21

Currently, if you vote to close a question as off-topic, you're given a choice of a few sites that the question might be suitable for. e.g. on Serverfault, you can choose Superuser, Stackoverflow, Databases, and Unix/Linux If the majority of the close votes also say the question is suitable for one of the other sites, the question is migrated there.

As far as I can see, this semi-automatic migration causes more trouble than it solves, especially now that StackExchange has grown beyond the original SO, SF, SU trilogy. Some of the problems:

  • bad questions sometimes get migrated, rather than just closed
  • questions sometimes get bad answers before they're migrated
  • questions get migrated to the wrong site, e.g. to SF instead of SU Once that happens, they need moderator intervention to be migrated
  • there are so many overlaps between sites now that it's not possible to have every reasonable target site in the off-topic dialog, e.g. SF should arguably include Ubuntu and Security as target sites
  • people don't always follow their questions to the other site when they're migrated

Proposal

Do away with migrating questions. Instead, when someone chooses "off-topic" as a close reason, there would still be a follow-up dialogue, but it would have the following options:

  • off-topic (same as current)
  • may be suitable for Other Site #1
  • may be suitable for Other Site #2
  • may be suitable for _______ (close-voter fills in a site name)

If the question gets 5 close votes and 3 or more are "off-topic," it would be closed with a message saying it's off-topic. If the close votes include suggestions of other sites (either the #1 and #2 that are listed or any other site someone types in), then there'd be a longer note suggesting that the question might be appropriate on the other site, but the asker should look at those sites first before copying and pasting the question.

15
  • 6
    Also: when someone acknowledges that their question is off topic and belongs elsewhere, instead of waiting or flagging for migration they repost it there instead. The main issue with this is content duplication (double-posted question) and fragmentation (different sets of answers to the same question on each site), which itself is quite a strong case for migration over letting the asker repost. That said, migration is indeed a problem with certain sites that have high rejection percentages. Sep 23, 2012 at 21:59
  • It will also stop the "this is more suitable for programmers/dba/tex/unix" etc comments from resulting in multiple cross-posted questions. Sep 23, 2012 at 22:01
  • 7
    I kind of support this. The one exception should be migrating questions to their per-site Metas. That should always be automatic.
    – animuson StaffMod
    Sep 23, 2012 at 22:01
  • +1 This is a radical, but IMO clever idea.
    – Matt
    Sep 23, 2012 at 22:02
  • 5
    If anything, at least moderators and developers should retain the privilege to migrate anywhere, making it sort of a high-level operation similar to merging questions. Sep 23, 2012 at 22:03
  • @animuson oops, I forgot about that situation... yeah, if 5 people on the site pick migrate to meta.Site, migration is fine. Sep 23, 2012 at 22:03
  • 1
    @BoltClock'saUnicorn: I figured mods would keep the power. I'm more in favor of letting the destination site moderators approve/reject migrations before they actually get migrated. Like "Hey, 4 or 5 people thought this question belongs here, do you want to accept it?" There's a similar feature-req for that somewhere, but just IMO, that's a better idea.
    – animuson StaffMod
    Sep 23, 2012 at 22:05
  • @animuson: Agreed. Sep 23, 2012 at 22:06
  • @BoltClock'saUnicorn: Answering off topic questions implicitly encourages them. If something is closed off topic for site X that should be good enough reason to delete it. This solves the problem of content duplication.
    – user147520
    Sep 23, 2012 at 22:07
  • @animuson: This wouldn't need to be a moderator level privilege, there would also need to be feedback to the migrators so that they are educated.
    – user147520
    Sep 23, 2012 at 22:12
  • @Iain - moderators usually check with the other site before migration to try to make sure that bad migrations don't happen.
    – ChrisF Mod
    Sep 23, 2012 at 22:19
  • 2
    @ChrisF: That's not universally true we get quite a few bad mod hammered migrations on SF. We usually feedback to the relevant mod but it still happens. Being a mod on site X doesn't give you a better insite into what is on topic and relevant to another site unless you're also a power user there.
    – user147520
    Sep 23, 2012 at 22:20
  • 1
    Maybe migration should be made a privilege that has to be earned in a way different from mere rep? Although I can't think of a way to do this, short of quizzing people about how they would decide on this or that migration.... but I agree, too much terrible questions get migrated, often even by seasoned users.
    – Pekka
    Sep 23, 2012 at 22:30
  • I think there was a feature request somewhere to allow migration only to targets on which you, the user, are active. That might help as well
    – Pekka
    Sep 23, 2012 at 22:31
  • @Pekka: Maybe have a similar system to flagging, where you have a certain small number of migrate votes per day, which gets decreased if you vote for a bad migration or increased if your migration votes are consistently good. Sep 23, 2012 at 22:58

