Regarding the high number of rejected migrations from Stack Overflow to Server Fault

Question

Currently, there is a 40% (updated Nov 9, 2012) rejection rate on migrations from Stack Overflow to Server Fault.

Here are some samples of rejected questions from the last few days:

• This rejected/deleted question about Amazon EC2 for personal use.

• https://serverfault.com/questions/436794/unable-to-connect-2-virtual-machines-in-a-private-virtual-network-in-hyper-v

• https://serverfault.com/questions/436883/routing-domain-over-lan

The first example is from 2009 and already had an accepted answer (in addition to being bad and off-topic on SF). There was no reason to migrate it. The second two, while seemingly on-topic for SF based on the title, are borderline unintelligible in regards to the content. They contained no details, no relevant information, and amount to "Halp! Halp! Itz broked!"

I realize that there is a much larger number of users on Stack Overflow that are capable of voting to migrate and as a result, it takes 4/5 votes to migrate, while every other site is 3/5. This, unfortunately, does not appear to be an effective solution.

Questions should not be migrated based on their title, they should be judged on their content and topicality on the target site. This, obviously, is not happening as there is a 40% rejection rate on Server Fault.

What can we do to further educate Stack Overflow users about what is and is not on-topic elsewhere? Should SO have a higher reputation threshold than the other sites for vote-to-migrate privs? Should a migration require 5/5 votes instead of the current 4/5?

Answer 1

Update: The % of rejected migrations has been getting steadily higher for quite a while now. Coupled with the small number of migrations that actually result in good (defined as up-voted) answers, keeping this path seems increasingly pointless. ServerFault has been removed as a default migration target on Stack Overflow, replaced with Share Point.

Moderators can still migrate to SF on request, for those rare high-quality questions that are simply asked on the wrong site.

---

First off, just a sanity check: out of 824 migrations from Stack Overflow in the last 90 days, that 31% rejection rate means ~ 255 were inappropriate. During that same time period, over 12 thousand questions were closed as off-topic on Stack Overflow.

IMHO, 31% isn't insanely high period - but when you consider that those bad migrations constitute a paltry 2% of all off-topic questions on SO, you're a long way from being seen as a trash bin.

Also keep in mind: with the introduction of the /review system, the number of questions closed every day has more than doubled in the last few months - on Stack Overflow and Server Fault. This'll tend to drive up both the number of migrations and the number of rejections. It's also likely to be temporary: there's a gigantic backlog of questions on Stack Overflow with pending close votes, and as those are either closed or aged away, the number of closures should even out a bit.

Over all, I kinda feel like the system is working as intended here. That doesn't mean we can't do better, but I'm not inclined to look at this and think the sky is falling.

What can we do to further educate Stack Overflow users about what is and is not on-topic elsewhere?

That's a good attitude. IMHO, the best place to start educating is right there on the close dialog:

"Q&A for system administrators and desktop support professionals" may be accurate enough, but time and time again I hear complaints about questions that, while possibly server- or desktop-related in some fashion, don't really fall into the domain of questions you'd ask a sysadmin or help desk for assistance with. In other words, they're probably just fine where they sit (or if not, should be closed and improved on SO rather than shuffled off elsewhere).

The Server Fault FAQ begins with:

Server Fault is for Information Technology Professionals needing expert answers related to managing computer systems in a professional capacity.

(emphasis mine)

How about switching to that for the subtitle as well?

Answer 2

What about putting migrations in a review queue of some kind for that site's users to review and approve/reject instead of moving them instantly?

The people who know the site best are the site's users, so let them be the deciding factor on if the site gets a question or not.

Answer 3

While writing the comments above, I had an idea. Perhaps if one had to have a certain amount of reputation on the destination site in order to vote to migrate questions there?

Someone with (say) 1000 rep on Server Fault would have a pretty good idea of what the site's about. More so than people who've never been there, anyway.

Answer 4

I can take responsibility for about 15 of recent migrations that (should) have been rejected.

There was a user who decided to flag all Amazon S3 questions as Offtopic - Migrate to Server Fault, and there were a number of them that I didn't pay attention to the age of the question or whether it had an accepted answer. After about 15 or so I realized what was going on and started to decline the ones that didn't belong.

All in all I think that user flagged upwards of 50 posts to be moved to Server Fault that should not have been.

However, it looks like this user is still doing it; and another moderator accepted the flags. I'll make sure to spread the word among the SO moderators (again) about this user.

