Tl;dr: Replace the migration feature with by-design simultaneous posting questions on multiple sites, with links.

The Problem(s)

Computer science is a broad field that touches many others. We use programming and mathematics. Our methods and results are used in programming, mathematics, biology, physics and basically any other science or engineering discipline. Naturally, many questions are right on the border between areas or lie firmly within both -- and such questions create problems for Stack Exchange. Which site to post on?

Currently, there are two "solutions":

  1. The deliberate cross-post
    Even lazy and deadline-driven homework-dumpers aside, many a question warrants being exposed to experts of more than one area. So, why not crosspost my statistical algorithms question with bioinformatics application on Computer Science, CrossValidated, Computational Science and Biology?

  2. The migration odyssey
    Being mindful of the network-wide "don't crosspost"-policy, you pick one of the sites your question might be a good fit for. Depending on your chosen focus and wording (and the mood of the reviewers of the day), your questions flies or -- and that's the issue -- is migrated to one of the other maybe fitting sites. Iterate until frustration peaks.

I'm sure computer science is but one example for this, but it is a good one: the sheer number of existing and proposed (on Area 51) sites dedicated to fields that are part of or have significant overlap with CS is conclusive proof that people are unsure where to pitch their tents with some of their questions. This effect is not limited to computer science, either.

When I browse the list of SE sites, I can imagine that, for instance, History has overlap with all the religion sites; Academia with Workplace and all the science sites; Mathematics Educators with the other sites (plural!) about mathematics; Physical Fitness with Sports and Bicycles and probably even Seasoned Advice; AskUbuntu with Unix & Linux and SuperUser.

Proposal

My proposal derives directly from this observation, namely that many questions fit more than one site. It is this:

Make it possible to (virtually) post questions to multiple sites at once and connect the instances.

In detail, this is how I think this would work (all subject to discussion):

  1. As has always been the case, user Jane selects site A she thinks fits her question. She chooses a title, writes up her post and selects some tags.
  2. Optional: A data-driven miracle occurs: the system detects that the question may fit more than one site! It proposes, "Your question may be ontopic for sites B, C and D; do you want to mirror your question there?"
    It'd be great if it could work (e.g. via tags that exist on multiple sites, comparing only such with existing migration paths?) but I'd rather not have it if it's crap.
  3. Jane thinks sites B and D might work as well (she didn't even know D existed!) and checks them; C is too far away in her estimation.
  4. One post is created and published on all of A, B and D.
  5. Moderator John on A comes up with a better title and 10k user Susan on D fixes a formula; these edits are shared across all sites. Tag edits remain local, of course.
  6. Community votes on B close the question as offtopic; the other instances are not affected but no longer link to the closed one on B.
  7. Moderator Lisa on D thinks the question might merit from being on E, too. She "migrates" the question, but decides to keep the D-copy around. The community on E decides to keep the question as well.
  8. Helpful users post answers on A, D and E, all of which are shown on all three sites (with an icon indicating their origin). Jane is excited; the combination of perspectives solves her problem more completely than any one site could have managed alone.
  9. Users reading the question on either site are exposed to other perspectives and learn, both about the subject matter and the other sites' scopes (which enables them to migrate better in the future).

As I said, discussion is definitely in order, and I deliberately do not touch some issues (e.g. which vote can be cast where and gives reputation where). However, I think that a feature that follows along these lines

  • solves the problems I outline above,
  • changes little in terms of using SE (closing and migrating already exist; the rest is sugar) but also
  • lets each community remain in full control of their scope.

But crossposting is bad!

Is it, really? The "official" FAQ accepted answer does not give any justification, and several commenters echo my concerns. The only real argument I have found so far is this:

Crossposting wastes time because of redundant answers and/or moderation actions.

My proposal mostly circumvents this; only site-specific moderation is duplicated (but it would also be with the close-repost workaround).

Please consider this and put an end to network fragmentation due to ever narrower Area 51 proposals. Let us embrace interdisciplinarity!

