With summer, the push to migrate returns (just like in those Attenborough specials).

Over a year ago, we had an extensive discussion about migrating to Stack Exchange 2.0. I remember feeling that just about everybody was in support of migration at the end of it. (I'm going to re-read the thread in case there are any important pitfalls I've forgotten about, but I haven't yet.) My feeling is that migrating will be almost entirely positive, though I expect this post will generate some fear about moving to 2.0. For what it's worth, aside from writing and running a really good Q&A engine, the SE team has been exceptionally generous with their attention and resources over the last 3 years. They want the MO community to be happy.

The main objection last year was that people really liked our current meta, but the folks at SE said that they've baked the SE-style meta into the framework. So we decided to work out some tools for dealing with a two meta system. Then academia season started and we all had to get back to work. Since then, I've become increasingly of the opinion that a pure SE-style meta is the way to go. While our discussion-style meta was certainly invaluable for hashing out community norms, I don't think there was anything in the last year that wouldn't have been well-served by an SE-style meta. Not only that, there would have been a huge benefit: proper integration with the main site. Cf. Dick Palais's comment here: "[T]he problem is that almost no MO users look at meta.MO so it is nearly useless to post it there."

The moderators and I pinged the Stack Exchangers about migration recently. It seems like there is no obstruction to migrating. Here's a summary from Joel Spolsky (SE cofounder/CEO):

Hi Anton!

Here is a summary of my current understanding regarding migrating MathOverflow.net to Stack Exchange 2.0.

1. The terms under which MathOverflow is operated will shift from the "Stack Exchange 1.0" model (under which the site is operated by Fog Creek Software as a service but the data, users, etc. are owned by you) to the "Stack Exchange 2.0" model (under which the site is a community within the Stack Exchange network, owned and operated by Stack Exchange).
2. We will upgrade MathOverflow to the latest software and join it to the Stack Exchange network.
3. Unless explicitly mentioned otherwise in this agreement, MathOverflow will operate like any other Stack Exchange site.
4. Current MathOverflow moderators will remain MathOverflow 2.0 moderators.
5. Before we finalize the migration, we will create a sandbox for you to test the migration. This will be a fully-functioning, fully operational version of MathOverflow running under the latest Stack Exchange software, which you can play around with and test before we have actually moved mathoverflow.net over. Any changes made in the sandbox will be lost when the real migration takes place.
6. The moderator team may submit additional Javascript to Stack Exchange which, if it does not compromise the technical integrity of the network, will be inserted into the footer, allowing some minor modification of the site that is unique to MathOverflow.
7. You will retain ownership of the domain name mathoverflow.net, but you will delegate the DNS operation to us.
8. Should you choose to migrate off of the Stack Exchange network:
   • We will provide the usual creative-commons data dump (which removes all private user information such as passwords and email addresses) complete as of to the migration;
   • We will return DNS control to you;
   • We will implement a system by which MathOverflow users can authenticate with our servers in order to reclaim their account on your new server.
   • Note that our privacy policy would not permit us to give you any user's email, password, or other authentication data if you are not an affiliated entity, thus, we would essentially have to get each user's permission on a one-time basis to transmit their credentials to you. In practical terms this could be as simple as a permission dialog that we present when users first attempt to log on to your server authorizing us to transmit the user's personal information to you.
9. If you don't already have one, I recommend creating a foundation, corporation, or not-for-profit that would own the mathoverflow domain name and serve as the counterparty. That way if something happens to Anton we know who is taking care of the domain name and who has the right to migrate out.

Does this sound like a workable plan?

I asked Joel if we could add a "no ads" term. He said that wouldn't be a problem. Aside from that, everything looks good to me.

I noticed that the tag badges do not seem to be implemented in the sandbox example site.

Perhaps it would be good to clarify how the user accounts will be transferred, in terms of reputation, badges and so on.

In response to

I know you said that the appearance is not final

and

It'd be nice to have a layout similar to the current one.

Just to reiterate --

Styling isn’t complete - we are using our beta theme for the moment until our designer finishes replicating and updating your current theme.

My understanding is that the layout/colors/etc will be as close to the current layout as possible.

I noticed that my account in the sandbox example has more answers than I currently have in my MO account (668 answers versus 661).

Perhaps there is a bug in the transfer process, causing some deleted answers to become undeleted? I'm not sure which answers are extra...

Edit: Perhaps the explanation is simply that whereas in the MO system, multiple answers to the same question are grouped together as a single entry, in the new system, they are listed separately. Several times I had given multiple answers to some big-list community-wiki questions, and this might account for the difference.

