We have incentive to ask questions: getting answers and getting points for a good question.

We have incentive for answering questions. Points, badges, reputation.

But I see two problems with the voting system.

  1. The first is that there is little incentive to upvote. The consequence is clear on Stack Overflow. Plenty of good answers with a score of 0. Isn't it just as important to find these, and upvote them as it is to post and answer questions? Why no reward then?
  2. The second is that there is no scale, or metric, for how many upvotes an answer should get. If an answer has been upvoted 40 times, and another upvoted 4 times, what does that mean? (Assuming they are are both correct.) Is the first answer 10 times longer? 1/10th the code? Did it take 10 times the experience to answer? Is it in a language with 1/10th the userbase as the second?

Because we have no incentive to upvote, and no direction as to how much we should upvote, all rankings are somewhat meaningless.

A music website like thesixtyone.com solves this problem with the following solution:

  1. There are no downvotes.
  2. Upvoting (initially) costs you reputation.
  3. As more people upvote the same answers you've upvoted, you earn back that reputation, and then earn more reputation on top of that.

That way you have incentive to find things with low scores that should have higher scores, and upvote. You are rewarded for voting when your vote agrees with others, and penalized when it doesn't.

I'm not saying that this is the correct solution to Stack Overflow. But I am saying it's a step in the right direction.



There are currently 48,048 questions on SO with no upvoted answers.

  • 2
    Slightly out of date, but see also: meta.stackexchange.com/questions/128 – Marc Gravell Nov 13 '09 at 22:16
  • As the graph on the above link was indicating, ~95% of votes are up votes. – Stu Thompson Nov 15 '09 at 19:14
  • 1
    "What is the metric for voting?" Actually we use imperial voting around here. None of this SI junk, no sir, and we resoundingly denounce kibibytes and other such abominations. Every man should be content to get thirty votes per furlong. – Pollyanna Feb 6 '11 at 4:28

Upvoting is not done often, and should be encouraged, but the purpose of upvoting is and should always be

  • Wow, I really like this answer
  • This answer is better than one already voted higher

If we give reputation for voting, then people will start to vote indiscriminately just for the reputation - even if you only gave 1 extra rep for hitting the daily voting cap. It would adversely affect the overall system.

I would be fine with upvoting badges, or possibly displaying some sort of metric with the user's answers showing how often they upvote (though I think that too might turn out bad).

But upvoting itself should not give a rep boost.

  • I don't think you read my post completely. Read again how thesixtyone's voting works. Then explain how people will vote indiscriminately. – z5h Nov 13 '09 at 16:29
  • 3
    +1, I totally agree with your point on giving reputation for voting. That is a very slippery slope. – Heather M Nov 13 '09 at 16:29
  • 2
    As a single point of data, I have 141 out of 645 answers with a vote total of 0 (could be some that were voted on both up and down) which means that perhaps as many as 20% of my answers don't receive votes. This seems ok with me - it means that about 20% of my answers were not terribly interesting. But I've been here since the beta, where voting was pretty wild and free. I'd like to see a good user's stats that joined in spring of this year, after things settled down to a reasonable steady pace. – Pollyanna Nov 13 '09 at 16:32
  • To be clear. I upvote a question. That costs me reputation. The next person (Bob) who upvotes loses reputation, but I get mine back, plus a bonus. Then another person upvotes and loses reputation, but Bob get's his back plus a bonus. As a question approaches a point where no-one thinks it should be upvoted any more, an equilibrium is reached. – z5h Nov 13 '09 at 16:33
  • @z5h - I didn't comment on your particular suggestion because it cannot work without significant change to the voting system here, and because it inevitably introduces popularity contests - from a system controls standpoint, the output is unbounded. It may work for the site you mention, and perhaps the algorithm is more complex than you state, because what you've described, in a vacuum, results in a system that is easy to game, and depends on popularity and subjectivity. – Pollyanna Nov 13 '09 at 16:36
  • Ah, so it's an early bird gets the worm system, and each participant has to decide, "is the vote low enough that it's likely to get another vote?" – Pollyanna Nov 13 '09 at 16:38
  • So there is absolutely NO benefit to upvoting after the initial period. It will cost you. This may be ok for popular short term sites like Digg, and the music site - the items that fall off the front page are not longer important. However, SO is building an objective collection of definitive answers to questions that google will index for everyone to use over time. I haven't submitted too much over the summer, but if you look at my reputation graph I still grew steadily over that time - people do find old questions and vote them up. They wouldn't if it cost them. Look up "long tail". – Pollyanna Nov 13 '09 at 16:41
  • I claim: there is a problem with the voting system. The symptom is answers that deserve much higher votes, and answers which deserve much lower votes. SO is in this state. It is not good for anyone. I also claim that the right people to fix this, are the users. And we need to give them "material" incentive (points) for doing so. If you are saying there is no problem, or there is but it will fix itself, then we will have to agree to disagree. If you are saying it's hard to fix, and my suggestions may not fix it, then we are having a good discussion. What's next? – z5h Nov 13 '09 at 17:07
  • Votes are relative. What do you mean that some answers deserve much higher votes? How many do they already have? How many, in your judgement, should they have? Why are you second guessing the community by saying that you believe the community is assessing questions incorrectly - could it not be that you simply have a different perspective? – Pollyanna Nov 13 '09 at 17:15
  • No, it's not that I have a different perspective. As one of dozens of examples I've seen on SO: I got 0 points for linking to tutorials I had written to help a person deeply understand the Y-Combinator. Another person there got +1 for humor. And another -1 for a correct answer, although it verbatim Wikipedia quote. stackoverflow.com/questions/93526/what-is-a-y-combinator Is it "perspective" that causes me to believe it's wrong to award humor points, punish correct information, and totally ignore an in depth tutorial? Or common sense? – z5h Nov 13 '09 at 17:53
  • 1
    Typically links to in-depth tutorials are not as highly voted as actual information written in the answer. That particular question, though, has a lot of answers, and it's easy to get lost among the forest. Specifically, the first three answers give the "real" answer immediately: It's a way to include recursion blah blah blah. Easy, simple, to the point. Your answer doesn't answer the question at all, and then links to an outside site. The outside site is hard to read, and doesn't really answer the question as well as the other answers. This is, I would say, an example of voting working. – Pollyanna Nov 13 '09 at 20:19
  • Note that the highest VOTED question has a very good answer in the answer, rather than a link (which is the lower voted, but selected answer). Lastly, there are at least 68 votes spread across the question and answers given. That's a very good example of a question where people voted up what they liked, didn't vote on those things that weren't interesting, and voted down things that they didn't like. – Pollyanna Nov 13 '09 at 20:25
  • The wikipedia bit ( stackoverflow.com/questions/93526/what-is-a-y-combinator/… ) is laughably poor. Technically correct, it does absolutely nothing to help the beginner due to its terse use of technical high-level jargon. Additionally, there are some users that frown on wikipedia cut-n-pastes, and will vote such answers down IF there are better answers that expand on what wikipedia says. Unfortunately stackoverflow is made up of people, and they have opinions on how things should and should not be done. It would be nice if we could encourage more voting without gaming. – Pollyanna Nov 13 '09 at 20:28
  • @z5h: About your claim: you do realize that this is a subjective claim. If we had a way to objectively tell what deserved to be voted up, after all, we wouldn't need voting. While I doubt anybody will claim the system is perfect, I'd like to see more discussion on what might be wrong before discussing fixes. – David Thornley Nov 13 '09 at 21:38
  • @Adam: perhaps I'm your "typical" user (I joined last fall but only really became active after late spring) - stackoverflow.com/users/40468/ether - 172 answers, with 45 of them at rank 0 = 26% – Ether Nov 13 '09 at 22:21