5 Answers 5

13

Are we sure this is a serious problem? I'm looking at the migration statistics for the last 90 days on Stack Overflow, and rejected migrations both to and from most sites are right around 10%. That means we're getting it right 9 times out of 10 on average. Do we really want to throw the baby out with the bath water?

Maybe migrations between other sites are worse so I'm just not seeing it?

I'm all for adjusting the migration mechanism if things can be improved, but I'd hate to see the total number of good migrations go down because we changed something that appears to be working.

13
  • 5
    I like the proposal, but having some real data always seems to have a way of surprising me. If 90% of migrations are OK, I doubt this is necessary. Sep 23, 2012 at 23:12
  • 5
    Just because 10% get rejected doesn't mean 90% deserve to stay. It just means they haven't all been caught yet. I'm a mod on SU and I very rarely see migrations from SO that don't suck. It's been said again and again that migration should only be for good questions that need a new home, not a catch-all for anybody who's misguided. As somebody who's sick of really bad questions being migrated without any of the 5 people even bothering to do basic editing before sending it, I think this is a very good proposal. Many SO users with migrate rights abuse them repeatedly and without consequence.
    – nhinkle
    Sep 24, 2012 at 1:50
  • The plural of anecdote is not data. That said, I see on Android Enthusiasts most of our inbound migrations are from Super User and fully a quarter of those have been "rejected". But we're talking about dozens of questions in the last 90 days, not hundreds.
    – ale
    Sep 24, 2012 at 2:13
  • @nhinkle When you see a question that shouldn't have been migrated, don't you close it? Any that you've seen in the last 90 days should already be counted in those statistics. Sep 24, 2012 at 2:17
  • 2
    @BilltheLizard The feeling of the "regulars" on SF is similar to what nhinkle says about SU: it seems like the majority of the migrations from SO are crap. There are a couple areas where the stats may be misleading: 1) SF has only about 250 3000+ users, so it's hard to close questions (e.g. the very slow decline of our close Review queue) 2) The # of migrated questions matters, I know I see a couple a day (and I read virtually all SF questions) Sep 24, 2012 at 3:55
  • 1
    @BilltheLizard Still, as you say, data is data, so maybe this isn't a big deal. Sep 24, 2012 at 3:56
  • @Ward It's possible the data is wrong, but then we have another problem. Now that people know what's being measured, we need to make sure these numbers aren't lying. If a question is migrated that shouldn't be, flag it on the receiving end for a moderator to close. Sep 24, 2012 at 11:14
  • @BilltheLizard: This gives people who do that a magic vote when it really should be a community decision and 1+mod isn't really the community deciding. Remember mods can (and do make mistakes) so having community involvement/oversight is important.
    – user147520
    Sep 24, 2012 at 11:46
  • AIUI rejected migrations only relate to questions closed TL/OT. We really need to know the total closed for any reason as NC/NARQ/Dupe is still a poor migration in reality. To get a full handle on it you also need to factor in things like did the OP reconnect with their question, was it answered on the destination site, did it get upvoted/downvoted and how well received were the answers.
    – user147520
    Sep 24, 2012 at 11:49
  • @Iain Dupe and OT are the only close reasons that don't count as a rejected migration. meta.stackexchange.com/a/126982/1288 (I understand duplicates not counting, but I'm not sure about OT.) Sep 24, 2012 at 12:27
  • @BilltheLizard: thanks. It says "belongs on another site" (but not generic "off topic") so I guess that means if a mod sends it on another migration (which we're asked not to) then it's not considered failed whereas a plain OT is.
    – user147520
    Sep 24, 2012 at 12:37
  • @Iain Oh good, then most of those should be counted correctly (if they do get closed). Still, getting people to act on the target site is key. If a high enough proportion of migrations are rejected, I'd recommend severing the link between two sites. Sep 24, 2012 at 12:48
  • 1
    @BilltheLizard: now there's a plan ...
    – user147520
    Sep 24, 2012 at 12:50
11

