@DL: Anton is currently discussing the issue with SE further. We should wait for an update from him.

@Asaf: I really don't understand why this is an issue.

• If all you care about is knowing what someone's "real reputation" is, you can figure out how likely it is that they got the boost based on their network profile.
• The privileges you get from 101 starting reputation are of no particular importance with the possible exception of CW editing (and that can be abused by users who "legitimately" have 100 reputation just as easily). Voting on MO is already broken.
• Anyone who is "not suited" to MO will probably find that out very quickly and leave anyway regardless of whether or not they get a +100 boost.
• Anyone who is "not suited" to MO and won't want to leave won't be particularly discouraged by not having the +100 boost.

Giving users from elsewhere on the network the ability to leave comments is a boon to moderation; without that ability users will frequently leave answers instead, which will then need to be cleaned up.

To provide one single data point (so with absolutely zero significance) for Kaveh's comment:

This is the currently most viewed, and highest voted, question on Math.StackExchange. You may be interested in some figures.

But Memes tend to be short lived and sporadic, so over all I don't think they can cause that much of a problem. Afterall, no one is seriously using question vote counts for any statistical analysis. In regards to the image a highly voted question may create for a site, in hindsight, I agree with G Edgar's admonition in the Meta.Math.Stackexchange thread I linked to: those type of popularity are windfalls for mathematics as a whole.

And yes, the migration problem. I remember reading here and there (something on Meta.SO) discussions about whether it is kosher to migrate something that has been open for quite some time. One time when this discussion happened was during the infancy of TeX when a lot of questions with ridiculously high number of votes were mass-migrated from StackOverflow. I don't recall whether any thing ever came out of that discussion, so I don't know what the current SE policy about that is.

Actually I think the batman question is a success story for Math.SE. The best answer is really really nice, is the highest voted answer, and is more highly voted than the question. The daily reputation cap assured that the asker's reputation didn't go high enough for them to cause any real havoc (and the user hasn't tried to cause any havoc).

If we migrate to SE and we get a question like that then we'd just migrate it over to math.SE. There's just no problem there.

If not having accounts associated is a thing you value, what's wrong with using different OpenIDs to do it? Automatic association reflects a philosophy that the different sites in the SE network shouldn't be completely unrelated, and if you don't want to use them that way, then you might as well use different OpenIDs, right?

Is it true that an edit to your profile on one SE 2.0 site necessarily propagates to all of your other SE 2.0 profiles? If no, then I can't imagine the objection to the accounts being associated, since it would be almost completely behind the scenes. If yes, then I think we can make a reasonable case for a feature allowing a user to prevent such propagation. It is completely reasonable to want to present different information to different communities.

I have now realised that under the accounts section, one can disassociate SE accounts, so they shouldn't appear in one's network profile. Hopefully this works permanently (I'm not quite confident), but as this appears to be the case, this particular artifact of the SE system that I brought up shouldn't be a problem.

I've nothing to say that hasn't already been brought up, so: for the people interested in a blow-by-blow account of changes made to the SE engine, here is a frequently(?) updated changelog for the SE engine, for those who want to know whether the added features are going to be hunky-dory for MO or not.

Some remarks:

Cons:

1) I view it as a big disadvantage that user profiles get linked on the SE network. I do *not* want to mix my professional life with any other activities (which would include an otherwise fine answer to an "interesting" question on skeptics, say, which will easily have more votes than any math answer and be listed first in my profile). (I actually view this as a weakness of SE that one cannot easily separate work and play, I realize that people a bit younger than me have grown up with facebook, but I suspect that many of them are going to wake up when they apply for jobs.) Note that if someone used the same email address to open an account on MO and SE, I am quite sure that they would be linked automatically even though someone might have used their real name on MO and a pseudonym on SE. Obviously, this issue can be dealt with by using different email adresses, logouts etc, but essentially, one has to try to subvert the idea of the network that *wants* to mix professional and personal aspects of life. It is not as simple as just clicking disassociate by now and if you associate something by accident, then it is linked, as I suspect that the migration will do.

2) MO is like a math journal that is looking for a new publisher, but the offer is to take over the *name* of the journal and the implicit right to change the editors with the understanding that there is no plan to exercice this power. I have no reason to believe that this particular publisher has any bad intentions, but that is no reason to sign ownership away. Obviously, this point can be addressed by the "escape hatch", but it is without any doubt a point that will be a concern to many mathematicians.


3) A more fundamental problem is that the philosophy of the SE network is depersonalization: By delegating reputation to points, it is possible to trust information by strangers. The philosophy on MO is personalization: A pressure to use realnames, for example (which I am against, but that is not the point). Even without realnames, people here have at least pseudonymous personas with reputation, typical for a forum with a humanly comprehensible size. The SO users enjoy this exchange with strangers, a bit like exchanging your apartment with strangers for holidays or flashmobs, but it does not go well with a research community. Many SO users do not understand at all why the research-oriented sites don't want to do what is necessary to grow really big. It is a bad idea to interpret this culture clash as hostility.

This issue is much more fundamental than just what Jeff may or may not think. MO as a community is only interested in half of the SE software: Using votes to do some sorting, but the reputation on MO comes largely from real-life reputation or direct impression from the answer quality.

I think that a lot of the problems on MSE can be tracked to this issue and in some ways, it is an issue of SE Inc. that they do not (yet) know how to reconcile their philosophy with research communities because they personally think that it is ok if they are just mixed up with everyone else. They have not yet figured out how research communities can work on their network, and this might be an incentive to invite MO because there is a high probability that MO will survive minor problems. MO will be a guinea pig.

