Discouraging frivolous close voting by adding a small cost like downvotes on answers

One of my frustrations with SE is the frivolous/rushed close voting. Things like,

  • Voting to close a newbies post as unclear, without leaving a comment asking for clarification, which the newbie won’t see and will be unaware of what they need to fix.
  • Voting to close as duplicate where it really isn’t but the close voter wrote the accepted answer on the dupe target and might get upvotes.
  • Voting to close instead of making the itty bitty edits that it would take to keep the question open.
  • Multiple rounds of voting on a question with no reason given as to why it should be closed.
  • Closing questions with helpful, correct, accepted answers as “unclear”. I realized this question was closed after finding it via Google and after applying the solution to my problem.
  • Closing because the question didn’t have information that belonged in the answer but the users didn’t want to look up that information or scroll down the page.

Close voting has it’s place, on Outdoors.SE I am one of the top close voters of all time. However, I think it would help if users were guided towards helping improve questions or nuking the truly horrible ones. I don’t want to stop closing, just slow people down a little bit and get them to ask if closing is really the best option.

What I would like to propose is that a Close Vote costs one point, just like downvotes do, and you get the point back if the question is reopened or is deleted.


I 100% agree with the problem and the need for a solution. The problem is, if we don’t have “traditional SE-style reputation” then there are no points to “give back”.


I think there definitely must be a way, going forward, to discourage this. Maybe some form of penalty for those who vote to close a question only for it to be reopened again later. This might lead to disputes and “wars” so care must be taken regarding how this penalization is implemented.

1 Like

No, no, no! We must not be doing that.

Close voting is a necessary mechanism for quality control. Even if we chose to implement CV different from SE, we would need some similar mechanism. Having a cost for that would discourage it. On SE, answers are downvoted quite seldom, due to the reputation loss.

This is alsoa cultural/human problem. We need to solve it, by fostering the right attitude in the community.

The idea was to implement granular bans (e.g. only disabling close votes), which allow us to mitigate this problem: When you see abuse of close votes, flag for mod attention. If the mid agreed, they erroneous close voter might be warned or banned.

Furthermore closure isn’t final. When a post has been wrongly closed, it can be reopened.


I can see the problem and agree with it. However I am afraid that the solution, some loss of points, is not going to hold back the voters. To make a strong impact on this behaviour a bigger change of the closing-system is needed.


One possible way to cut down on this would be to only apply the penalty to the initiating voter, not to those that agree with them, so that only one voter would have any incentive to care about this, which would greatly reduce warring. Unfortunately, this would lead to hesitation to be the first. To compensate for that, we might also add some small benefit to only the initiating voter if the closure stays correct. Even a simple count on a profile page of correct first close votes would be something, or we could add badges for hitting milestones or correctness+quantity thresholds, gate a granular privilege by a combined correctness and quantity threshold, etc.

The overall idea is to reward (or penalize) those who are best (or worst) at quickly and accurately judging posts by the eventual consensus that will emerge.


If it takes several close votes to close a question, one could also make a penalty for the close vote (each one of them) which is revoked back if the post is actually closed. That way, people will at least have to stop to think “will enough people agree with me that this post doesn’t belong here?”

If the post is subsequently edited by someone else but the close voter, retracting the close vote afterwards should also revoke the penalty (because it will usually mean the problem was addressed).

I have always spoken in favour of as much filtering as possible before the question hits the site. Things like SO’s “ask a question wizard”. Force the OP themselves to do as much of the tidying work as possible and don’t post things that will just get closed anyway.

As for excessive close-voting, it is a small problem on SO. Veteran users repeatedly answering bad/incoherent/obviously duplicate questions is a larger problem, because that erodes site quality and credibility. If questions that should be closed don’t get closed, then this is what you get. It’s a serious enough problem that incorrectly closed questions could even be regarded as acceptable “collateral damage”, if that’s what it takes to maintain quality.

A question that gets incorrectly closed just annoys the poster. A crap question that gets answered destroys the credibility of the site.


And then you list 6 items that are all highly subjective. I feel SE has often not been strict enough with what questions should be allowed to live. We both have rights to our opinions.

This is what voting is for. Voting is a good mechanism when there are likely to be a spread of opinions. You or I don’t get to decide, the crowd as a whole does.


Strong disagree.

This really goes against the whole concept of voting. People have different opinions. If not, there would be no need for a vote in the first place. Now you want to penalize those people that don’t go along with the crowd? You seem to have the concept that others’ votes are “wrong” when they don’t agree with your opinion, or that of the crowd. That concept is what’s really, seriously, wrong.


