Vanilla 1.1.9 is a product of Lussumo. More Information: Documentation, Community Support.
@Anton: You say that “There are two inseparable changes that come with a move to SE 2.0”. Why are they inseparable? Did you ask SE to separate them and they refused? Did SE make any specific statement to this point? I would be very happy if we could just upgrade the software from SE 1.0 to SE 2.0 and not join the SE network. Would it be possible for you to clarify this matter also?
“has a more idealistic spirit”: Perhaps, but we must also take into account the possibility that SE will be bought by a company with a very un-idealistic spirit, like Microsoft or Facebook. The story of LiveJournal very clearly demonstrates that the company that takes over can easily violate the explicitly stated social contract (no ads in the case of LiveJournal), and there is nothing that could be done about that. Similar stories happened in the past with mathematical journals. The only way to protect from this is via an actual legal contract between SE and MO foundation.
“SE's privacy policy dictates that they cannot share these private details with third parties.”: There are many ways around this situation, for example users can be presented with an additional privacy notice that they will have to sign in order to register on MO.
“Is there something currently owned by you or me which would be owned by SE after migration?”: Currently it is very easy to take the entire MO database (both public and private stuff) and move it elsewhere. This ability will be lost under the new terms. Thus this ability (call it “ownership” or any other term you like) is exactly what is being transferred to SE. With significant effort we might be able to crawl the MO site and extract the public data, but we will never be able to get the private stuff. I hope this also answers your question “explain how moving to another platform would become significantly more difficult after migrating to SE 2.0”.
“I hope this makes it very clear why the comparison to Elsevier is severely flawed.” I was fully aware of how MO content is licensed when I wrote previous comments and I disagree with the “severely flawed” part. The point of my comparison was not to illustrate the copyright issues (which are indeed very different in these cases), but to show that moving can become very difficult in the new situation (i.e., if we just have a social contract not backed up by an actual legal contract).
“What is not "free" about it?”: free as in freedom, not as in free beer. Sure, SE does not demand payments from its users, but there are many other aspects in which SE is not free. One of them you already cited: ads. Another one is persistent attempts by the SE management to micromanage its communities, very often unsuccessfully. (Yes, I know there are promises by them not to intervene with MO, but again these are impossible to guarantee without an actual legal contract.) In other words, “free” means the freedom to completely control the content of a web site. SE lacks this type of freedom. [I guess LiveJournal in its early stages could be cited as an example of a more free platform: no ads, no micromanagement, etc.]
“What is not "public" about it?”: The fact that tomorrow some third party entity can take over SE and suddenly make things very different from what they are now: put lots of ads everywhere, demand payment from its users etc. It will also be under no legal obligation to provide public database dumps etc. [Thus LiveJournal in its early days was not a public service, for example.] There is also a different sense in which SE is not public: public vs. private universities.
“nevermind that he has been given the task of caring to our technical needs, not our political ones”: I responded to his comment that was purely political in nature and had no technical content.
I still don't understand why it would be more difficult to "reboot" on a new platform if we migrate to SE 2.0. Please keep trying to explain this to me!
I would be very happy if we could just upgrade the software from SE 1.0 to SE 2.0 and not join the SE network. Would it be possible for you to clarify this matter also?
I would be happy to do that too, but there is no such option. Stack Exchange has made it very clear that the network aspect is baked into the new engine. They are only willing to migrate us on the condition that we join the network. Otherwise they'd essentially need to develop a separate product, which they're not willing to do even if we pay them.
Becoming part of the network entails compatibility with the rest of the network. Though I'd like it, I think a special user agreement for MO would set an unacceptable precedent for SE.
The only way to protect from this is via an actual legal contract between SE and MO foundation.
I thought that's the whole point. The conditions of the migration to SE 2.0 are supposed to make it possible to migrate to another platform in case SE becomes evil in some way.
With significant effort we might be able to crawl the MO site and extract the public data, but we will never be able to get the private stuff. I hope this also answers your question “explain how moving to another platform would become significantly more difficult after migrating to SE 2.0”.
There would be no need to crawl the site to extract the public data. Dumps of the public data are provided regularly. Even if they stop providing regular dumps of public information, clause 8 clearly states that SE would provide a such a dump if we migrate off the SE network.
It does not answer the question. Given that SE would be contractually obligated to allow users to recover their accounts, I don't really understand why it would be more difficult. Why is it important to know things like the originating IP of all actions? So long as you can match people to users, and identify who has voted on what, how is private data necessary to "reboot" the site on a different platform? It seems like a simple one-time dialog box "Please re-enter your email address" recovers all the important private data.
