Not signed in (Sign In)

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


    I'm porting so-if-internetsd00d81-answers-a-question-related-to-my-research-how-do-i-add-him over here.

    Adding as a coauthor is a higher step than I would think that answering a question on mathoverflow should warrant. I'm curious as to what question was answered that could lead to such a generous offer!

    Much more likely is simply acknowledging someone's help. Here I think that the best strategy is not to acknowledge the help of a particular person, but to acknowledge mathoverflow as a whole. So, for example, say:

    Here we need to solve the identity Ax = b. The following is distilled from the answers on mathoverflow to \cite{question on mathoverflow}.

    To do this, of course, someone should come up with a template for the bibliographic information. Looking at the basic BibTeX fields, I'd go for:

    • Title: Question title
    • Author: Mathoverflow
    • Eprint: Direct link to question in Mathoverflow
    • Date: Date asked (?)
    Yeah, almost surely questions and answers on mathoverflow shouldn't be sufficient for coauthorship. Certainly askers/answers should go away and write papers together, and if they do they should definitely give us a free plug by writing in the introduction that it's a "mathoverflow baby".

    I'm looking forward to the first mathoverflow citation! I'll buy the author a beer, for sure.
    If the answers came out of a large discussion, then I think acknowledging MO as a whole might make sense (though I might still mention explicitly the main contributors, if possible), but if the answer came from a specific person, I would acknowledge that person (and perhaps say that the answer was given on MO).
    • CommentAuthorfedja
    • CommentTimeDec 28th 2009
    I guess it should be handled separately in each particular situation but what, in my opinion, should be done in every case when the discussion value was non-zero is to add thanks to MO and its developers in the aknowledgement section of the paper (you know, the one that contains thanks for useful discussions and such).
    This to me also shows why people should write on MO under their own name and not some alias, it makes acknowledging much easier. But I suppose that is a different discussion.
    Well, yeah, but that's already encouraged by the staff and the community, so I don't see how we can encourage it any more than it already is.
    • CommentAuthortheojf
    • CommentTimeJan 3rd 2010
    I've been treating MO as an extension of 1015 Evans (the math lounge at UC Berkeley). If I include a fact I learned from someone else in a paper, I'd certainly acknowledge them in the acknowledgments section. What can be hard in both cases is that I don't always know if their answer to my question is in print _anywhere_, as certainly I didn't find it when writing said paper.
    It is already weird enough to read some MO answers when people have edited or removed responses. Citing an MO post seems almost as tricky as citing a blog post or Wikipedia article. If there were snapshots of MO posts and responses which could automatically get LaTeX'd and compiled into a single file every couple of months, maybe they could be archived somewhere with a more standardized e-citation system. Maybe even arXiv'd?

    This may be a temporary solution, but it's also possible to include the question/answer into the paper (perhaps as an appendix) with full attribution (certainly fine in the people involved agree).

    I think this paper by user JSE and colaborators does a reasonable job citing Mathoverflow (see footnote 7). Though I think the link to the specific question might have been an improvement. At any rate I think something small like that is much more appropriate than including appendices or compiling questions to the arXiv, etc.

    I was wondering whether something like the following would be possible: to have a link which gives the BibTeX data for an MO question/answer.


    @figueroa, this is a great idea. It would require a bit of custom javascript to insert the link.

    Want to propose a sample BIBTEX entry? I or someone else can probably generate a little script that lives at say that produces whatever is required.


    Scott, I thought that Andrew Stacey's suggestion at the top of this thread was a good one, where Date could be the date of the latest edit.


    I've been meaning to learn enough jquery to efficiently implement this for a while. In principle, it shouldn't be too hard if we use the javascript functions already provided by SE. But it's a bit tricky because you have to do a fair amount of walking around the page to figure out the post number, author, question title, etc.

    Regarding links, what is the MO policy on permalinks? Will there be link rot if MO switches to another platform?