Isn't a warm fuzzy feeling enough? :)


Flippant answer:

300 votes = 1 silver badge.

Better answer:

By voting you are showing participation in the community, by showing this participation you might make it more likely that people will vote for your answers.

  • "you are showing"--You are "showing" nothing: to see your vote rate one should go to your profile and divide your vote numbers by the time you are registered. – P Shved Nov 13 '09 at 18:12

I, personally, upvote when I like the answer. It might be clever, informative, or simply contain something other answers overlooked. I don't need incentives, because it isn't actually difficult. I read, I like, I click on the arrow. (Alternatively, I read, I dislike, I click on the other arrow. That takes a little more consideration.)

Nor does it matter to me how many votes an answer has. The basic idea here is to make the better ones float to the top, not to judge the objective quality of the answer. This, of course, means that answers to some questions will garner more rep than equally good answers to other types of questions, but I don't think that's worth fiddling with.

I don't think we need an incentive to vote, because we already get votes, and I don't want people voting more or less at random because they're rewarded somehow. I don't think we want to mandate groupthink either. If people have incentive to upvote answers that will get lots of upvotes, they're going to be voting based on a basis other than "Is this a good answer?", and I think any such incentive will cause more problems than it solves.

I think the biggest voting problem right now is the sympathy upvote, and I think the biggest problem it causes is rep. If somebody posts an incorrect answer, and somebody votes him down, and somebody votes him up, and somebody votes him down, and so forth, the answerer gets 8 rep per cycle. Arguably, this is an easier way to get rep than posting correct and well thought out answers. I think this can be dealt adequately with by not giving full rep to upvotes received while negative.

Finally, I'd strongly urge against complicated schemes. At least on SO, we're all part-time highly literal perfectionists who want to know the rules exactly by virtue of what we do, and this does tend to attract people who are normally perfectionists who want to know the rules. I think most of us would be more comfortable with knowing exactly what a vote is likely to do.

  • You have good points, but I'm still not convinced. You claim we don't need incentives, we shouldn't judge answers objectively or via groupthink, and we should avoid complicated voting schemes. Consider the extreme interpretation of that: Remove voting and reputation all together. That is much simpler, and completely subjective. Each reader has to judge every question on their own. My feeling is that this would kill StackOverflow. People would lose interest, get lost in the sea of answers and become apathetic. But this is actually happening. There is a sea of unmoderated answers. – z5h Nov 14 '09 at 0:49

You upvote helps other people. If I see more votes, I tend to trust the answer. So you vote up something that deserve it.

You must log in to answer this question.

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