I was thinking about the "pull the plug" option and came to the conclusion that actually there's no real danger here. We don't really need all the history of who voted for what. We could just create a frozen version of MO at the point of "jumping ship" and then restart with a whole new platform and a whole new database. If wanted, it would be possible to transfer everyone's reputations at freeze point. MO has enough momentum that that would work, I think.
I'll miss the forum format for meta - the SE format is almost completely unusable - but we seem to use it less now than in the early days so perhaps it has served its purpose. On TeX-SX we tend to use the "chat" for what we used this forum for - checking if our initial reactions to a question were right and things like that. It's harder to look back at discussions and to thread them but then we have less controversy on TeX than MO would have. But that's no longer the deal-breaker that it used to be for me as I find myself using MO less and less so I'm curious as to whether the migration will make MO useful again.
“SE would be contractually obligated”: So there are plans to prepare an actual legal contract (not an informal declaration of intents) that will explicitly provide for public and private data recovery in case MO migrates to another platform?
In this case I withdraw most of my criticism, except that we need to be very careful about private data (individual votes, emails etc.) that must be made available if/when we migrate.
“Stack Exchange has made it very clear that the network aspect is baked into the new engine.”: I am still not convinced. I bet it's very easy to create a new separate network of which MO is the only participating site. This is essentially how MO works now.
“Even if they stop providing regular dumps of public information”: Is there a reason for this? Theoretically it can be extracted by crawling (putting much more stress on SE servers), so why not provide it directly? Public data dumps provide an opportunity for data mining, which will be lost in case dumps are not available. Can we include a clause about regular dumps of public data in the contract?
Andrew Stacey:
I was thinking about the "pull the plug" option and came to the conclusion that actually there's no real danger here. We don't really need all the history of who voted for what. We could just create a frozen version of MO at the point of "jumping ship" and then restart with a whole new platform and a whole new database. If wanted, it would be possible to transfer everyone's reputations at freeze point. MO has enough momentum that that would work, I think.
Except there's no way for people to reliably and automatically claim their accounts from Stack Exchange without Stack Exchange's cooperation. The public dumps don't include email addresses or passwords.
In any case, I agree. In the unlikely "Stack Exchange turns evil" scenario, a seamless transition would be nice, but switching to a new database seeded with the publicly available data would be a relatively minor discontinuity, in my opinion.
Except there's no way for people to reliably and automatically claim their accounts from Stack Exchange without Stack Exchange's cooperation.
Right. But in my model, that's not needed. The SE version of MO is "frozen" in that we copy the publicly-available data and create a static site with all the existing information but in a non-modifiable form. Then we start a new MO on a new platform.
After all, the only reason for really wanting to maintain continuity is so that people can maintain their reputation. But who really cares about that?
After all, the only reason for really wanting to maintain continuity is so that people can maintain their reputation.
No, that's not the point. It's to maintain continuous operation of MO. Freezing MO at some point and restarting it from scratch would work, but that's not continued operation. As Anton pointed out earlier, under the current terms, there is still a certain amount of data that would be missing without which some transition glitches would occur. We are currently working on a way to fix that.
No, that's not the point. It's to maintain continuous operation of MO.
No, that's not the point. The point is that if SE do turn out to be Evil Overlords(TM) then it's not as dire as it would seem that people are making out. MO really could survive being restarted: it has the community and (I judge) the loyalty of the community is to MO, not to SE, so if MO had to restart the community would follow and put in the effort needed to restart it. What would really harm the mathematical community would be if we lost all the archive material, but that is what is safeguarded by the licence and the public dumps.
That's the lesson of Elsevier: don't let anyone hold us hostage by holding on to our prior work. We did the work. It's ours. We allow SE to publish it on the web, but if we don't like the way that they do it we can legally republish it elsewhere. All the "private" data is immaterial to that: as I have no access to it now, I wouldn't miss it if we had a "frozen" version of MO that we could simply refer to: all the actual mathematical information would be there. And if I had a new question I could ask it on the new site.
So I see no danger in migrating.
I think we're talking about slightly different things; I don't think we actually disagree anywhere.
I'm talking about the formulation of clause 8, which is designed so that we can continue to operate MO without too much hindrance in the event we leave SE after transition. (There are a few little things that have been added since the initial draft, e.g. regaining our twitter handle. We'll keep everybody up to date in due time.)