That’s true for votes on e.g. public policy, where the question is whether to pass such-and-such a law, or whether to spend money on this or on that. It is not true for votes on whether a question meets the standard to receive quality answers.

People can have different opinions, but, certainly in most cases, and arguably in almost all, some of those opinions can be shown to be wrong in practice.

Yes. The vast majority of questions are straightforwardly either worthy of being closed, or not, and it’s valuable to accurately predict this as soon as possible. I’m not talking about rewarding or penalizing all voters, but only the first one to give the signal. So, in point of fact, the exact opposite of going along with the crowd.

It’s common when discussing voting in the context of Q&A for participants to accidentally bring their reverence for voting as a foundation for democratic polity into a setting where the goal is fundamentally different. We are not making decisions about what crimes to punish with 5 years of prison, or what to spend $10B in tax money on. We’re guessing what questions will result in good answers and trying to guess accurately. Votes are a useful tool for that purpose, but they are not sacred.

All that said, it may be worth canceling out all benefits and penalties for early votes if the question becomes sufficiently contentious that it’s clear a simple algorithm is no longer understanding the state of consensus well. For example, after multiple close-reopen cycles.


Not in my experience. You seem to be wanting to impose your view of when a question should be closed on everyone else. My bar for a good question tends to be higher than the average, but I strongly resent your attempt to punish me for that. Let the voters decide.

You are basically saying that if I see a question that I think is bad, then I have to consider what everyone else might think before casting my vote on whether it is bad. That’s just plain wrong. It will lead to fewer questions being closed, which is apparently your real objective. That will not be good for the site.

How about the flip of this? Punish those that answer a question that was later deemed bad?

And also add some punishment for those that attempt to circumvent the will of the site by posting an “answer” as a comment. This is rampant on SE. There would be an “Answer in comment” flag. If enough (3 maybe) such flags are cast, then the comment is deleted and the commenter looses something.

I am sometimes one of those people. There are times (for me this is primarily on DIY, but it could happen anywhere) where I am fairly sure of the basic answer, but I am not certain of a few important details. I know that eventually one of the real experts will provide a full answer. So I know that if I answer first then eventually my answer will fall to the bottom in voting because it pales by comparison with the experts’ answers. At the same, if I put in that comment then OP can have a head start on key things to look for (i.e., the things I know the experts will ask for more details on) and also prepare them for the likely answer (which may point OP in a different direction than what they had originally been thinking. So I feel it really won’t be a “archive for the ages” level answer and yet I feel I should do “something”. I will often go back and delete my comments if somebody else posts a good answer that includes my ideas (which are not “my ideas” per se - they are ideas that I noted but which come from reading similar-but-not-duplicate Q/A from others over the years).

1 Like

This is an abuse of comments. It is particularly bad if you do this because you think the question will soon be closed. Either way, it is still something we should strive to avoid.

The correct action in your case is to write a short answer. As long as it doesn’t get downvoted, you still solve the same problem. The OP still gets his “head start on key things to look at”, you can still “prepare them for the likely answer”, etc. Eventually your answer will be pushed to the bottom as other more complete or accurate answers are written. You won’t get much rep from that, but then again a comment is guaranteed to get none.

The only downside of writing your rough answer as a real answer is that it might get downvoted. Personally, I never downvote answers unless they are outright wrong, misleading, or badly written. I would not, and I think most others would also not downvote a answer merely because it is incomplete or approximate, especially if it is marked as such.

I have on rare occasions done this myself. Sometimes I don’t have time to write the answer that does the question justice, or the time or inclination to research it, but I think I can provide a useful hint. I’ve found simply saying so is all it take to prevent downvotes. Start with something like “This isn’t a full answer, but roughly …”. People tend to understand that and not ding you for it.


For bad questions, there are downvotes. Close votes are for questions that don’t meet certain quality standards that are clearly laid down in the rules. Using close votes for anything else in my opinion is a misuse of close votes.


“The voters decide” is what I’m going for here! That’s literally the point!

This entire paragraph is bizarrely reversed. The only view on whether the question should be closed that I’m “trying to impose” is … the same one that would have determined whether the question was closed anyway? Accusing me of trying to make my view/vote binding is unfair and rather hostile.

Actually, I’m now wondering if you misunderstood my proposal somehow. (I know I misread posts embarrassingly often.) It’s basically just extending SE’s helpful flags count and flag ban system to the first close voter (not any subsequent voters), then adjusting from there.

Absolutely. Isn’t that what agreements on close reasons are for? Many electrons have been spilled on SE metas over the years in order to establish sound criteria and educate (close) voters on what questions qualify, and why. This increases transparency and consistency, which encourages better questions and reduces feelings of unfairness.

