Not signed in (Sign In)

Vanilla 1.1.9 is a product of Lussumo. More Information: Documentation, Community Support.

    • CommentAuthorNoah Snyder
    • CommentTimeNov 30th 2009 edited
    We should have a central repository somewhere with links to papers that had a significant mathoverflow component. This will be useful when someone needs to apply for a grant to pay for Math Overflow (or when Ravi renews his grant).

    There's a wave Writing an article about mathoverflow with a little of this, but there's been no traffic recently. I'll add anyone who asks me -- just IM or email.


    Some random items, duplicated from the wave:

    • CommentAuthorNoah Snyder
    • CommentTimeDec 5th 2009 edited
    Passing remark now I have a meta userid: I think that 25 or so years ago some people thought that (internet newsgroup) sci.math would change the world and would be heavily cited in maths papers forever more. I think that 20 or so years ago people thought the same about sci.math.research . I don't think I've ever seen a paper that credits sci.math or sci.math.research as being crucial to its existence.
    "people thought the same"->"some people thought the same"
    Searching Google Scholar, I find 3 results that mention "mathoverflow" and 29 that mention "sci.math.research" ...
    I did not manage to search "sci.math" without also matching "sci. math" with myriads of results since this is contained in many journal names.
    Funny, Gerald: a result *you* showed me on SMR is in my paper

    which I will be revising (and submitting to an appropriate journal) one of these days. I think a lot of folks probably just use "personal communication".

    I think Steve Huntsman may have hit the nail on the head here. How many "personal communication" citations are really "sci.math.research" for example, but not cited as such because of perhaps lack of guidelines on how to do that. I confess that I am new to the whole scene so this is just a hunch...

    I have cited Andrey Gogolev's answer to a MO question as a "personal communication" in something I'm writing now as well. I don't see myself citing MO explicitly, but one of the biggest reasons I go there is to contribute in the expectation that my own actual research problems will get some attention. In fact next week I'll be returning to something I put on hold for about three weeks and if I can't knock it out quickly I'll be asking it there. It's another Markov partition question BTW.

    So how might we convince Steve that he should cite MO more directly?

    Personally, I wouldn't hesitate to cite a particular MO page, using the convention of adding "Accessed DDMMYYYY" when citing (potentially mutable) webpages. What objections do people have to doing this, and can we resolve these objections?

    I'd like to see such references for two reasons. One, I think it's more useful to the reader. They can actually go read the discussion, instead of being potentially frustrated by an opaque reference to a private conversation.Two, it's great advertising for MO!


    Polymath5 is now farming out some subquestions to mathoverflow. I think this is a great success story! See my blog post about mathoverflow and polymath, or just jump directly to the list of questions, tagged polymath5.

    @ Scott: I believe that the people contributing to an answer that is helpful should get the citation. MO might get the reference, e.g.

    Gogolev, A. [MO identifier]

    which I might end up doing anyway. I think the situation now is quite a bit different than it was on SMR even a few years ago, not least because of the peculiarities of Usenet and the ubiquity of the web. But this brings up another issue. What is the long-term picture for MO? If the site goes away in a few years or gets supplanted by a new one (again, think SMR) then citing MO would be problematic.

    Steve: I don't think there's anything problematic about specifying a MO URL as opposed to not doing so. The worst that can happen is that at some point in the future the URL stops working. Meanwhile, it's useful.

    Personally I think that if MO helps you with your research, then acknowledging that in print

    • is polite,
    • is helpful to your readers (who can look it up),
    • helps MO to carry on being funded,
    • attracts new people to MO,
    • costs nothing.

    I probably agree with you that in the reference list, the citation should be for the person, not MO itself - but the source should be given as a particular MO page, e.g. in the way that Scott describes.

    Perhaps if we kept one of those "cleaned database" files lying around and compiled every so often and if the site goes down for good, we could store the database backups somewhere safe, like on servers at a university willing to host them, then that way, at least the information isn't gone forever.

    A quick remark because I didn't know this until I tested it out: changing the title of a question does not break the link with the old title in it.