In addition to having no ads, any chance of getting rid of the +100 reputation bonus for associating another stackexchange account, and of keeping question migrations to MO from happening without MO moderator consent? (Maybe migrations already work that way in 2.0.)

@Henry Cohn: Different SE sites already have different cultures, and even different features (e.g. some sites have MathJax and citation links). While there may be some initial confusion/friction when joining SE 2.0, I don't see why MO would ultimately require more latitude than any other SE community. It may be that there are some legitimate areas where some SE feature is disruptive. In that case, I'm confident that the SE team will be receptive when presented with the supporting data. It's true that they really want to engineer more successful communities, but this includes academic communities. If there's something they don't understand about academia, let's help them out by explaining it. My guess is that the 100 point bonus and question migration will not be a problem. I certainly don't find myself wanting to go disrupt SE communities I don't belong in just because I'd get a 100 point bonus for doing so.

I agree that the discussion-style meta has mostly outlived it's usefulness.

I think that we forget that some of the people on MSE will be very against migration of advanced questions. I can come up with three or four of those people instantly. I will start a meta thread on meta.MSE to raise awareness of the upcoming issue (and it will rise...).

The main advantage, though, of the migration to SE 2.0 is that the same user can be used in the two sites simultaneously and experts can go back and forth to answer on MSE.

It may be that there are some legitimate areas where some SE feature is disruptive. In that case, I'm confident that the SE team will be receptive when presented with the supporting data.

Based on their history with other sites, I doubt it: when a feature is disruptive to an existing community, but they believe the feature will be an improvement in the long run as the community adapts to it or changes, they seem extremely unsympathetic. Maybe that was Jeff rather than Joel, in which case this would be less of an issue, but in principle I don't think we should expect the future will be substantially different from the past.

And even if Joel is the most accommodating person we could possibly work with, he may be under pressure in various ways. For example, other SE employees or VCs may argue that it's bad for morale in the SE network as a whole when they don't enforce uniform policies across the network. In that case, we're better off if Joel can say the company is obligated to treat MO differently, rather than having to make a business case for why it's a wise idea.

I think it's really important to negotiate well now, because we're never going to get another chance to negotiate over these issues. Once the transition takes place, we're not going to leave the stackexchange network any time in the foreseeable future. It's just too much work, and it's not going to happen unless they are really destroying MO (which presumably won't happen). We won't even be able to credibly threaten to leave. All we'll be able to do is to ask them to live up to the bargain we struck, or to renegotiate the bargain. The former is much better than the latter.

In particular, the line "Unless explicitly mentioned otherwise in this agreement, MathOverflow will operate like any other Stack Exchange site" is an enormous red flag. I strongly believe there should be a corresponding sentence saying something along the lines of "Except to the limited extent necessary to use a common code base, Stack Exchange will not attempt to impose consistency between Mathoverflow's culture and that of other Stack Exchange sites, for example on topics such as community wiki, comments, etc." Of course this is hopelessly vague, and there's no reasonable way to make it precise, but I think it's an important philosophical principle. They ought to be willing to add such a sentence, in which case why not? And if they aren't, then that's an extremely bad sign.

It's true that they really want to engineer more successful communities, but this includes academic communities. If there's something they don't understand about academia, let's help them out by explaining it.

They do have a vision that their network will be useful for academia, but it's utterly unrealistic. They seem to have almost no understanding of how academia works or what academics care about or value. Their goal isn't to adapt to academia or build tools for this purpose, but rather to attract academics to use the tools they already have (with a few minor tweaks if necessary). I don't think they are going to have much success with this, and I don't expect much flexibility.

My guess is that the 100 point bonus and question migration will not be a problem. I certainly don't find myself wanting to go disrupt SE communities I don't belong in just because I'd get a 100 point bonus for doing so.

It's not a matter of intentional disruption (the worst likely scenario is something like a batman question, which could attract a lot of voters from outside MO, but that wouldn't be a disaster). Rather, saying that all active stackexchange users should be allowed to vote things up on MO is philosophically wrong: it really makes sense to say you can't vote unless you've made some minor contribution to the community, and contributing to another stackexchange community shouldn't count. Fixing this would add only a trivial amount of work to the transition, and it's another thing where if the SE people refused to make this small change, it would be a bad sign.

I agree with Henry Cohn's concerns. In particular, for each way in which we expect to be treated differently from other SE networks in the future, there should be a corresponding written statement. If it is difficult to craft a precise statement, then a vague statement is better than none at all. To the extent possible we should not rely on the SE folks being reasonable. I also think it would be best if users from other SE sites (with sufficient reputation) were not automatically given MO privileges via a 100 point bonus.