One situation this doesn't seem to address is when the question gets good answers before being migrated.

That is a dealbreaker for me. Assuming that the migration would be valid... a few things potentially happen:

  1. The question is closed as off-topic on the site it was originally posted on and is not reposted in the proper location, since the asker got their answer already. This diminishes the value of the question for future readers, since now it's closed and will never be reopened. It's effectively "done".
  2. The site that would host the question post-migration loses out on something that would be on-topic.
  3. We possibly end up with the same question in two (or more) locations with different sets of answers. This isn't helping future readers who might discover one site but not the other.

Also, if the migration would've been invalid and the suggestions are wrong, the asker's experience isn't much better - now they're going to repost a question somewhere they were told would be fine except that they'll get shut down there as well. (This already happens with the way things are set up right now, by the way. It's far from ideal.)

Last but not least, what about several possible suggestions for a better site? The asker then has to figure out where to go with their question. It's not a terribly friendly experience. Again, this can be/probably is an issue with our current system as well, but my point is that these are still things that have to be considered in any redesign.

Beyond that, like Bill the Lizard said, the vast majority of migrations are never contested. And ones that are, already result in the original question being closed as off-topic.

(As a side note, I'm curious to see how many people don't follow their questions to the other site when a migration happens. My perception is that it used to happen a lot more often in the past, but now we have notifications and easy account association in place, so it should be less of an issue.)

13
  • 1
    "I'm curious to see how many people don't follow their questions to the other site when a migration happens" - Hmm, interesting question. Should be easy enough to query for, although SEDE is a bit broken at the moment.
    – Tim Stone
    Sep 23, 2012 at 23:26
  • 1
    How do you know they are good answers? The question has been answered and voted upon on the wrong site. There are definitely issues with disproportionate and incorrect voting/incorrect answers when a question is asked on the wrong site meta.security.stackexchange.com/questions/862/….
    – user147520
    Sep 24, 2012 at 6:27
  • @Iain I don't know, necessarily, but it does happen. Usually on sites when there's scope overlap in the first place. This is sort of my point - we can't guarantee that we're only abandoning bad answers.
    – Adam Lear StaffMod
    Sep 24, 2012 at 13:20
  • If there's overlap, I don't see any of those 3 situations as a problem, or at least no worse a problem than the overlap already causes. Sep 24, 2012 at 13:29
  • @Ward Let me try again. There are sites (for example, SO and Programmers) where things off-topic on one are on-topic on the other, but the core audiences are essentially the same - programmers. So, migration still makes sense. I should've said "audience overlap".
    – Adam Lear StaffMod
    Sep 24, 2012 at 13:35
  • @AnnaLear: AIUI there were so many crappy migrations from SO to Programmers that they got themselves taken off the SO migration list. If the core audience for the two is the same then that's not a good advert for migrations at all and especially when the core audience is different.
    – user147520
    Sep 24, 2012 at 15:28
  • @Iain The issue with SO->Programmers migrations wasn't that the sites had the same audience. Programmers started out as an "everything that's off-topic on SO" kind of site and it took a while to shake that image. You can see how that would lead to a ton of crappy migrations.
    – Adam Lear StaffMod
    Sep 24, 2012 at 15:30
  • @AnnaLear I don't see your 3 situations as being dealbreakers in the case of sites that have some overlap in users and topics. For example, the "other" SF migration sites are dba and unix/linux. If a question gets some good answers and then closed on SF, it still gets indexed and if it really is a better fit for unix/linux there's a good chance that someone will notice it later and suggest a mod migrate it over. Security isn't a migration path for SF, but there's at least one security mod who's active on SF and questions do get moved around. Sep 24, 2012 at 16:15
  • 1
    @Ward: to be clear, the rejection rate wasn't the only reason Progse got blackballed. They also had fewer questions migrated to them than other sites which required moderator intervention to migrate! For all the talk about ignorant migrations, the folks on SO do tend to listen to reasoned arguments - rather than being the malicious action so often alleged in these discussions, it is more often an honest effort to salvage good but off-topic questions; no one wants to waste time and effort directing folks to a site that is overtly hostile to new users.
    – Shog9
    Sep 24, 2012 at 17:10
  • ... Educating the OP is the way forward, advising them to ask their question elsewhere and providing them with information about the other sites does just this. When they ask their question they get the benefit of 'Questions that may already have your answer' and the sidebar etc.
    – user147520
    Sep 25, 2012 at 10:25
  • I know, @Iain, it's almost like I read the rest of the comment threads here, and the 100 or so previous discussions on this topic before leaving that comment. Each time around, slightly more bitter. But y'know what's funny? No data to back it up. You've got the stats - as long as you're rejecting more questions asked directly on the site than those migrated in, it's kinda tough to argue that migration is doing such a bad job.
    – Shog9
    Sep 25, 2012 at 14:46
  • For the record, I'm all about less migration in general. But you're unwise to try following in Programmer's footsteps. If you're having that much trouble explaining your scope to folks who already know what you do, you have bigger problems than a few crappy migrated posts.
    – Shog9
    Sep 25, 2012 at 14:48
  • @Shog9: I commented on that blog post at the time, it was that that made me pretty much stop migrating stuff myself.
    – user147520
    Sep 25, 2012 at 16:57
6