You seem to be talking only about the possibility of leaving SE after transition. Yes, that is a possibility even without anything like clause 8, even with significant loss of user data and other stuff that would prevent continuous operation of MO, since we have legal rights to the post data regardless and we are the user base, which is what MO is all about. I also agree that the MO community would probably survive such an event without critical wounds. Of course, that doesn't mean we shouldn't have a clause 8 at all.
So there are plans to prepare an actual legal contract (not an informal declaration of intents) that will explicitly provide for public and private data recovery in case MO migrates to another platform?
Yes.
“Stack Exchange has made it very clear that the network aspect is baked into the new engine.”: I am still not convinced. I bet it's very easy to create a new separate network of which MO is the only participating site. This is essentially how MO works now.
I guess I should have said that they've made it clear they aren't interested in that sort of arrangement.
I do wish I understood why SE wants to only have a single network. It seems to me that there would be big advantages to having a few networks (e.g. an academic network).
Perhaps this has already been brought up, but mathoverflow has an extensive, well-thought-out section on "how to ask" that is, I believe, also linked to from the Theoretical Computer Science stackexchange faq. The sandbox does not have a separate "how to ask" link; the current faq does have a section on this, but it is much shorter (even after clicking "show more") and not at all specific to mathoverflow. Will we be able to keep the "how to ask" section we currently have, and in a way that is easy to get to and does not break existing links?
Perhaps this has already been brought up, but mathoverflow has an extensive, well-thought-out section on "how to ask" that is, I believe, also linked to from the Theoretical Computer Science stackexchange faq.
The How To Ask page is a testament to the quality of our early meta discussions, squabbling notwithstanding.
+1 Harry (squabbling notwithstanding)
mathoverflow has an extensive, well-thought-out section on "how to ask"
It hasn't already been brought up (this year), but it is one of the points we brought up in our email chat with SE. They said that it won't be a problem to incorporate our FAQ and "how to ask" pages into MO 2.0. We should be able to either replace the stock how-to-ask page with ours, or have a separate static page.
We should be able to either replace the stock how-to-ask page with ours, or have a separate static page.
Anton, is there any reason why this couldn't be done for meta? SE2.0 meta is absolute garbage, and even this stripped down forum is a dream to use in comparison. If this is something that the SE team sees as a deal-breaker, could we at least have an explanation from the SE team about why it is so inconceivable that we might be allowed to keep our meta?
I already feel that the SE team's insistence on our abandonment of meta.MO borders on the kind of bad faith that others have been talking about. It just seems completely arbitrary, like we're being forced to use SE2.0 meta because Jeff Atwood refuses to admit that SE software makes a crappy discussion platform (objections of SOBlog fanboys again notwithstanding).
Incidentally, Jeff Atwood no longer works at Stack Exchange. My impression is that Jeff Atwood's style was somewhat more interventionist than the rest of the Stack Exchange moderators, at least on MSE; the other members of the company generally did not participate on MSE beyond occasionally answering questions on the meta site. I think that if MathOverflow migrates, there would not be much reason to expect interference from SE moderators (e.g., in meta) that might be contrary to MO's culture.
@Michael: Yes, that's what I'm talking about. However, this raises the following question: Where should I go to discuss mathoverflow policy, rules, guidelines, how to ask, etc. with other people? Right now, I go to meta.MO, where we can comfortably have this conversation and read posts in chronological order.
I see that you've used SE's native meta. Is it any good for a thread like this one? Of course not.
I mean, take a look at the meta.math.SE thread about migrating questions to MO and try to follow what's even going on. There is a discussion going on that spans multiple answers, and it's not really clear in what order they should be read. Contrast this with the MO discussion, which I was able to pick up having skipped nearly the entire second page without any trouble.
Harry: Actually, there's a fairly simple solution to that: community policy should be (or should have been) that on meta, comments should only be placed under the question itself--not under answers.
While I agree that the meta.math.SE thread has ended up as something of a mess, I also found it quite useful to be able to see the votes on different answers. The comments are made by only a relatively few people, while the votes give a much better impression of the position of the community as a whole. In reading a meta.MO thread such as this one, it is hard to tell with any reliability where the community stands on migration to SE 2.0 and which of the comments actually are important to more people than just the commentator.
