Should we get rid of the "cross-posted" close reason?

Question

TL;DR

Should we remove the "cross-posted" close reason and instead flag for mod attention as they do everywhere else on the SE network?

---

Cross-posting on multiple sites of the Stack Exchange network is frowned upon. The general feeling of the community seems to be that having the same question on multiple sites leads to duplicated effort and wastes everyone's time. Perhaps. However, we take it to a whole new level.

As far as I know, U&L is the only site of the SE network with a close reason specific to cross-posts:

As a result, our users tend to close cross-posted questions on sight. I have a number of problems with this approach:

• New users have no idea that cross-posting is discouraged. How could they? It's buried in the help pages and is very easy to miss even if the user in question has actually read the help.
• Consider the case of Ask Ubuntu (http://askubuntu.com), Super User (http://superuser.com) and U&L (http://unix.stackexchange.com). There are many questions that can be on topic on all three. How can I, as a new user, unfamiliar with Stack Exchange, know that those three are part of a single network of sites? They all have different UI styles, completely different URLs and different titles. It is very easy to assume they are independent sites. Yes, they share some similarities but so do http://biostars.org and http://ask.debian.net/ neither of which is affiliated with SE.
• Most importantly, I don't like losing questions. If a question has been cross-posted on multiple sites, is on-topic on all of them and is a decent question, one of them should remain open. Because of this close-reason, that is almost never the one that is here!

Now, I already mentioned that other sites don't have such a close reason. Back when I first joined an SE site, I had asked what the correct way of dealing with cross-posts was. One of the mods on SU answered thusly:

• It's usually enough to just say that it's been cross posted. Don't forget to include a link to the other version(s), so we can take a look ourselves.

• If you want to be super helpful, tell us whether the question has already received an answer on the other site(s) it's been posted on. In that case we can usually close it or even delete it here.

• If it doesn't have an answer on either site, please only flag on one site. We'll deal with contacting other sites' moderators and resolving the issue. Alerting moderators on all sites will probably lead to even more confusion. In the worst case all instances of the question would be closed.

I still think that is the best way of dealing with cross-posts. For example, this question was recently asked here, on AU and on SU. Because it was cross-posted, it received close votes and downvotes and flags on all three sites. That is messy and unhelpful to the OP and would have resulted in all three being closed. The question here had received a good answer so I contacted the mods on AU and SU, asked them to close their version of the question and then I re-opened the one here.

Granted, that is not a particularly stellar question but it is useful. It should exist on the SE network and the current system of closing cross-posts blindly would result in its loss network-wide. I feel that flagging is a much better way of dealing with cross-posts than close-voting. If the post is flagged, the mods can

• Check which of the cross-posts is most on topic on its respective site.
• Check whether any of the cross-posted questions has received an answer.
• Check whether any of the others has been closed.
• Contacts the mods of the other site(s) in question.

And decide which to close in each particular case taking into account the particular situation. With all this in mind, I suggest we remove the cross-posted close reason. Instead of immediately closing cross-posted questions, we should leave a comment like:

Cross-posted here. Please don't post the same question on multiple sites of the Stack Exchange network. See here for more information.

Then, flag the post for mod attention, giving us the link to the cross-post. We can then coordinate with the mods of the other site to choose the best home for the Q.

So, what does everyone think? Should we get rid of the close reason and flag/comment instead like everyone else? Or do we hate cross-posts so much more than anyone else on the network and really, really need this close reason on U&L?

Answer 1

To address your concerns individually:

New users have no idea that cross-posting is discouraged. How could they? It's buried in the help pages and is very easy to miss even if the user in question has actually read the help.

People that cross-post the same question across multiple sites generally have little interest in reading the help pages anyway: they just want their question answered, and answered immediately, and the most expeditious way to acheive that is to put it, word for word, in as many places as possible because, like dropping dynamite in a pond, it is far more likely to get you the damn fish, irrespective of the side effects.

And there are side effects to this sort of behaviour. It isn't just about the this site generating more questions (see below), it is about the impact upon the community of volunteers that make this network of sites what it is. Cross-posting is an expression of selfish disregard for the time and effort that the community put in to maintaining these sites; it is a practice that should be, as the help centre notes, actively discouraged. And by actively, that means that, we the community, not just moderators, should have the ability to make that decision.

Most importantly, I don't like losing questions. If a question has been cross-posted on multiple sites, is on-topic on all of them and is a decent question, one of them should remain open. Because of this close-reason, that is almost never the one that is here!

While I understand this position, I think in this case, it is misguided. The sorts of questions that are cross-posted, because of the mentality that drives that behaviour, are almost always of poor quality. The sense of entitlement that leads someone to spam every community they can with their important question, is the same sense that renders them incapable of reading the documentation, searching for other answers already posted or taking the time to read the help centre pages. They are so busy farming out their question and then furiously hitting refresh with a sense of mounting indignation that people are actually ignoring their life threatening issue that they need to find yet another hapless community to inflict it upon...