  • 1
    This proposal seems related but never got much of an response. – Raphael Jun 8 '15 at 14:37
  • think the crossposting issue on SE is old & related to another issue. nobody reliably knows what is going to "fly" on any particular se in the sense of not getting negative votes or closed. even mods! as long as there is a substantial chance that new posts get closed or negative voted, it appears that "quality questions" are in fact a scarce resource and random users are not a reliable/ consistent source of them.... seems even mods who are long/ well aware of other site scopes may find question migration an unpredictable exercise... – vzn Jun 8 '15 at 15:17
  • 2
    One very important consideration that would need to be thought of for the workflow for this is "How do you handle a question (from A) being closed as a duplicate on B (and C?)? Can you still answer on sites A or D?" – user213963 Jun 8 '15 at 15:53
  • Let us continue this discussion in chat. – Raphael Jun 8 '15 at 15:54
  • 1
    @Raphael I was reminded of an older post I did about a cross site feed: Add a limited cross site question feed to the main page that may help with collecting appropriate eyeballs on other sites while keeping the answers together. – user213963 Jul 20 '15 at 22:04
  • 1
    To the downvoters: why? If you think this is a bad idea, that doesn't make this a bad question! – reinierpost Oct 10 '16 at 9:44
up vote 9 down vote accepted

On users deciding what to cross post

tl;dr: Mods have a poor enough hit ratio. History shows an abysmal hit ration for topicality. You trust end users to say "yep, I'll post this on all the sites" and have it be on topic and up to expected quality standards?

  1. As has always been the case, user Jane selects site A she thinks fits her question. She chooses a title, writes up her post and selects some tags.
  2. Optional: A data-driven miracle occurs: the system detects that the question may fit more than one site! It proposes, "Your question may be ontopic for sites B, C and D; do you want to mirror your question there?"
    It'd be great if it could work (e.g. via tags that exist on multiple sites, comparing only such with existing migration paths?) but I'd rather not have it if it's crap.
  3. Jane thinks sites B and D might work as well (she didn't even know D existed!) and checks them; C is too far away in her estimation.
  4. One post is created and published on all of A, B and D.

My perspective on this is from a ~30 question/day site that is related to programming - Programmers.SE.

Programmers.SE and Stack Overflow have a bit of history with migrations from the days when it was an 3k close vote target. Even with a mod approving every migration, I'm still seeing 23% reject rate from Stack Overflow to Programmers.SE.

So theres the thing. Look at the migration stats and consider "what would this be like if users could do this with cross posting." No moderator in the way. Not even a majority vote from five community users who should have a better idea of the scope.

Guess what CS.SE, big-o had 59 posts this month, and every user will cross post those to CS.SE. Along with computation-theory, computation, complexity-theory, and of course, everything tagged computer-science (which includes everyone's computer science homework who doesn't know how to tag).

Just based on these tags, CS.SE would see a tripling of questions/day. Thats a good thing if they are good questions. If the quality of Stack Overflow continues to retain its current quality standards, well... CS.SE is getting maybe a dozen close vote reviews per day you will need triple that rate assuming that all the questions that SO gets are comparable to the quality that CS.SE currently gets.

On Programmers.SE, based on the quality of currently cross posted questions, we would have a significant crisis of community close votes being in short supply. Stack Overflow suggested "post this on Programmers.SE" can easily overwhelm the community moderation for the site (there are proactive measures for this).


Lets assume it is in place and most everything is well

Programmers.SE has different expectations for the quality and content of a post than Stack Overflow does. Our classic saying is "the whiteboard doesn't have a compiler" and we down vote and delete code only answers with surprising frequency.

Consider that question that is on topic for both Stack Overflow and Programmers.SE.

On editing

The question is meh, but with some heroic edits it can be fixed.

Programmers.SE has a culture of aggressive edits while Stack Overflow will reject the edit if it so much as touches the code. Who gets to edit the question? When a site does a massive edit to the content of the body because its not up to the standards and changes it to something that is in conflict with other site standards - what happens?

Comments

If you poke around MSO you'll find the occasional "where did all my comments go?" questions. Consider the additional moderation when an answer on one site gets its comments cleaned on another.

If moderators can't remove comments on other sites, consider the situation where there is "show 53 more comments" and you can't do anything about it.

Answers

Answers from Stack Overflow would show up on Programmers.SE and vice versa. And how many code only posts does Stack Overflow get in the first few minutes of an easy question?