Charles, Vanilla supports a voting module, if that's something you'd like to see. It's also worth noting that SE meta requires a certain number of points to post on it, which means that the function of meta as a place to ask if a question is okay is severely limited in the case of new users.
The downsides to an SE-style meta are well-documented, and the existing advantages could easily be implemented on vanilla (I can only think of two: OpenID login and voting-on-posts).
However, I'd argue that the votes we see on meta.mathunderflow do not represent well-thought-out positions. A voter need not have read the arguments against the position expressed in the post. Going by this sort of "at-first-glance" voting completely removes the deliberation and discussion aspects of decision-making and can actually be rather harmful.
(Vanilla being the name of the software that this forum uses)
Harry, while old timers like you and I have very fond memories of great meta discussions that really helped shape MO in the early days, there is much less need for a discussion outlet these days. Over the last year, it became clear that this type of discussion oriented forum was less than ideal and sometimes even counterproductive to resolve day-to-day meta issues on MO. Though it is by no means a panacea, we have seen many ways in which a Q&A style meta would be a better tool. When we started thinking about migration again this summer, the unanimous opinion among moderators was that keeping our meta was no longer essential for a successful transition and we would be happy to try the Q&A style meta. Of course, there will still be a need for discussions like the many we had here, but we are confident that we will be able to find a way to have those when we need them even if they are not accommodated by the SE platform.
We plan to have a strong creative impact on the SE network once we become part of it. The SE team has always been receptive to our ideas, we plan to keep it that way and make good use of that. My personal feeling is that we will always get what we need when we need it, even more than we do now, after we move to the SE network.
@François: All the more reason for the coexistence of an SE-style meta together with this discussion forum. For instance, at the top of the page, have the usual links, (FAQ, meta, How to Ask) together with a link to discussion.mathoverflow.net, which could just be this board renamed. If it really ends up not being used for anything constructive, then I'd be fine with seeing it go. However, I think that if we don't insist on maintaining this board in one form or another (with a link at the top of the page), we might end up in a situation where we can never get it back. If we insist on at least coexistence right now, we leave our options open.
It's really not a lot to ask. I don't see why a transition to SE2.0 makes it necessary to completely abandon meta.MO.
The option to have both was discussed last year. While the idea had gained some support, the problem is that having two main meta outlets for MO would probably be more confusing than beneficial overall.
@François: Even if we called it discussion.mathoverflow.net and linked it as "Discussion"? I am really uncomfortable with giving up this meta without having at least a chance to get it back. If it's not in writing from the SE team that we can bring back a discussion-style meta/forum if the admins request it, the SE team will probably give us the run-around. I'm just saying, we're better safe than sorry.
This meta will not suddenly vanish when we roll out MO2.0. We will keep it as a reference and we could keep using it for some other things. What to use it for and how to link it to the main site are very good questions. Do you want to start a new thread where you and others can pitch ideas? I don't think this is the ideal place for this since it's not a critical transition issue.
Would it be possible to have a certain, reasonable amount of space in the mathoverflow.net domain that we could use however we want? Limiting the amount of space would hopefully prevent us from overtaxing the SE servers with extraneous material, while having the space would allow the moderators to put up things that do not really fit within the SE format, and link to it from e.g. the FAQ. [The material might even be stored on non-SE servers, if this is technically possible.]
Such extra space would ensure that MO has the option of bringing back this version of meta (as a supplement to the SE-style meta) at a later time, without having to maintain it meanwhile "just in case."
We will have to delegate DNS operations to SE, though we will still own the mathoverflow.net domain. I don't see why they wouldn't want to add a few DNS entries for us (provided there aren't any clashes with their normal use). I'm glad you brought that up Charles, we may want to clarify item 7.
Harry, +1 at-first-glance for
However, I'd argue that the votes we see on meta.mathunderflow do not represent well-thought-out positions. A voter need not have read the arguments against the position expressed in the post. Going by this sort of "at-first-glance" voting completely removes the deliberation and discussion aspects of decision-making and can actually be rather harmful.
But I think this applies to non-meta just as well as to meta (one example being the obvious bias of votes towards early answers). Which is a good reason to look for a real solution to this problem, not just a temporary fix, now that there is an opportunity for a serious change at least in the meta. That could be a single system that includes both voting and an opportunity for deliberations, rather than a combination of two types of metas as you suggested in another thread.