I agree with Henry that SE approach to academic sites is completely broken. Namely you first must collect a large number of non-academic non-experts to seed the site and only after that can you move on to recruiting experts. Their science sites are generally not very good (Biology is the one I know best). The Theoretical Physics situation seemed to make it pretty clear that SE strongly prefers research questions be merged in with low-level questions.

That said, I just don't think the 100 rep thing will be a problem at all. Is it a problem at TCS? Another good comparison point is Mi Yodeya, which is a SE1.0 migration which is actually very technical, but which you could easily imagine attracting problem users from the rest of the network. But it really doesn't seem to. I think they're a very reassuring example, they were successful in SE1.0 and the migration doesn't seem to have resulted in SE trying to change the rules or culture.

Clause 3 is necessary and there is nothing wrong with it. They need us to work with all the explicit and implicit SE service agreements, unless we have special needs for which those agreements need to be amended. This is implicit for many contracts, but it doesn't hurt them to make it explicit. It's just a way for them to make sure that they can continue normal operations unless otherwise noted in the contract. Note that the exclusion clauses are very permissive. Carefully read clause 6, for example: it prohibits them to object to our code insertions for any reasons other than network integrity. Imagine all the crazy abuses they trust we won't do!

That said, there are good reasons to try to protect MathOverflow as much as possible. Right now, we have the option to jump ship. This is clearly only a last resort, but it's nice to have this kind of safety net. We could do a lot more, but this contract is not necessarily the best place to do that. One thing we could do is to ask them to make a real commitment to the advancement of science (or academia, or even just mathematics), perhaps we can even ask them to create a permanent scientific advisor position on their executive board (or whatever their top administrative unit is called) to promote and protect the needs of their scientific communities. Of course, there is no need to put that in this in the MO contract. In fact, it's probably better to have them do something like this in a highly visible place.

I'm not arguing against migrating - I'd be happier if there were another option, but moving to 2.0 is better than trying to stick with 1.0 until it becomes untenable, and there is no better option unless someone wants to put in an enormous amount of work and/or money.

I think part of the issue is that stackexchange 2.0 just doesn't appeal to me. I don't like the network or many of the other sites, I haven't liked the culture when I've participated in it, and I don't like the way I've seen the stackexchange team run things in the past. (For comparison, M.SE is working much better now than it used to be, but it has not been run nearly as well as MO, and part of this is because of active mismanagement in the past on the part of the SE team.) I think MO benefited enormously from being a stand-alone site that was clearly run by mathematicians and for mathematicians, and being tied into the network and subject to SE policy changes will make it look and feel less appealing to at least some mathematicians. I think MO has a big enough group of core users to overcome this, but it's worth doing everything reasonably possible to try to keep this transition from damaging MO.

Adding a sentence that you admit is hopelessly vague and unenforceable is purely a show of bad faith.

I don't see it as a show of bad faith. From my perspective, none of this agreement has much precise meaning except for the exit clause, and that's the only part where enforceability is really an issue anyway (nobody's going to be taking them to court over policies). I don't think the purpose should be to try to craft detailed rules for exactly how things should be run - that would be incredibly complicated and would end up ambiguous anyway. Instead, the goal should be to capture the spirit of the arrangement, make sure everyone's on the same page, and record what the expectations were. The idea isn't to make disputes impossible, but to minimize the chances of misunderstandings.

That's why Joel's version troubles me. What it captures is "MO will be run exactly like every other stackexchange site, except for grandfathering in old moderators, being able to add some javascript, and being allowed to leave the network in an emergency", which seems extremely one-sided to me. If I didn't think they would be flexible and understanding, I'd be extremely unhappy with this phrasing. I believe they will be accommodating in the short term, but I don't think we can count on it in the long run. Adding a sentence or two would at least record more clearly that MO is more independent than other SE 2.0 sites and should be treated that way.

One practical place where this matters is future SE employees. Eventually decisions regarding MO will be made by people who were hired long after the MO migration and have no idea what the back story or expectations were. If there's a written record, they will presumably do their best to follow its spirit. However, if the written record doesn't agree with the MO moderators' recollection of what they had in mind, then that's a problem. Maybe Joel could intervene and sort it out, but it's best to keep these issues from even arising by being clear.