    Also what about the usual issues with links to versioned content? Namely, when the link is resolved it may not point to the intended version but instead to a much-revised later version.

    By the way, this is not relevant to this thread, but... how do I go about displaying my full name instead of just "figueroa"? I would like this to be the same as my MO id "José Figueroa-O'Farrill", but for some reason I didn't understand the distinction between real name and username in Meta. Also I believe I created another account "José Figueroa" in Meta which I didn't use. Any possibility of merging both accounts (or simply getting rid of one)? Thanks!


    Fresh off the presses, I arXived earlier today this paper with Paul de Medeiros. It's a hep-th paper, but some results in the appendix are thanks in no small part to the answers I received to some of my recent questions concerning finite groups. We acknowledge MO in a couple of places, and in particular Robin Chapman, Keith Conrad and Theo Johnson-Freyd for their help. There are no \cite's to MO in this version of the eprint, partially because I would like us to establish a BibTeX format for MO questions/answers. I'm happy to add such \cite's to a second version.


    @Jose, click the "Account" tab at the top, then "Personal Information" on the left. You should then be able to change your username.

    Let me know if you also want to merge the other account. It would take approximately 5-10 minutes of my time.


    Okay, I hacked out a decent-looking solution on faketestsite. There should now be a "cite" link on every post which produces a popup box with a BibTeX entry. Please go try it out and report back here if you find any bugs or have any suggestions.

    I feel like the BibTeX entry should somehow include the user number since user names are not unique. Maybe this is unnecessary since you can get to the user page once you know the post ID. If we decide not to include the user number in the citation, I should remove that requirement from the attribution page.

    Question: What is the appropriate entry type (I've used "misc")? What are the appropriate fields to fill out for that entry type?

    That looks quite nice Anton. I think the main point of concern people might have before using that technique of citation is the stability of the EPRINT link. Is it a goal to keep those links stable?
    • CommentAuthorWillieWong
    • CommentTimeJul 28th 2010

    I don't think there are any standard ones that are really applicable.

    If you are willing to go non-standard, you can use "electronic". Only the bibtexkey is required, the optional fields are author, month, year, title, language, howpublished, organization, note, and url. (So in that sense it is not really that different from misc.)


    Another question: should we include some citation information besides BibTeX data? Should the popup also include an sample formatted citation? I'd imagine yes. Something like

    Soo Key Foo (, The question title,


    I think one of the standard entry types should definitely be used (misc being probably the best default lacking something obviously more appropriate). It should be made easy to drop this into a .bib file and know it will compile more or less reasonably, preferably without annoying warnings.

    • CommentAuthorWillieWong
    • CommentTimeJul 28th 2010

    If we insist on standard types, then we should also insist on standard fields where available.

    The Date data field is non-standard. The Month and Year fields are. But in regards to the crowd-editing nature of MO, we should also add a Note field that states "accessed on YYYY-MM-DD".

    Also, as the URL and eprint fields are both non-standard, it'd be nice to have both present. (Also being a bit selfish here, since JabRef supports url, but not eprint.)

    A fixed Howpublished field may be nice, but we should come up with a nice wording for it. Maybe something like

    HOWPUBLISHED = { appeared on the website MathOverflow }

    Though a bigger problem is how the information will be used by BibTex. In other words, if the included information doesn't make it into the .bbl file, it is all a bit moot.

    • CommentAuthordanseetea
    • CommentTimeJul 28th 2010

    What would happen with CW posts?


    @Anton: You should also record which revision of the post was used and link to the revision in the history page rather than linking to the actual original post. That way, it will be easier to pinpoint exactly what the person is citing.

    • CommentAuthorWillieWong
    • CommentTimeJul 28th 2010

    Ah, and for the amsalpha style (the one I usually use), the URL and Eprint fields are ignored for the Misc entry type. It will print

    Author, Title, Howpublished, Month Year, Note

    So perhaps the url should be also included in the Note, something like

    HOWPUBLISHED = { MathOverflow ( },
    NOTE = { URL (accessed on YYYY-MM-DD): }

    which is similar to how I deal with LivingReviews articles.