So how to devise a system that forces, or at least strongly encourages the user to review and digest all opinions before casting a vote? There should probably be research about this in political sciences, http://en.wikipedia.org/wiki/Online_deliberation - though I didn't have the patience to penetrate through general words to find any concrete online examples. But to begin with, one could imagine some alternative voting systems (which need not be any harder to use than the existing one). Say, if you vote for an answer at an early stage, the value of your vote could "fade out" as new answers appear, proportionally to their quantity. Probably the fading should not affect reputations, but be only comparative to other upvotes. Then you could still upvote another answer later, which would count at 100% value at the moment, or perhaps reconfirm some of your earlier upvotes to bring them back to the full value. Of course, this is just one possibility, one could also experiment with the Australian system (presumably the user who has to rank all the answers is more likely to read them all), etc.
@Sergey: I think that our chances of actually getting the SE overlords to change the voting system would be rather difficult, since it would require them to do work. While I do agree that your ideas would improve the voting system, I don't think that we have a realistic chance of getting something like that implemented. I think we actually have a realistic chance of having the overlords put a link to our discussion meta, which is why I'm focusing on that.
Can we please try to avoid inflammatory language like "overloads"? I'd like us to try to get things going on as much of the right foot as possible.
Someone needs to correct Noah's spelling of "overlords." (Incidentally, I agree with him completely--even if you have a highly negative view of the SE administrators [which I do not], antagonizing them does no one any favors.)
@Noah: "SE Overlords" is old old old terminology and is meant tongue-in-cheek. It is part of the same joke as calling Anton "our benevolent dictator" and Ravi "our mysterious benefactor" (I think this was it, at least).
Have a look: Click
Regardless Harry, it's pointlessly inflammatory.
Come on, why does everyone always assume that I'm trying to be nasty?! That was obviously not my intent.
The term also appears on other SE sites: Click
It doesn't matter what your intent is, your comment is certainly going to be read that way by a large number of people.
A few points regarding migration on SE2.0 and some other issues:
There are two kind of migrations on SE2.0: by moderators and by users. Moderators can migrate questions anywhere, however the usual practice is to request the opinion of the moderators of the target site before migrating a question. We have not had any migrations to cstheory that has not followed this practice recently, I don't think moderator migrations will be a problem for MO. Regarding the migration by users, when a users votes to close a question as off-topic the systems allows the user to select another site where the question might be suitable. However the list of sites that can be selected is limited, the default setting is just to the meta of that site. A site can ask SE to open a migration path to another one for users but AFAIK this also needs the approval of the target site's community. In short, if you don't want migrations from another site that will be the case.
Requiring moderators to confirm all migrations can be problematic as explained here, it can create too much work for moderators, particularly between the trilogy sites which have a huge number of users/questions. (ps: The feature request question is still open and it seems they haven't made the final decision regarding it).
Another issue that you may want to discuss is that the votes on a migrated question is transfered from the source site to the target site. There is an open feature request to reset the number of votes after migration.
AFAIK, we haven't experienced any real trouble because of users from other sites coming our the site.
Over the last two years SE employees have allowed the community to run the site as it wants and have not interfered with our decisions. They have posted a few times on our meta about SE's ideas, but that can hardly be called interference IMO, (and most of the time the ideas were welcomed by cstheory community, e.g. 1, 2). They get more involved in the new sites that need help and advice to get running, they don't get involved much on established sites. So experiencing any interference by SE on how you run MO2.0 is unlikely IMHO.
I think the structure of MO's meta is kind of similar to SE2.0 chat rooms which is designed for discussions (you can create a new chat room each time you want to start a new discussion topic). You may want to test to see if it can be the new place for discussions.
From my experiences at TeX-SX:
I think the structure of MO's meta is kind of similar to SE2.0 chat rooms which is designed for discussions (you can create a new chat room each time you want to start a new discussion topic).
Nothing like it. The chat is what we use in place of the forum and it certainly has its place and is very useful for getting a quick consensus of a few users before acting. We also use it for special "answer the unanswered" sessions to dig up old posts that need attention. The big drawbacks of chat over a forum are related to the lack of organisation: it's difficult to trace back a conversation or to record some decision. Creating new chat rooms is no replacement as I doubt many people browse the current list of chat rooms to see what's going on (and most have inspiring names such as "Chat room for X and Y.)
Over the last two years SE employees have allowed the community to run the site as it wants and have not interfered with our decisions.