Please do not make baseless assumptions about people’s inner motives. I have years of SE meta history that clearly shows this is not my objective. Nor do I believe my suggestion would lead to fewer questions being closed as an unintended consequence, either, unless of course there was no reward for correctly predicting the eventual consensus. Which is why I mentioned the need for such a reward, which could be cranked up higher if need be to encourage earlier close voting more.

Speaking very broadly, questions should be closed early and often, and edited and reopened early and often too.

I have to admit I’m not entirely sure whether these are serious suggestions or not.

I’ve mulled the idea of penalties for answering blatantly close-worthy questions in the past, but never come up with any specific scheme I was comfortable with. Most obviously, it runs up against problems with new answerers (who don’t have enough trust to even flag, much less vote to close, and don’t know what is and isn’t close-worthy yet).

I don’t think there’s any point adding penalties to comments flagged as answers; that would just discourage flaggers from correctly flagging them. In fact, it would probably be desirable to a) make it clear that comment deletion is the worst that can happen, ever, in such a case and b) make it straightforward to convert such comments to actual answers.

Strong disagree.

The herd-think will win in the end anyway. There is no need to punish people for having a different opinion. You are actually trying to stifle diversity of thought. Bad.

No, you’re going further and trying to punish people for thinking differently.

But your idea of closed for unintended consequences could be someone else’s idea of took way to long to get rid of this turd. That’s why there is voting.

Answers in comments do real harm to the site. That’s not how comments are intended to be used. This exact abuse is also rampant on SE. It’s a real problem. On a few sites (The Workplace comes to mind), the mods clean up such comments aggressively. Even better would be a mechanism that discourages them in the first place. Right now there is no cost, so those inclined to answer in a comment have nothing to lose. It would be good to fix that.

As for discouraging flaggers, that makes no sense. If I see a comment that is trying to answer the question, I’ll flag it as such. I don’t see anything discouraging me or anyone else from doing that. The OP and the comment writer would be only ones that have any reason to leave the comment.

So, for clarity, do you or do you not like having community consensus on close reasons and the right ways to apply those reasons?

“Let’s all agree on what it means for a question to be Too Broad” sounds a lot like “stifling diversity of thought” on the surface. But there is not a small difference between, on the one hand, accepting that not everyone sees things the way you do and that some sort of compromise is necessary, and, on the other hand, being told from the top how we are all going to do things.

I’m not sure what you mean by this.

I suspect I was unclear in how I phrased that, though, so how about this: I don’t believe my suggestion (as a whole) would have the effect of reducing closures, even as an unintended consequence.

Since no one has yet mentioned the actual size or type of penalty the only mention so far of a specific penalty has been a rep penalty, which may or may not even be applicable, and since I specifically recommended increasing rewards for early accuracy as much as needed to compensate for any fall-off in early close-voting, I don’t think there’s any real grounds for supposing that the net effect would be a fall-off in early close-voting. In fact, this might be a good way to encourage faster closing without becoming too sloppy and careless, even though the original motivation was to minimize sloppy and careless close votes without slowing down too much.

Taking my earlier remark about similarities to flag correctness statistics, perhaps decreasing available close votes per day by 1 every 5/10/15/however many clearly wrong first votes a person casts (and, likewise, increasing votes per day for clearly correct first votes). Most users aren’t reliably casting the first vote most of the time, and the majority of votes aren’t contentious, so it would likely take quite a while for most people to begin to see direct consequences. (Maybe too long, in fact, for the feedback to be as immediately useful.) I would expect most of the minority of users who routinely cast the first close votes to be tag experts patrolling diligently, and their votes should logically be quite reliable, so they shouldn’t have any problems.

Unless, of course, they have a systematic disagreement with the community about how or whether to apply particular guidelines. But if you have someone routinely digging up questions that most other close voters agree don’t need to be closed and casting close votes on them, then, especially if there’s some way to see a list of close candidates as in /review, this is a bit of a burden on the community, so gently discouraging them seems reasonable enough. If there’s a specific disagreement about a principle or rule of thumb to use, one that keeps coming up over and over, it’s better to resolve this by discussion than by protest vote. (The fact that tracking this could actually uncover such disagreements early should not be neglected. Catching it soon enough could prevent it from getting stuck as a point of contention and instead the whole community considers and maybe adjusts a little bit.)

