14

Could we know the details of how the current migration to ELL from EL&U works? I think that in general the migration path is a good thing. However, I'm a bit worried that there might be some counterproductive aspects to the way it currently works.

The impression that I get is that questions are being migrated as soon as 5 vote-to-close votes are assigned to the question. I also think that maybe some mods or high-rep users may be getting these questions migrated unilaterally or with another vote or two. In principle I'm not averse to this, but in practice, it seems to me that there might be a bit of a problem here.

The problem is that some questions might at first look like language learning questions, but may in fact be of huge significance and interest for linguists etymologists and serious English enthusiasts. In particular this applies to questions about grammar and syntax (yes, I know, syntax is part of grammar!). It can sometimes be very easy for people who care about the site, but aren't familiar with certain developments in modern grammar, or classical problems, or just interesting questions to migrate questions because a) they know intuitively what is grammatical, and b) because the OP is not a native speaker.

The kinds of questions that I'm worried about use losing are questions like What is a noun, What is a question? Can a word or phrase be a subject and an object at the same time?, What is an adverb? and so forth.

Now, when questions get closed unilaterally by a mod or get five down-votes, there is the opportunity for the community to vote, and if necessary appeal here on meta, to get the question reopened.

If a question gets migrated to ELL, does this happen instantaneously after it gets the necessary vote? If so, could this be changed so that there is an incubation period so to speak, where the community has a window of a couple of weeks to get the question reopened if it is in fact of value to us here at EL&U?

4
  • 3
    An "incubation period" would be a network-wide change (albeit possibly implemented with a flag on certain migration paths) so realistically the only way it would happen is by withdrawing the current migration path. Appealing against a wrong migration would be by flags on the created "target" post, I think, for mods on the target site to reject the migration. Rejected migrations end up as closed on the original site, where they can be edited and re-opened. [A comment, because I'm not sure if this is the answer.]
    – Andrew Leach Mod
    Nov 22, 2015 at 21:11
  • 3
    That said, your rescue of one recent question was masterful and has resulted in upvotes for the OP. Perhaps surprisingly, I don't think ELU has a similar question, and I decided I wouldn't summarily migrate the original on account of that.
    – Andrew Leach Mod
    Nov 22, 2015 at 21:30
  • If we're talking about network-wide changes, then all questions on this site should start out closed and require votes to get opened ;) Nov 23, 2015 at 3:21
  • +1. On A51.SE: 4. There's a high probability that users of site Y would enjoy seeing the occasional question about X on the FAQ page
    – AAM111
    Dec 1, 2015 at 4:45

1 Answer 1

9

tl,dr: An incubation period of several weeks wouldn’t work out very well because the original asker wouldn’t get a timely answer.


Migratory Mechanics

Questions immigrate to another site when four out of five non-binding close votes are for that particular migration.

In the normal course of things, this is perceived to occur “immediately”, but what really happens is that the question is immediately locked on the original site and its answers deleted (leaving a migration “stub”), and a message sent to the target site’s queue requesting that other site to create the migrated copy. Once that happens, the origin site annotates the stub to that effect with a pointer to the migrated copy.

This creation of the migrated copy by the other site usually happens so quickly that it appears to be all one action, but sometimes it isn’t exactly so. I’ve seen it take more than a minute for the queue to be processed, and in cases of network troubles, even double-digit numbers of minutes.

An incubation period of several weeks wouldn’t work out very well. The original asker might need an answer right away, even though it’s off-topic on the source site where it was. If it’s on-topic for the destination site, there’s no reason to have a waiting period for answers to appear — and if isn’t on-topic there, it shouldn’t’ve been migrated in the first place.

Migrated posts have downvotes cleared in the migrated copies. The reasoning behind this is that the downvotes on the question may have been incurred because of its perceived unsuitability on the origin site.

If the target site closes the migrated copy of the question, it locks its own migrated copy and sends a message to the site where the question emigrated from. It does not migrate new answers on the migrated question back to the origin site.

Once that original gets around to processing that notification message, it automatically unlocks (but does not automatically reöpen) the original copy and undeletes any answers it had previously deleted due to migration. This makes the now-unlocked original again eligible for reöpening under the customary rules and new answers being added once that’s done.

A locked migration stub is deleted by the system after some period of time that’s intended to be long enough for the original asker to follow the link to the other site; I believe this currently stands at a month out from the migration. However, if the stub becomes unlocked due to a failed migration, then it becomes subject to the normal deletion rules for closed questions.

All this locking and unlocking is there to make sure there is only one valid copy of a given question across the network.

Although upvotes on the migrated question (and any on migrated answers) go along with it, those votes are not attributed to you on the target site, and so you may vote “again” should it please you to do so.

That means you do have an opportunity to vote again on the migrated version (and any answers), even if you had previously voted on the other site. That’s by design: votes on one site do not count against the “one vote per post per user” rule on any other sites. You may also upvote any migrated comments despite having voted on them on their origin site.

3
  • Thanks for a detailed answer. Re " This makes the now-unlocked original again eligible for reöpening under the customary rules and new answers being added once that’s done": How does one get to cast a reopen vote on these? If I try clicking on a question that has (migrated) next to it, it automatically transfers me to the new site, not the unlocked original. Nov 24, 2015 at 13:35
  • Is it bad that I knew this was a tchrist answer as soon as I saw that diaeresis on "reöpen"? :)
    – Marthaª
    Dec 1, 2015 at 18:12
  • @Marthaª Did you mean 'that diëresis" on "rëopen"'? Dec 2, 2015 at 14:34

You must log in to answer this question.

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