don't get to set policy about what the social rules are. But (b) is just wrong. There are plenty of ways to tell my behavior apart from a sock-puppet account. If you're only looking at this one brief pattern of voting it might be hard to tell, but SO has a huge amount of data at its disposal to make this determination. And again, I'm not asking for the rules to change, just for a little courtesy and respect in informing me what the rules are.
    – Glyph
    Jun 18, 2010 at 17:44
  • 3
    Glyph re (b), how? Considering that there are hundreds of thousands of active users, and this would have to be done automatically? How can you tell apart a user signing up a sock puppet account to leave 10 upvotes for their real account, or a new user signing up, and leaving 10 upvotes for their very competent friend? Re informing you about the rules, that's what the question I linked to was about - but Jeff obviously wants things to stay the way they are.
    – Pekka
    Jun 18, 2010 at 18:26
  • 2
    Well, off the top of my head: how long has the account been active before the "abuse" began? What is the ratio of upvotes to downvotes? How many other votes does the account have? Does it have over a certain reputation threshold itself? SO could also implement the sort of reputation algorithm which Advogato had, which made "clique" votes worth a lot less (more documentation here: <advogato.org/trust-metric.html>). Some of these may work better than others, some may not work at all, but to say that it is "impossible" to tell the behavior apart is just a failure of imagination.
    – Glyph
    Jun 19, 2010 at 6:48

You must log in to answer this question.

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