    @Harry Indeed, I mentioned that way above but it seems to have been lost in the noise.To reiterate, link rot, version drift, etc are all well-known issues with permalinks that deserve to be addressed if we promote such linking.
    • CommentAuthorWillieWong
    • CommentTimeJul 28th 2010

    @Harry: I think if we include an "accessed on" date (and maybe a time?) it should be easy to find which version of a post the author read. On the other hand, since the revision histories are public, it will also be nice to have "(revision X, accessed on YYYY-MM-DD)" to further simplify matters.

    But I disagree that the link should be to the precise revision. Since the posts are expected to improve over time, I think we should follow LivingReview's model of linking to the most recent version, with the caveat that the author read a particular, possibly outdated version.


    The problem is that revision information is not available unless you visit a different URL, so I don't have sufficient javascript-fu to include that information. I'll include HOWPUBLISHED and NOTE as WillieWong suggested (Edit: done). Hopefully the date will be sufficient revision information. I can include an exact time too if that would really help. I've removed DATE since it seems to be completely nonstandard and added URL (exactly the same as EPRINT). It seems a bit over the top to include the URL so many times.

    @danseetea: Right now the last user-link to appear is the author. This normally means the original author rather than the person who last edited the post. In the case of CW, I think it will actually be the person who last edited the post. But I think the question of who to list as the author of a CW post is quite a bit trickier than the technical question of how to extract the right name from the page. In an ideal world, what should the author field say?


    I've also added an example citation to get a feel for how it looks to have that next to the BibTeX.

    • CommentAuthordanseetea
    • CommentTimeJul 28th 2010

    @Anton: I think handling CW is indeed tricky - and regardless of the sophistication of the script that will be built, human intervention will sometimes be needed to extract the information of who the real author is (or in certain cases, multiple authors). The best solution IMHO would be that also in CW posts the original author should be cited by default (if that's possible), and that people citing CW posts always be careful to look at the revision history to see if any changes need to be made to what the bot generated.

    • CommentAuthorMariano
    • CommentTimeJul 28th 2010

    As an amsrefs user, I wonder how much work it woud tak to support it too! :)

    • CommentAuthorWillieWong
    • CommentTimeJul 28th 2010

    @Anton: ah, you are using the last modified date as the access date? I was originally thinking putting the current date as the access date, and the last modified Month and Year into their respective fields. But this way looks fine too (maybe modulo a tweaking of the phrasing). You did a great job there. Thanks!


    I'd propose dropping both the URL and EPRINT field, if we're including the information we really want in the NOTE field.

    Some editing by hand is going to be essential for most people anyway, and if they have a system that makes BIBTEX usefully display the URL or EPRINT fields, more power to them, but they can do the copy and paste easily enough. For everyone else, who just relies on the NOTE field, these are just cruft.

    Also --- great job!

    • CommentAuthorWillieWong
    • CommentTimeJul 28th 2010 edited

    @Scott: First, I don't object to dropping the URL and EPRINT fields. The reason that we provide the url in the BibTeX entry should be that it gives proper credit--gets displayed in the bibliography. So for that the non-standard, non-displayed URL/EPRINT fields are just cruft.

    But I want to note that for people who use reference management systems (JabRef, Zotero, Mandeley [I think I spelled this last one wrong]...) the URL/EPRINT fields has nice semantic uses: it allows the software to associate an entry to a resource on the web. Of course, I agree with the sentiment that copy and pasting is easy enough. I just want to point out that the fields can be useful beyond just BibTex's handling of them.


    Good point, Willie, I guess I've just been frustrated in my attempts to use such systems: they all expect me to change my established BIBTEX database too much, to fit their expectations.