In short, we have nothing to fear from closing these questions on all the sites they appear on: we, as a community aren't in danger of losing some precious resource, we are actually contributing to the greater good of the FOSS ecosystem by helping to establish and maintain some basic standard of civility and regard for the longer term good of our collective efforts.

So, my view would be that it is important that the ability to close these questions remains in the hands of the community, and is not solely left to the discretion of the moderators (fine, upstanding people that they are).

Answer 2

U&L is the only site I know of where multi-posting¹ is one of the moderator-defined close reasons. However most other sites do close multi-posts. In the old days, we used the badly-fitting “too localized” reason (on the basis that the copy of the question here is not useful to anyone since it has answers on another site). Nowadays, we use the worse-fitting “off-topic” reason with a custom comment.

The one site I'm aware of that doesn't have this policy is Mathematics (which tends to accept multi-posts as long as all copies link to each other). Its research-level sister site Math Overflow does have a policy against multi-posts.

We should close all but one copy of multi-posts as fast as possible so that people don't waste effort writing similar answers independently on different sites. Waiting is the worst possible reaction. I agree with slhck: users should be encouraged to vote to close in addition to flagging. (I do not encourage downvotes for multi-posts, however, unless the user doing it has already been told not to in the past.) If the question ends up closed on both sides (which rarely happens, usually moderators react faster than the question can get 5 close votes on both sides), a moderator can easily reopen it.

This doesn't need to be a moderator-defined close reason, but I would only support removing it to make room for a different moderator-defined close reason. We have a policy against multiposts and should keep it.

¹ _{Technically, this is not cross-posting, since it isn't the same post appearing in multiple locations but identical yet unlinked posts.}

Answer 3

I don't have much of an opinion on how the cross posts get closed, as long as it is fast, before it generates the reintegration work associated with multiple answers on different sites.

My first thought was to suggest including something in the "Welcome" text that you get when you post your first question (not sure if that is true and you get this text, I just followed a link in a post on meta.stackexchange and applied the path to our site), but as others indicated people just click through that.

However there is IMO another approach: it is easy for me to check if someone cross-posts if they use the same account on multiple sites, and more difficult, but still possible, if they do so using multiple accounts (by matching title, content). To do that one can write a program to look up new questions by following the profile link of the question's OP to see what other (recent) activity there has been on other, subscribed, sites and give notification of possible cross-posts. Matching titles across questions on all SE sites independent of poster would be more expensive but not impossible. The delay of a question showing up on google seems so small you could even use that.

With the extra information that is available to the application observing new posts within the StackExchange network (easy database access on OPs recent activity, but checking other activity from the same IP should catch most cross-posters using multiple account), checking for cross posts should be relatively simple and not very expensive. Based on that analyses:

• the OP could be warned
• the OP could be altogether prohibited from posting the question that is too similar to a question somewhere else (or even similar to one the same site by the same OP, we have those every once in a while as well)
• moderators could be notified
• An automated close vote could be cast, with a comment "possible cross-posted" so reviewers can directly check.

If we need a proof of concept, I can offer looking into creation of a client side tool for that. I can have the tool add answers to some special meta-post created for this purpose, for potential cross posts found, so everyone interested can see there how well this works (instead of directly putting comments on real questions).

Answer 4

Since we are talking about cross-posting among the sites in the SE network which necessarily share the code and inner structure -

- why not do a provision in a UI for the cross-posting that would legally merge all the sites' answers and form a interdisciplinary question?

Let the scoring be independently counted on all sites (upvotes/downvotes) but with points that are an arithmetic mean of the points given on each site. This way the scoring system will automatically discourage publishing the question on sites where it would get low scores and give no bonus to the act of cross-posting.

Such questions will obviously be visible on all SE sites into which it is cross-posted, but with references to the other sites as well. And each site would conveniently merge all answers together, but only local answers would be active.

I think most of this functionality can be rendered on the client side. Think about it as a layer of Greasemonkey scripts that are applied when the question is flagged as cross-posted. The only time the sites from the SE network will have to contact with each other is to calculate the common scoring rule I proposed.

I know it might be a challenge to program this. But don't say it is impossible. And we should strive for the best, right?

And if you still have your doubts - think about this very question. I bet there is more than one SE community that has this "cross-posted" flag. Why not it be shared on all of these sites' meta to get a collective discussion?

---

Being an another interdisciplinary person (physicist turned-statistician who also administer a cluster of his own Ubuntu Linux servers and fluently program in several languages) I too often have a feeling that tailoring a problem into a question that fits exactly on one SE network really hurts its potential. And maintaining the proper cross-posting etiquette is tedious and time-consuming.