My subjective impression is that there is less outrage about questions being closed than there used to be. I feel like the main function of vote trading was to mitigate this outrage. If this function is no longer necessary (or is less important), the method of starting a meta thread is strictly better than posting a "vote to keep open" comment.

If nothing else, the sooner you vote to close a question, the sooner somebody else has the opportunity to vote to reopen it. Of course, I'm not suggesting you vote to close/reopen willy-nilly. As always, you should be able to defend your vote to close/reopen. Preferably, you should defend it in the comments or on meta before you are even asked to do so.

Policies have to be formally decided upon by the "ruling class" and made clear in some obvious way, with a good way of notification of updates. No such method exists on MO (the FAQ certainly is not normative) so policies are hard to implement. This is fine by me as I don't want to have to click an EULA to use this site, and I like the ease of entry that these sites afford.

Conventions are simply things that a group of people agree to abide by. We have a convention that when downvoting, we leave a comment. Not all of us do it, it is unenforceable, and there are no penalties when you don't do it. But nonetheless, it was discussed here and thought to be a good idea, so those who are aware of it try to do it, and hope that by doing it then they make MO a better place (and that others will see it and do likewise).

A policy is imposed from on high and is a "you must". A convention is agreed by the masses and is an "I will".

This vote trading is a convention. There are no strictures saying that people have to abide by it, it is merely a way to avoid certain unpleasant situations and those who prefer to avoid them join in with the trade. But no-one has to abide by it. If someone puts a "I cast a virtual vote to stay open" and then the question is closed anyway, there is nothing to stop them making that virtual vote a reality by casting a reopen vote. This convention is here because there were some heated arguments about questions being closed early, and we prefer to work in a calmer environment. But if someone chooses to ignore it, that is their choice.

That said, there are a couple of things that I would change about this convention. I don't like the "pre-emptive" votes to stay open. If someone feels that a question is in danger of being closed, but no-one has actually voted to close yet then the right thing to do is start a meta thread and link to it.

I also think that those putting a "don't close this" vote should include their current reputation. Since reputation is not displayed on comments, one needs to click through to see that the person really does have enough reputation to participate in the close/open debate. We sometimes get copy-cat comments and those are annoying.

If the vote-trading practice is implemented at all, I would prefer not to have to write things like: "I vote to close, cancelling X vote". IMHO, it will make things more personal than necessary.

Moreover, my impression is that MO started out with certain goals, not as a manifestation of "the community"...

@Harry's most recent comment: do you mean the post about the URL and javascript? (Sorry, as this thread has gotten a bit long and hard to keep track of.)

How do other SE sites manage this issue? StackOverflow has 87 pages of users who can vote to close, and MO has a mere 2.5 pages (two and a half!!!)... I have not looked, but I expect SO users do not end up in longuish debates about whether to close or not to close questions.

</fake french accent>

Also I'm pretty sure the Mods have the power to send e-mail to all users. A curt missive stating "some changes to voting policy has made. See the FAQ" is not out of the question.

Anyway, my point is that since it is impossible to make anything completely foolproof, why not just settle for something that is easy to implement?

(Yes yes, people are unpredictable and hard to control. That's why I do maths and not sociology.)

I'm voting to close this question because X, Y, and Z. If you agree, either vote up this comment or vote to close (don't do both). - Anton

I'd be puzzled. Maybe after reading

I don't think this question should be closed because R, S, and T. I disagree with objections X and Y because of W. I can see where you're coming from with Z, but I don't think it's worth closing on that basis. If you want to pre-emptively vote to reopen this question, please vote up this comment. If you want to vote to close and this comment has more votes than Anton's, please consider voting up Anton's comment instead of voting to close. - Yemon

I would understand the mechanism. More probably, I would step back and leave the whole business of closing to other people.

Ultimately, if there's a moderate amount of disagreement about closing a question, I think somebody should start a thread on meta.

Perhaps we should make a rule that if the Vote To Close and Vote Of Support comments each gets more than X votes, someone who sees it should open a thread on meta?