The main practical challenge to enforcing this policy has always been well-meaning flaggers who look at what seems to them like a perfectly reasonable answer, just in slightly the wrong place, and balk at even the modest punishment of simply deleting the comment. Adding additional penalties on top of that is a great way to ensure they simply won’t flag such comments, because that would seem to mean punishing someone unjustly for trying to help in a way they thought best (but were wrong about).

Now, you may not care about that personally, but most flaggers do, in my experience on sites that have tried to crack down on this (especially ELL).

1 Like

Here’s one suggestion (it can be improved):

  1. Suppose it takes 4 votes to close or reopen a question, initially. So, 4 random users go and cast their vote for a certain question, which is then closed.

  2. Later, 4 other users will go and cast their vote to reopen it, so it’s reopened. If, for some reason, one - or more - of those users who initially voted to close the question have a change of heart, they should be allowed to vote for reopening. This is specially true if the question was edited in the meantime of course, but that’s not a barrier, yet.

  3. Now it should take 4 votes from different users to close the question again - whoever cast their vote in step 1 cannot do so again here. Those who voted to reopen in step 2 can vote to close again here, as long as the question was edited. (Except users who already had one change of heart.)

  4. If, as a result of step 3, the question is closed yet again, 4 different users can still vote to reopen it one more time. From now on, no one who previously voted to either close or reopen this question should be allowed to vote on it again.

  5. it should now require more votes - say, 5 or 6, to close the question one more time.

  6. At this point, it’s a matter of increasing the vote count requirements for reopening and closing ad infinitum, or locking it for moderators only (single-handedly or as part of a voting process of its own). Or even leaving it closed forever and that’s the end of it.


  • Those who voted to close in step 1 receive a “strike type 1” if their decision is reversed in step 2. (optionally: except if they had a “change of heart”.)
    A “strike type 1” means nothing by itself and there’s no penalty, it’s just a number that will be stored internally associated with those users.

  • Those who voted to reopen in step 2 will receive a “strike type 2” if the question is once again closed (in step 3), unless they have the change of heart, in which case they get a “strike type 1”.
    A strike type 2 results, by itself, in some immediate, but mild penalty being applied.

  • The exact same logic applies to anyone who votes to close in step 3 if the closing ends up being reversed in step 4.

  • Whoever votes to reopen in step 4 will get two type-2 strikes in case the question is closed in step 5.

As I said, a strike type 1 has no consequences at first, but it will accumulate. Say an user accumulates an n amount of these over an x period of time - as a penalty, they can have their vote weight reduced by half for a while (say, 2 weeks), or receive a “close-voting cooldown” that lasts a few days. The duration of such penalties shall increase for repeated “offenses” and the counter will not reset until the x amount of time has passed. The value of n is quite adjustable and can be pretty relaxed - or not.

Anyone who accumulates strikes of type 2 shall receive equal penalties to those applicable for strikes of type 1 with an equivalence of, say, 2:1 or even more. In addition to that, strikes of type 2 should take way longer to reset (like, 3x) and repeated “offenders” might have their close-voting privileges permanently revoked.

I know this sounds complicated and is still a rough proposal. Please let me know what you think, and whether it can serve as a basis for something.


Most of your suggestion seems fairly good, although admittedly surfacing the back-and-forth escalation in the UI in a natural way seems like it might get a bit sticky.

I assume the time periods would be rolling windows, probably either 7 or 30 days long to start.

These particular penalties seem stiff enough that you’d need a comparatively high value of n, and if that’s the only correction that would take a while before most people got any pushback from their mistakes. (Besides the purely advisory, probably mostly-hidden stats, at any rate.)

That said, having an automated temporary ban based on being clearly far out of sync with community close criteria is still worthwhile, say in addition to something else. That could be votes per day, as I mentioned earlier, or maybe something else, although I’m having trouble coming up with other alternatives that tie into close voting.

One final thought: are we focused mostly on absolute number of errors, percentage of errors, both, or what? I’d recommend going for a hybrid that requires a certain volume of errors, as well as dropping below a certain percentile floor in reliability, in order to trigger the maximum sanctions from the system. Otherwise you end up with odd results where a single careless close vote triggers maximum response, or contrariwise, someone who’s fairly reliable, but whose high volume of votes takes them along the ragged edge of getting temp-banned (or de-weighted). For example, with n of 10 in a 30-day rolling window, someone who usually casts 30-40 votes/day could end up being 99% accurate but still get penalized, whereas, checking my review history on SpEx shows the most recent 30-day window with only 91.4% accuracy (but far fewer votes). I would have been surprised to be sanctioned for the disagreements at 91.4%, but much more so if in fact I was agreeing with eventual consensus 98-99% of the time.

1 Like