    So is there a consensus on how to quote MO properly (I feel I may have to do something like that soon) ? Thanks.


    Re: What if MO disappears.
    As soon as possible, we'll provide sanitized dumps of the database which should contain basically all the information that you normally have access to. Given that the SO dumps are only a few hundred MB, the MO dumps should be quite small.

    There is an attribution page (linked to from the bottom of each MO page) listing what you should do if you want to republish any content. If you're just citing or paraphrasing something that happened on MO, I'd include the following information:

    • The name(s)/user number(s) of the person who posted the material you're using (e.g. "Greg Kuperberg (").
    • A link to the question/answer, of the form, where 12345 is the number of the post. As Qiaochu points out, you don't have to include all the stuff after the number; that stuff is just there to make it easier for Google to understand what the page is about. Also note that if post number 12345 is an answer rather than a question, the above URL will redirect to the right place.

    I don't see any downside to linking to the specific post or page that you're using, even if there's a chance it disappears. It doesn't stop people from citing impossible-to-find papers. Btw, does anybody have a copy of Artin's Algebraization of formal moduli I they could send me?

    Let's agree on some citation guidelines, and I'll include them in the attribution page. Is there some way I should make the attribution page more prominent?

    I think something along the lines of

    Gogolev, A. "Proper families for Anosov flows." (2009).

    is entirely appropriate. I will try to remember to do this.
    Anton: Your attribution requirements state that, when republishing an MO answer, I must link each author's name to their profile. What if I republish in a format that does not support hyperlinks, such as a physical hand out or lecture slides? Does the requirement not apply, or should I type out the URL manually?

    @David: I would be happiest if the user number appeared somewhere, even if it's not the full URL, like this

    See this very similar discussion on meta.MO.
    answered Jan 15 at 11:52, David Speyer (/users/297)

    But I think it should be fine to just include the number of the post and the name of the user if it's a pain to include the user numbers, since that is enough information to track down the source. I'll look into the possibility of including the URLs of the posts and the user numbers automatically whenever you send an MO page to your printer (or print to a file). It's also worth noting:

    • I'm not a lawyer, but I think including an answer in a hand out or lecture slides would count as fair use, so you're not legally required to clear all the hurdles the copyright holder sets up. Those requirements are there to keep people from blatantly taking MO content and selling it in some way. Citing MO precisely is primarily a courtesy to your audience (so they can track down the source) and to the author of the material.
    • You can always skip the hurdles if you get permission from the copyright holder (the author of the post).

    I think there are two different concepts here: copyright and citations.

    The first concept is only applicable if you're copying the text; the copyright law (I'm not a lawyer, but this idea is quite clear and well-known) doesn't protect ideas, mechanisms or algorithms (note that patent law sometimes does).

    For example, suppose nobody knew how to calculate some function f(x) at x = 137 and finally Smarty S. was able to provide a long calculation that established that f(137) = 12. It doesn't matter where s/he did it — in a journal, on the web, in a talk: both before and after, anyone is able to talk about the value of f(137) and speculate about it being equal 12, or claim it being 12 or 13. After the result is posted, nothing in the copyright law prohibits saying "f(137) is known to be 12 (see reference)".

    I don't think therefore that there could be any legal requirement for citations that Math Overflow could impose, unless some part of the text is copied. However, ethically, it's well-established in mathematics that you should cite other people's ideas, for two reasons: (1) you shouldn't pretend you did something when it's actually another person and (2) if the statement is non-trivial, you need to give readers the ability to see its proof or the source.

    Since we aren't sure about the sitewide policy, perhaps the best thing would be to ask the person who answered in each case, especially if a person is normally pseudonymous. I personally would go for something like

    [10] Communicated by David Speyer on MathOverflow (/questions/11851).


    Let's agree on some citation guidelines, and I'll include them in the attribution page

    I think it may be a good idea, but it should make clear the distinction between republishing (action possible under cc-wiki license; and I'm not aware of how we could change any of its requirements) and citing (something done voluntary by mathematicians on ethical basis; and the format of which this thread discusses).