I just looked through the FAQ, and realized we don't have any guidelines for Migration. Or if we do,they don't have their own FAQ post and they don't show up on search. So at the risk of posting a duplicate, I posted a Q & A that spells out the guidelines for migration. Feel free to add to it or to add your own.

Answer 5

I believe all of these migrations are well-intentioned -- we want the askers to get answers to their questions.

The problem here seems to be a fundamental lack of understanding of:
(a) What makes a good, answerable question on SF ; and
(b) What is actually on-topic on SF

---

Item (b) is easy to fix: We spent a LOT of time revising our FAQ to address this.
I implore anyone migrating questions to read at least the first section of the FAQ to determine if the question is on-topic. If you haven't read the FAQ in a while you may be due for a refresher - it does change.

If you feel the question is on-topic based on the FAQ, item (a) merits some thought.
General guidance: If the question would be a good, answerable question on SO, it will probably be OK on SF (or other SE sites). If it's a lousy question we're going to kick it back and wonder aloud how 4 out of 5 people thought it was migration-worthy.
_{ProTip: If the question is deleted less than an hour after we rejected the migration it was probably too lousy to migrate :-)}

Specific guidance: We have a meta post about what it takes to write a good question.
(Much like the FAQ this can't be comprehensive - in fact it's based on the assumption that you have a specific type of problem, and not all good questions fit that mold - but it's a good starting point.)

---

When in doubt I suggest closing the question and leaving a comment for the OP pointing them at Server Fault (and the links above to the FAQ and Guidance for asking good questions). They can post a good question, or dreck, and we'll deal with it appropriately without all the angst of it being "Yet Another Bad Migration".

_{Note that I'm assuming the migrations being made are happening in good faith. If instead there's simply a subset of Stack Overflow users who are hitting the migrate button to make the question go away the only viable solution may be those shocky collar gizmos and 30kV shocks every time someone votes to migrate...}

Answer 6

To make a suggestion on my own question:

What if voting to migrate requires that you have the Analytical badge on the target site. That way, you have to have at least seen the FAQ. It won't be a panacea, but it could be a step in the right direction and it will guarantee that users know that you should only migrate topical things.

Answer 7

I think there's two elements to this, one is social and one is technical.

Between SU and SF, we occasionally have mods pop in to check if moving a question is ok. We have communication and we try to not move stuff the other site doesn't want. This is a GOOD thing - migration without thinking whether the other site wants the question is bad.

While its tempting to dump questions that 'don't belong here, and but I feel bad about taking them out back and shooting them' to other sites, its very likely it will happen. Think before you hit that vote to migrate button. When in doubt, it doesn't hurt to ask someone who is active on the site you are migrating to. And please, don't move really old questions with inactive users. If its unanswered, and unlikely to be answered, we're better off having someone re-ask the question.

On the Technical side I'd love to have a 'migration review queue'. Have users on the other site check through migrated posts before they get integrated into the site they are migrated to. We should know what belongs to our own site, don't we?

Answer 8

Since there hasn't really been an official response to this and we're over 40% rejected and still rising, I've suggested that Server Fault be removed from Stack Overflow's migration list.

I don't know if anything will come of it, but it seems that our most active mods are on board, as well as many of the high-rep users that are the ones vetting these migrations.

Answer 9

I think that we need to change SO so that it loses the marketing blurbs when it asks where to migrate, and replaces that with a list of site names and a short (two lines or three of text) description of what questions should and shouldn't go there. Actually with the capabilities of today's browser GUIs, you could hover over a potential migration site and see a much larger overlay with more info about what belongs and what doesn't.

Alternately, don't show a list of sites, instead ask a list of key questions about the topic of the post and automatically choose a site based on which questions are answered YES and which are answered NO. The question material is already there in the various site FAQs. Of course you can't just stick the FAQs up as they are, but the material in the FAQ can be used to craft some specific content questions. In this alternative, there would be no mention of destination sites, just yes/no questions about content and a backend decision table that sends the posts to another site. The same amount of screen space could hold a dozen questions and they could be crafted so that they all have a default "no" answer which means "discard this post". The user would click a few radio buttons to indicate "yes".

And the list of migration targets needs to be longer, i.e. there are other sites that should be in the list.

And lastly, the UI needs to make it clear that it is OK to simply vote for migration without specifying a destination.

The root of this problem is that the migration voter cannot be expected to have memorized all the current site FAQs and remember all that info when they vote. We need something that provides a reminder and these content questions would do that. They are like the questions used to key out a botanical taxonomy or to do medical diagnoses.