Pros:

1) If the guinea pig phase works, MO will have some influence on the standards.

2) Yes, 2.0 is much better.

3) SE 1.0 will not work for 20 years, either, so it is not a good argument that SE2.0 might not be a 20 years solution. We might migrate to holodecks by then.

Conclusion: If you can negotiate a possibility to leave, then it is an interesting thing to try, but there *will* be problems, especially along the M.SE/Math.Overflow boundary. Many soft questions work well on MO because there are only MO users answering.


Note that the software problems are not really problems. There are *programmers* on SE, lots of them, there are greasemonkey scripts for anything you want, sorted by votes, including disabling of the enter-comment function. And you are not even allowed to say thank you :).

I strongly disagree with the negative view on Jeff Atwood's personality. Given the numbers of interaction with communities and the circumstances, a conflict like this was very likely, regardless of personality.

And finally: I am annoyed by the "we don't want a democracy"-lanugage. If I don't want to have citizens of other countries decide the public transport policy in my village, then this is not undemocratic. And there was a time when a policeman could certainly not remove a hobo from the math common room because policemen were not allowed to enter universities at all without special permission. And a small enough community should decide as much as possible by consensus not just by vote, because everyone who has been in a committee knows that it depends on who formulates the text voted upon.

The blog post by Jeff Atwood seems to encourage posting a question and immediately answering it. This is not as much of an issue as a quiz question. (And we already had one of those http://mathoverflow.net/questions/71092/how-many-integer-partitions-of-a-googol-10100-into-at-most-60-parts, which caused quite a stir http://tea.mathoverflow.net/discussion/1091/how-many-integer-partitions-of-a-googol-10100-into-at-most-60-parts/.)

Any particular policy is subject to abuse or to use beyond and contrary to the original intent of the (principles of) the forum containing it. If this particular policy were promoted on MathOverflow, here is how I might (ab)use it.

I would start posting a mix of questions, some of which I knew and some of which were related to the ones I knew, but had not worked through, whether for laziness or lack of cleverness or what have you. Those who paid attention would see some of the struggles and successes with my current dabblings on Jacobsthal's function, whether they be in the literature or not. Then I might resurrect my work on the Hadamard matrix conjecture, then on Frankl's union closed sets conjecture, followed by hyperidentities and other work related to Murskii on finite basis problems. I figure each of these topics to be good for 10 posts at least; depending on my strategy, I might let half or more of them be questions to which I know the answer.

My intent would partly be to circumvent (or totally alter) the process for doing graduate level research; I tackle the stuff I know I can handle, and the stuff "I don't feel like doing" I post on MathOverflow. If I don't get shut down by the moderators, I amass enough material for a dissertation or two and present it to my advisor.

The discussion of how ethical this is (or not), how it changes the role of the advisor (who might be compelled to check the references and attributions made or not made in the submitted draft of the dissertation), and how it affects doing research can be saved for other threads; I think this is a lousy way to make a mathematician. It takes away some of the struggle I feel is necessary to build one's abilities. (Supporting anecdote: I recall the time I proved to myself that the real numbers were separable and how this could be used in forming sequences of functions that were used in regularity results of Leray on Navier-Stokes; never mind that I did not really understand the PDE course I was taking nor that I had not yet taken topology; I'll always remember the flash of "there is a rational number between any two distinct real numbers", and how from that followed many of the claims Leray made in his paper. The struggle to reach that insight as much as anything from the claim of certain properties of countable sequences of functions was formative for me.)

It's possible a version of such a policy might be useful to MathOverflow, but I am not seeing that yet.

Gerhard "Is Feeling Somewhat Expansive Today" Paseman, 2011.10.28

Gerry, indeed it should also work that way on MathOverflow. My concern is that someone will take an interpretation of the policy similar to what I outlined above, especially if one only sees the (moral equivalent of the) phrase "it is encouraged to ask and answer your own questions". I have no problems with a policy about updating your questions when you figure out the answer, nor (subject to an advisor's proper guidance) with graduate students asking questions pertaining to their particular dissertation topic on MathOverflow.

Gerhard "Ask Me About System Design" Paseman, 2011.10.28

I would just like to add to this rather old thread, in the hopes that Anton or Scott is reading it that to my mind the "Interesting" tab on SE, if it works (I don't use the site enough to know) would be worth a great deal from my point of view. As the number of users has grown the site has become less and less usable for me, as there are simply many topics that for one reason or another are not interesting to me, or even close. The "Interesting" tab tracks a user's history of interacting with the website and suggests questions tailored to them.

@Asaf: no it's different. The interesting tags on MO allows you to indicate a discrete set of tags that you find interesting. The "Interesting" tab on stack Exchange factors in that user input, but also systematically tracks the user's activities, in terms of what questions they ask, what questions they answer, and what questions they view, and uses all this information to create a model for the user's interests. So it requires less organizational effort on behalf of the user, and it does a better job (hopefully) of tailoring the site to users' needs.

It seems that the SE2.0 has not some features that current MO has. For example, the graphic representation of reputation growth/decline and multi-line comments. The latter one is quite important for me, because some of my comments may be extensive.

In fact, multiline comments are explicitly discouraged by the SE2.0 software: hitting the carriage return would trigger comment submission; and newline characters are ignored in the presentation of the comments.

@David: ... and Anixx claims 2.0 lacks this feature.

@gerald - whoops! I didn't read that correctly. Usually the complaints are 'MO doesn't have this feature of SE2.0...'