As for academic communities, I think math is almost uniquely well suited to this sort of system, since one frequently runs into short, unambiguous questions that may be critical obstacles to one's work but might easily be answered by someone with different expertise. Plus there's a long history of problems and problem solving. This seems to line up with the limited data available so far regarding stackexchange: math works well, as do branches of mathematics that don't call themselves math (e.g., theoretical computer science), but even adjacent fields like physics already start to have difficulty building/maintaining a research-level Q&A site. In most academic fields, I don't think people want or would benefit from research-level stackexchange. One of the weaknesses of SE sites is that they tend to overestimate the degree of expertise involved (e.g., english.SE describes itself as being for "linguists, etymologists, and serious English language enthusiasts"), and I think this leads the people involved to underestimate the gap between hobbyist and professional sites (especially since CS is a field where this gap is particularly small). This may make the SE team more optimistic about research-level activity than I think is warranted.

Busy right now so cannot contribute to the discussion as much as I would like; but I wanted to pick up on something Anton says in his last comment. I don't like this distinction between high and low levels. My filter tends more to be whether the question is asked with some evidence of thought, effort or maturity. I'm sure I could concoct a question involving high-level definitions which nevertheless would be "too facile" for MO-as-I-would-like-it.

Indeed, some of the questions that have got my goat recently are probably from graduate students with poor training and under-developed problem-solving skills. (The recent question on amenability of tensor products comes to mind; compare it with one of Tom Leinster's old questions about proving in detail that commutative groups are amenable.)

This is getting a bit off-topic. However, I would be concerned if the planned upgrade led to more pressure from SO-central/collective on us, to refrain from closing things asked by people who should put some damn effort into thinking through their question. (Obligatory link to http://mattgemmell.com/2008/12/08/what-have-you-tried/ ) This fear may, of course, be unfounded.

Carefully read clause 6, for example: it prohibits them to object to our code insertions for any reasons other than network integrity. Imagine all the crazy abuses they trust we won't do!

Yes and no: my understanding is that they will still have no obligation to run MO at all, so if the moderators asked them to do something truly outrageous, they could shut down the site instead and blame the unreasonable moderators. Simply threatening to do so would be pretty effective, since voluntarily leaving would be much better for MO than being shut down.

One thing I would like to add that I haven't see mentioned is that MO will be advertised on other SE sites. For all intents and purposes MO is currently invisible to the SE network, and M.SE is their maths Q&A site. We will have to deal with people being confused and picking the 'wrong' site, i.e. MO, for their mathematics questions, including questions that are migrated from other sites in the network by well-meaning moderators. Perhaps in practice this will not be a major nuisance, but note we will have a little less people with access to moderator tools.

David, questions can only be migrated along established channels, we cannot not get questions migrated from random SE sites where moderators are unfamiliar with MO's scope.

@Francois: That is not true. Users can migrate only in open channels; moderators can migrate everywhere.

@Tom - that is a hard one. StackOverflow has a special lock for questions of this sort, see for example http://stackoverflow.com/questions/3748/storing-images-in-db-yea-or-nay. A moderator can lock the question (instead of closing it as mods do here) and the text

This question exists because it has historical significance, but it is not considered a good, on-topic question for this site, so please do not use it as evidence that you can ask similar questions here. This question and its answers are frozen and cannot be changed. More info: FAQ.

appears. This would be good to apply, after migration, or indeed in some automated way given a list of question numbers, to all the usual big list suspects:

http://mathoverflow.net/questions/23478/examples-of-common-false-beliefs-in-mathematics-closed

http://mathoverflow.net/questions/63221/ideas-on-how-to-prevent-a-department-from-being-shut-down-closed

http://mathoverflow.net/questions/22299/what-are-some-examples-of-colorful-language-in-serious-mathematics-papers-clos

http://mathoverflow.net/questions/53122/mathematical-urban-legends-closed

etc...

I wanted to point that so far it seems that many of the users on MSE are strongly opposed to migrating even research questions, without an explicit request of the OP.

http://meta.math.stackexchange.com/questions/4880/

I'm pleased to see that you don't have an account on TeX-SX so I can except that site from your comments about the SE network as a whole - with which I broadly agree except in so far as they concern the TeX site

Yeah, I probably should have phrased things more carefully (although I did say "many of the other sites"). There are a few sites I really like, for example TeX, where the only reason I've never asked a question there is that all my questions have already been asked and answered. And even among the sites I find frustrating, it's more a matter of whether they appeal to me than whether they are intrinsically valuable (for example, math.SE seems to be providing a valuable service to many people).

Emil, actually it's not broken -- it's fixed. :-)

Just the hacks from MO are not needed, and therefore not working anymore.

@Asaf: I am perfectly aware that this amounts to a fix of the situation for new posts. However, it breaks a lot of old posts, so it needs to be handled more carefully by the import procedure.