I disagree that this is necessary for the simple reason that, as Bill notes, it would be throwing the baby out with the bathwater. But I applaud your simple and elegant solution - it's a refreshing change from the often-byzantine schemes proposed as knee-jerk reactions to a few sub-par migrations.

And I'll note that most of what you propose is easily possible today: just leave a comment noting which other site(s) might accept the question when voting to close.

There are even some handy shortcuts to help you out...

3

I would definitely agree that the problem is real, but think the solution that you propose is too radical. I think we can solve most of the issues by restricting the migration vote to users with sufficient reputation on the target site.

Migration is a process that touches two sites - the source site and the destination site. It would be logical to require anyone casting a "migration vote" to have a certain level of privilege on the target site. However, it is not required at the moment. I think this is the root cause of issues 1, 3, and 4 from your list: requiring users to have an "Established User" level* on the target site should significantly reduce, if not eliminate, these issues.

Issues number two (migrating answers along with the question) and five (not following the migrated question) are independent of the migration-by-vote problem, in the sense that it would be there even when the migration is done by moderators, so they should be considered separately.


* Meta site should be exempted from this requirement.

2
3

Speaking as a party of one, I'll offer that I virtually never propose migration. I've read so many variations on "don't migrate crap" that I've come to a conclusion: I'm not a regular on any of the target sites, so I'm not qualified to judge what would be welcome there.

The esteemed lizard's statistics are unimpeachable. I wonder, however, if the opposite problem is significant: people like me declining to migrate at all, leaving questions closed and marooned, as per Ms. Lear. I have no idea how to measure this; if significant, it would perhaps motivate consideration of one of the proposals in which off-topic questions sit in limbo but can be retrieved by other sites that like them.

2
  • It's fairly easy to get a list of closed questions in a given tag; if someone wants to watch, say, Stack Overflow for closed but not migrated questions in their area of expertise and flag for migration, that's certainly possible. And of course, leaving a comment for the author of a recently-closed question ("Hey, we can help you over here!") is a great way to handle decent questions without good answers.
    – Shog9
    Sep 23, 2012 at 23:30
  • You are very much in the minority. Many people migrate away based on what they think another site does rather than having any real idea. Often they have no rep on the receiving site at all.
    – user147520
    Sep 24, 2012 at 6:45

You must log in to answer this question.

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