So, now that post is visible on Programmers.SE, and it requires 10k or 20k or some super power to say "nope, that answer is crap and not up to the standards of the site." I've got nearly 40k rep on P.SE and I can easily see burning through all the delete votes in a day on answers in this case. We've got maybe half a dozen active 20ks who will cast delete votes... we wouldn't be able to handle them to say "nope, don't show this here."

Consider the rare occurrence that a question that is perfectly fits multiple sites. A question that is on topic for Stack Overflow, Programmers.SE, CS.SE, and Math.SE. Something about complexity analysis.

Programmers.SE and Stack Overflow don't have mathjax enabled. And we won't - it takes too much time for rendering and affects too few questions and answers. And Programmers.SE thinks funny greek letters look funny (and wander back to our IoC containers...). So, there's an issue that needs to be thought of.

You're going to get an answer from Stack Overflow of:

The answer is O(n) <!-- extra space -->

The answer you're going to get from Math.SE is:

Hint: you only go through the loop once

CS.SE will duplicate it to "Is there a system behind the magic of algorithm analysis?" and comment that this is probably homework.

Programmers.SE will either down vote it, close it, link Open letter to students with homework problems in the comments and delete it... or write an essay length answer on it (which has a 50/50 chance of having some logic errors in it depending on how many decades its been since the person answering was last in a CS theory classroom).

Having all of these answers together will result in a significant cultural clash between the sites and/or confuse readers about what is actually being answered on the site.

On bumping

Programmers.SE gets 30 questions per day. The active 50 shows about 12 hours currently. It moves at a rate where each question has some time in the sun. Consider how much activity Stack Overflow questions would bring to the site. What would this do to the quality design and architecture questions that we want to see here?

On closing on which site

Every site has its custom off topic reasons. Something that might be on topic on one site has been found to be problematic on another. Just picture every recommendation question getting cross posted with Software Recommendations (I hear the cringing from here)... and getting closed on SO and P.SE as a recommendation question (one of the defined off topic reasons). When a question is closed on one site, what does that do? This needs to be answered.

The other challenging one is for the duplicate question (I'll let mods familiar with the process also ponder merges). Does this prevent new answers on other sites? If not, do the new answers show up on sites that closed it already? How do you prevent the question from getting 30 more answers after its been closed on all but one site?

On topic for more than one site

I want to bring up a post by fbueckert (from a post on Gaming.SE): Are cross-site questions acceptable in rare cases?

The reason cross-posting is frowned upon is because it's the hallmark of a vague question. If it's broad enough to be equally applicable to multiple SE sites, you haven't refined the question enough. More thought needs to be invested to see which group would be best able to answer your question. You need to decide which site you think has the best chance of answering the question. If you don't manage to get an answer, feel free to delete and ask on another. That way, there's no duplication of effort, no bickering, and if you get an answer on another site, it will be applicable to those that view it.

Lets go back to that complexity question - it is probably too broad for at least one, or too specific for another.


Summary

  • Yes, migrations are a mess. However..
    • Mods do a meh job for migrations.
    • 3k rep users do an awful job
    • Letting 1, 101, or even 1001 rep users crosslink questions shows a very poor history based on existing cross posting anecdotes.
  • As soon as you include Stack Overflow, you overwhelm everyone else.
  • The culture of sites is too disparate for us to agree on the quality of answers or questions.
  • Closing needs to be handled and thought about
    • Custom close reason for different sites?
    • Duplicate on one or more sites?
    • Closed question still getting answers on other sites?

Between the architectural, cultural, and additional moderation tasks required I really don't see this being feasible.

  • You make valid points. I think some subset of your concerns can be addressed through proper design, but I don't see a way to do away with all at the same time. Too bad. So, what can we do? Maybe stop at linking duplicates/reposts on other sites in a nice way (say, icon + k answers), and don't try to show answers and comments on other sites? – Raphael Jun 9 '15 at 9:35
  • I disagree with fbueckert, by the way. If you can nail down your issue so well that it's ontopic on only one site, great. But sometimes, you can't see yourself whether you make invalid domain assumptions, pick a bad model, make a mistake in that model, or interpret the results wrongly. One of the experts at least will know, but you can't reach them in one place. – Raphael Jun 9 '15 at 9:36

You must log in to answer this question.

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