The way I first imagined vote trading would work is this. Suppose I find a question that I think is terrible. I vote to close it and leave a comment to the effect of

I'm voting to close this question because X, Y, and Z. If you agree, either vote up this comment or vote to close (don't do both). - Anton

Maybe other people come along an also vote to close, but eventually Yemon comes along and decides that the question shouldn't be closed, so he leaves a comment to the effect of

I don't think this question should be closed because R, S, and T. I disagree with objections X and Y because of W. I can see where you're coming from with Z, but I don't think it's worth closing on that basis. If you want to pre-emptively vote to reopen this question, please vote up this comment. If you want to vote to close and this comment has more votes than Anton's, please consider voting up Anton's comment instead of voting to close. - Yemon

Then people follow Yemon's suggestion. The advantages of this approach:

• We don't need to do any of this comment deletion. I know I'd get annoyed by the clutter, but I think I'd also be annoyed by feeling like I have to clean it up.
• All instructions for how to behave are contained in the comments. You don't have to send people to meta or to the FAQ or anywhere else. You don't have to explain that there's a system. They know what to do the first time they see it, and they'll get used to the system after a while. It feels more organic and cleaner to me.
• The vote trading mechanism doesn't get engaged unless somebody cares enough to leave a comment. So the process of closing blatantly bad posts isn't bogged down by bureaucracy.
• People can vote on comments long before they can vote to close/reopen. Maybe this will make people feel like the process of closing questions is less cliquey. Then again, it also means that it should be acceptable for high rep users to circumvent the system and vote to close if they really feel strongly about it -- that's the power that comes with 3k rep.

Ultimately, if there's a moderate amount of disagreement about closing a question, I think somebody should start a thread on meta. It would be a bad idea to replace thoughtful discussion on meta with some impersonal vote trading system. Since votes to close/reopen expire after four days, slowing down the process of closing/reopening will have the effect of fewer questions being closed/reopened.

I'm intrigued by Andrea's suggestion to implement a solution with javascript. It's certainly feasible to rebind the close/reopen link to automatically talk to votes.mathoverflow.net instead of submitting a vote to close/reopen. That way we might be able to make the UI work essentially exactly as it would if the feature were implemented in the SE software.

1) because MO traffic has increased, a question on the front page which has got votes to close can sink off the top page before other people who might vote to close have a chance to see it;

2) I think that the geographical spread of MO's users means that there can be apparent consensus to close a question when a significant number of people (say >5) might feel otherwise, but not be wasting, er I mean investing, time on the site. Perhaps this is a factor in some people's frustration with the apparent clique-driven nature of closure?

These are two very ill-thought through notions and so no offence is meant, and all rebuttals are welcome.

I still think that if we are looking for a "people"-based solution (as opposed to a software one), the simplest is best: if everybody be a little less trigger happy with closing questions, especially when a constructive comment has been already placed and the question may stand a chance after a second edit, then we won't have to worry about this problem at all.

When it comes to the obviously off questions, no one will complain about closure anyway; and for the contentious ones, no matter what kind of system we put in, someone will complain. So I don't see what this complicate system is actually doing for us.

On the other hand, the fact that the cookie is available on Scott server opens interesting possibilities. I think I will open a new thread for this.

I think if the users with sufficient reputation just be a little bit more hesistant in wielding the closure axe, especially in the case that a question has up-votes, this should be a non-issue. This jerry-rigged system that we are discussing, unless implemented in software, seems more trouble than it's worth:

(a) A person who voted to close or against closing has to keep checking back the comments to see if his vote has been cancelled. (a') Noah's suggestion for the moderators to keep track of this really will, I think, overwork them.

(b) Users have to keep track themselves who actually has sufficient rep to vote for closure. (b') Honor system for double votes is problematic

On the one hand I would like to see the ability to "vote to cancel", and perhaps even a graduated response system coupled with that: something like if 10 people voted to close the question, the question gets closed immediately. If <5, the question stays open. If the question maintains a state of having more than 5 but fewer than 10 votes for a set period (say 12 hours), then it gets closed. So lovers can still "vote to cancel". But things like this really should be in-software rather than in meat-ware.

duh! I'd forgotten that at a subdomain of mathoverflow.net on my server, we can probably see the "real" cookie!

While I agree that if this really works it does open up some opportunities for more features, personally I prefer the "low-tech" solution Pete suggested at the beginning of this thread. Given that it appears there's some consensus around that, I think we should try that out first.