Fairly similar on TeX-SX except that I suspect it's more to do with the fact that none of them know we exist! There was a video posted where members of a certain team within SE tried to name as many sites as possible. Only one mentioned TeX and even then the video editor missed it and didn't cut it into the right place in the sequence. But we have had occasions where the SE team have stepped in and where we didn't like it.
Slightly more seriously, I think we get away with our oddities because we don't cause much trouble (moderator flags get seen by the main team so they can keep an eye on the overall balance of the sites - ours is particularly low) and because tex is seen as being a bit ... odd ... by the programming community. No-one understands tex who doesn't use it.
My point is that it is very hard to extrapolate from one site to another as to how much the local flavour is allowed to flourish. If SE feel that it isn't harming the site, they'll let it go. If they feel that it is, they'll step in. But this is probably where their lack of understanding of academics counts against us. I realise that the experience of cstheory is probably more applicable for MO than TeX, though.
@Andrew Stacey,
I agree: your comments regarding both issues are more accurate than mine.
Here's an updated letter from Joel (split over two posts because of the character limit). He said he'll have it converted into legalese for us. We're currently looking into the possibility of borrowing the legal team of some math-friendly organization to look over everything for us (if you have some suggestion, feel free to email moderators@mathoverflow.net). As far as I can tell, all the important issues are resolved in a way that makes (almost) everybody in the MO community happy.
A few comments:
Dear Anton:
Since MathOverflow opened almost three years ago, it has turned into an amazing resource for the world mathematics community. We recognize that what's needed to drive success for an academic site such as yours is different than what's needed for some of our other sites, and it is important to us to preserve and protect that resource.
We look forward to upgrading MathOverflow to the latest version of the Stack Exchange software and bringing it into the Stack Exchange network of sites. As we discussed, there are several changes that this will bring about.
- 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 Inc, with data governed by a Creative Commons license).
- We will upgrade MathOverflow to the latest software and join it to the Stack Exchange network.
- In most respects, MathOverflow will operate like any other Stack Exchange site. As with all Stack Exchange sites, cultural issues such as moderation approach and site norms (such as appropriate use of comments, etc.) will continue to be determined by the MathOverflow mods and community.
- Current MathOverflow moderators will serve as the initial MathOverflow 2.0 moderators. From time to time, and after consultation with existing moderators, we will hold elections for additional moderators as the need arises.
- 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.
- In general, Stack Exchange Inc. endeavors to make software that meets the needs of all its users. However, in certain instances, MathOverflow moderators may wish to make reasonable adjustments to the operation of that site by means of extra client-side javascript. Thus, they 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 reasonable modification of the site that is specific to MathOverflow. Unlike SE1.0, the SE2.0 HTML is changing all the time, and so such JavaScript will need to be actively maintained or it may stop working. It is understood that this script maintenance will not be the responsibility of Stack Exchange Inc.
- It is understood that the MathOverflow community in general desires an ad-free environment but might be open to certain announcement services like job listings in the future. Thus Stack Exchange will not run advertisements, including internal "house ads", on MathOverflow, without specific and advance permission of the moderators.
- The DNS operation of mathoverflow.net and .com will be delegated to Stack Exchange but owned by you.
We are acutely aware that other commercial organizations, like journal publishers, have not served as good stewards for the mathematical knowledge with which they were entrusted, and, even though we are a for-profit entity, we are determined to ensure that will not happen to MathOverflow. Since the inception of Stack Overflow our corporate ethic has been a categorical belief that the content on our website was created by our users and is owned by our users, not by us. That's why all user-generated content on our network is governed by the Creative Commons license and we make frequent data dumps available.
In that spirit, we understand that the current stakeholders of MathOverflow would like to maintain an escape clause, allowing them to move MathOverflow's content and users out of the Stack Exchange network at some point in the future, at their sole discretion. Because such a migration might be technically tricky, this letter will serve to memorialize some of the principles that we have agreed upon in advance to enable it.
We hope that you and the MathOverflow leadership will create some kind of foundation or not-for-profit corporation which can serve as the owner in trust of the mathoverflow.net and .com domain names, and which can serve as the counterparty that decides, under its own by-laws, if and when it is necessary to migrate out of the Stack Exchange Network.
In the event of such a migration, we will provide you with a complete and current database dump which contains all the data necessary to recreate MathOverflow on your own servers and software. (We will obviously not provide any servers or software, just the data). This provision notwithstanding, due to the privacy policy we maintain with our users, we cannot provide you with any personally identifiable information that is not already visible on our site. Such private information includes, but is not limited to, the user's email addresses, passwords, voting records, IP addresses, and authentication data. In order to enable a smooth migration, we would build a simple system that would allow each user to authorize us to transfer their personal and authentication data to the new organization upon their first logon.