    I've added a citation recommendation to the attribution page.


    William Stein thinks we're awesome.

    • CommentAuthoryfarjoun
    • CommentTimeFeb 26th 2010
    would it be possible or desirable to create an "export to bib format" button for questions?
    • CommentAuthorMariano
    • CommentTimeFeb 26th 2010

    and/or amsrefs... to which I am partial :P


    There's an article in "The Tyee", a newspaper from British Columbia, that talks about mathoverflow at length, including quotes from David Brown and Anton Geraschenko.


    The publicity is nice, but the fact that they mention us right before "psychics" is something of a double-edged sword.

    • CommentAuthorMariano
    • CommentTimeMar 2nd 2010

    Well, "Equations and extra-sensory perception" makes for a great tag line...


    I just noticed that the arxiv's experimental full text search pulls up a few articles mentioning mathoverflow that I hadn't known about:

    • CommentAuthorjc
    • CommentTimeJun 26th 2010

    Very cool. It seems worthwhile to add comments to the questions cited in the arxiv articles to complete the trackback circle.

    Interesting list, Scott.

    I am amused that this article:
    ends with the line "We do not have strong evidence for this conjecture to be true, but it is mathoverflow-hard."
    • CommentAuthorjbl
    • CommentTimeSep 11th 2010 edited

    The conjecture to which it refers is posted (by one of the authors) here:

    I've mentioned and cited MO in this physics preprint:
    Remark: I mention MO on p24 of and this paper does not seem to show up in Scott Morrison's search link above, indicating that perhaps the link needs to be refined in some way (either that or I missed the paper in the list). It's in tt font with a .net after it. In fact searching for at arxiv seems to give more hits than Scott's link (which searches for other things like mathoverflow)

    Apologies if these have been mentioned elsewhere, but I just wanted to add 2 links to newspaper articles to the repository:

    1. San Jose Mercury News, Stanford and UC Berkeley create massively collaborative math
    2. The Daily Californian, Student's Site Connects Mathematicians Worldwide

    I've mentioned this in another thread -- but cannot locate it now. I cited and acknowledged MO in the following eprint:

    Just today :

    (Though unpublished and not really of the same spirit as some of the others, still thought it was worth tabulating here).

    Is there a "mathoverflow in the news" thread elsewhere? Anyway, a minor mention that I thought was interesting.


    And an article all about mathoverflow at

    • CommentAuthorSam Nead
    • CommentTimeNov 11th 2010
    I think this has to count as a success:

    @Sam, indeed, that is pretty neat. (For those in a hurry: Michael Freedman posted a paper on the arxiv specifically to answer two mathoverflow questions.)

    • CommentAuthorWill Jagy
    • CommentTimeNov 11th 2010
    I wrote to Ian Agol, who expected this method rather than having someone here typeset it into MO itself. When I saw Michael's post on one of the questions, I edited in the link within his answer, which may have had a depressant effect on readership as they saw my name as most recent modifier. Joseph O'Rourke did it better on the other question, he simply put the link in a comment below, so the main "Active" sort showed Michael as the modifier. Better for the purpose.
    @Will: Thanks! :-)
    • CommentAuthorWill Jagy
    • CommentTimeNov 11th 2010
    Hi, Joseph. My impression is that Freedman is not at all comfortable with the MO software. Now Scott M. has edited his typing, which was pretty bad. So I think you had the right idea, minimal interference, but there is also the issue of leaving a post (as I did as well) with poor spelling. Question of how well you know a person, I guess, or your perceived role on MO.
    • CommentAuthorBen Webster
    • CommentTimeNov 11th 2010 edited

    I also cite a MO question in the most recent version of one of my preprints:

    • CommentAuthorWill Jagy
    • CommentTimeNov 11th 2010
    reference Tsu, page 14
    • CommentAuthormarkvs
    • CommentTimeNov 11th 2010
    A. Dranishnikov and I cite an answer from MathOverflow here: (page 20, reference to Dmitri Panov).