Hopefully, it won't come up often: for clear cut cases, by definition it will work as usual. For anything controversial, it adds an extra layer of complexity, but hopefully also results in a more consensual outcome. I still strongly advocate for creating meta threads, and I'd say even that it is "good form" to create a meta thread any time you vote to close and aren't absolutely sure that everyone will agree. So much so, perhaps, that I'd say we should attach some opprobrium to voting to close without creating such a thread, and then having people disagree.

Of course, this might frequently put me in the sin bin...

I understood that the other system was based on trust too (I could vote to reopen after my vote has been cancelled). But yes, on a second thought it may become a mess.

how would we check for sufficient reputation, and double voting? If users OpenIDs were public we could insist that you're logged in at the other site, with the same OpenID.

I'm dubious however that the implementation effort would be worthwhile, even if we could agree on a new system.

a) implement a simple system with a database which contains a unique table deletion with fields (id, count). id refers to the is of the post on MathOverflow and count would be the number of votes to keep open - votes to close.

It would be called by visiting an address like

http://votes.mathoverflow.net/open/50000

http://votes.mathoverflow.net/close/50000

The first URL

1. adds the entry (50000, 1) if 50000 is not present in the id list, or

2. increments count by 1 for the id 50000.

The second URL

1. decreases count by 1 for the id 50000.

2. If this is already 0, it answers with an error

b) create a javascript bookmarklet for voting up or down. The bookmarklet is like a normal bookmark, but has the effect of calling one of the two addresses above. There would be two different bookmarklets, one for voting to close and the other for voting to keep open. If the second URL gives an error, the bookmarklet notifies the user to cast a normal vote to close, or even better calls itself the official MathOverflow address for casting a vote to close.

In this way we could have our own vote to close system without waiting for the SO guys. The bookmarklet way is the easiest one to have compatibility with all browsers, but we could try other ways: Firefox-Chrome extensions, GreaseMonkey scripts or whatever.

I don't know what the best way to keep the tally is. Even at the rate of one or two such questions per week, it's not realistic to have a moderator tally up the votes. Harry is right, MO should be self-moderated so the proposed system should ideally maintain itself.

A lot of the point of MO is that it is self-moderated. This seems like a step in the wrong direction on posts that aren't contentious.

Another way of implementing it would be to have one CW answer that contains all the votes.

Wouldn't this break the "rule" that MO answers/questions are not to be used for meta MO purposes.

As a third-order effect, that change would probably be enough to make me want to be elected moderator the next time around. (I don't say "bonus", since probably not everyone feels that way.)

Another way of implementing it would be to have one CW answer that contains all the votes. For some reason I am not quite enthusiastic about this -- I worry that it will be too easy to miss this. Possibly a CW answer plus a single, eye-catching "look down" comment is a way to go?

Two options:

1. Once your "stay-open-vote" has been cancelled, delete your comment. Once you see that the "stay-open-vote" that you cancelled has been deleted, deleted your comment.
2. Give blanket permission to moderators (i.e. me, if I'm the only person bothered by this clutter) to delete cancelling pairs.

Thanks, now I know what's going on. Yes, I agree that the current setup makes it too easy for a question to get quickly closed even when the majority of the voters want it to remain open. Having a question get closed and then reopened (and sometimes reclosed!) is a lot of unnecessary drama.

I gather though that we are stuck with our current platform for the forseeable future, so however tempting it may be, it doesn' t seem to be productive to propose "rule changes" that cannot yet be implemented.

Brainstorming on what we could do with the current system, I came up with the following idea: vote trading. That is, if I see a question that I like but for whatever reason feel is in danger of being closed, I leave as a comment: "I cast a vote against closure." Then, the next person who would have voted to close, instead of doing so, leaves a comment saying "I vote to close, cancelling Pete's vote" or something to that effect.

Among other things, a certain amount of "honor" is necessary to pull this kind of thing off, and it brings some people's votes out into the open. But it might be worth a try...

Noah Snyder asked whether this procedure should take place in the comments to the question itself or should redirect to the meta site, pointing out that there are pro's and con's to both. If I had to guess, I would say that it would work better as comments on the question itself, but I don't see why I have to guess: we could try out multiple formats and see which, if any, is to our liking.