We have enjoyed our small part in supporting what you've been able to do to advance the state of mathematics through MathOverflow, and look forward to continuing and strengthening the fruitful relationship we have had.
All the best,
Joel Spolsky
CEO and Cofounder, Stack Exchange
One Exchange Plaza, 26th Floor, New York, NY 10006
I see two potential issues with this agreement:
The mention of voting records as private information that will not be made available to us is particularly disturbing. What will happen to votes on all questions? Will all of them be reset to zero? Will voting on older questions be allowed? Can't we arrange for the voting records to be owned by the MathOverflow foundation? Specifically, the SE privacy policy states “if such a day should come, we will first give you the opportunity to explicitly consent (opt-in) to such disclosure or to any use of the information for a purpose other than the one for which it was originally collected or previously authorized”. Can't we require MO users to opt-in to allow their voting records to be shared/transferred to the MathOverflow foundation?
“we will hold elections for additional moderators as the need arises”: What does this mean? Why can't we continue to hold moderator elections without any involvement of SE as we did before?
I don't see it as an urgent issue, but I do think that in the long term it would be good to follow their suggestion of creating a MathOverflow foundation or something. Another question I have (probably one for the lawyers, if you can have some look at this) is what would happen if SE goes into bankruptcy or something? Can we count on being able to use the escape clause under such circumstances? (I certainly don't mean to suggest that I think such a scenario is likely, but the whole point of the escape clause is to be prepared for the unexpected.)
@Dmitri: experience on other SE sites says this about your second question.
Remember how we last voted for moderators? It was kinda hectic, because there were no established infrastructure for it.
SE2 has such an established infrastructure. Except that the switch to activate the machine is only within the reach of SE. So basically, if we were to want more moderators, we just need to agitate on meta until one of the staff goes over and pushes the red button to start the election process, at which point a well-defined procedure will come in to play: a nomination phase of (some number of) days, a primary phase (for the first round of voting; only needed if the number of candidates are large) for (some number of) days, and the final vote round. More info here. (I am not sure whether the 150 rep barrier includes or excludes the 100 bonus for associating the account; I know that for certain functions on SE that 100 additional reputation does not count.)
Also implicit in "as the need arises" is the fact that every year SE Inc will e-mail the current moderators asking them whether they are overworked and whether they need more hands on deck.
@Dmitri: Perhaps the following clarifications will help; otherwise please clarify what the potential issues are.
The vote counts for each post are completely public, so there would be no need to reset anything to zero, and voting on old post will work just fine. The only reason we would need access to the vote owners is to make sure that nobody is voting twice on the same post. If you re-read the stuff after clause 8, you'll see that Joel proposes exactly the solution you are suggesting: if we migrate off the SE platform, they will set up a system whereby MO users can opt-in to transferring all their private data. We can simply require users to opt-in if they wish to continue using their old account.
The SE moderator election mechanism is way better than our dinky ad hoc system. Why would we not want to use it? It's certainly the case that new moderators are needed from time to time. I can only imagine this clause being objectionable if it is read with a super sinister eye, in which case basically every one of the terms becomes a landmine. But I think that requires Humpty Dumpty arguments like "when we said 'election' we meant that we'll appoint somebody from outside the community."
@Mark: Yes, I'd like to discuss the formation on a legal entity to own MO with whatever lawyers we end up talking to. I've also thought about what would happen if SE were to go under. Unless it both turns evil and goes bankrupt, I don't think it would be a huge issue. Presumably there would be some warning, since a non-evil entity wouldn't simply destroy the work of several dozen communities overnight. We could start the "opt-in for private data transfer" operation as soon as possible. A couple of months should be enough to get all the regular users in, and moderators could reasonably collect the emails of stragglers with non-trivial account activity so that we could identify them as the owners of their accounts in the future (though they'd have to fill our their profile pages again).
@Anton: So “personal and authentication data” does include individual vote owners? This is not clear in the current formulation.
What if somebody registers a new account instead of transferring the old one and starts voting on the old questions for the second time?
I don't have anything about the new moderator election mechanism per se, I just don't understand why we should relinquish the control over the “red button” (see Willie Wong's comment above) to some outsiders. Why can't the MathOverflow foundation control the “red button”?