    Thanks, Anton! It looks pretty good. I'm happy to do some editing by hand later, so it's great to get the basic information already in this form. Cheers!


    Scott: thanks. Now I remember the problem: my full name "José Figueroa-O'Farrill" is too long for meta :( so I'd like to use "José Figueroa" which is the name of my old account which I never used. So, if it's not too much of an inconvenience, could you merge both accounts? Thanks!


    I took the liberty of simply renaming the defunct account to José Figueroa(2), and renaming your main account to José Figueroa, which took 30 seconds instead of 5 minutes.


    Thanks a lot, Scott! I'm glad it did not waste too much of your time.


    Any objections to running the current solution on MO?

    @Mariano: would the corresponding amsrefs entry be obtained by replacing "@MISC {MOnnn," by "\bib{MOnnn}{misc}{"?

    • CommentAuthorMariano
    • CommentTimeJul 28th 2010 edited


    @MISC {MO76,
        TITLE = {Animated Proofs},
        AUTHOR = {Jon Awbrey (},
        HOWPUBLISHED = {MathOverflow},
        NOTE = {URL (accessed 2010-03-14):},
        EPRINT = {},
        URL = {},

    would become

      title={Animated proofs},
      author={Jon Awbrey (},
      note={{URL (accessed 2010-03-14): \url{}},

    I would add


    (Notice the trailing comma in the third argument to \bib)

    (By the way, the pointer in the popup is coming out wrong for some reason)


    +1 for amsrefs.


    It shouldn't be too hard to present BibTeX/amsrefs information depending on a choice of radio button or something. I'll deal with that tomorrow or this weekend.

    (By the way, the pointer in the popup is coming out wrong for some reason)

    I did change the cursor from "pointer" to "auto" because "pointer" communicated (to me) that something would happen if you click anywhere in the popup box.

    • CommentAuthorMariano
    • CommentTimeJul 29th 2010 edited

    Ahh. I see.

    I don't know how 'auto' selects the pointer, but since the main purpose of that popup is to have some of its text selected and copied, it's a bit weird that it ends up using a pointer used all over the place for something else.

    You could wrap the text (ie, all contents of the popup except the Close link) in a div, and set "cursor:text" on it.


    There should now be cite links on MO proper (you may have to clear your cache to see them). I've added a radio button which allows you to choose whether you get BibTeX or amsrefs. There were some surprisingly annoying situations that come up, so I'm sure there are some bugs that I haven't worked out. If you'd like to help out by finding bugs, here are some things to look for:

    • Weird usernames/titles caused trouble. Specifically, usernames with a ' in them, like A'nton. This should be resolved now, but there might be other similar bugs.
    • If the author has been deleted (rare, but occasionally people request that their account be deleted), then there was trouble. It should now correctly fish out the name of the author, but it won't handle the user page link gracefully. There just shouldn't be any user page link, but right now it uses the uid "xx".
    • Community wiki posts and posts with "multiple authors" are a likely source of bugs.
    • There were some problems with deleted posts and answer that you have accepted. The reason is that I find the postId by looking at the nearest flag link since this is usually the only link I can be sure will be there. But you cannot flag deleted posts or answers that you've accepted (for some reason ... this has been changed on SO), which caused some trouble. This should be resolved now.

    Please post bug reports and further suggestions on this thread.

    @Mariano: does the cursor look right for you now that it's a textarea?

    • CommentAuthorMariano
    • CommentTimeAug 1st 2010

    @Anton, perfect.

    Minor point: currently we have

    note={URL (last changed 2010-08-01):}


    note={URL: (last changed 2010-08-01)}

    end up looking slightly better grammatically?

    Also: the 'last changed' date is slightly unuseful, in that it stops being correct exactly when the post is changed, while an 'accessed' date conveys exactly the same information (what is the version being cited) and never stops being correct.


    @Mariano: I agree the parenthetical should come after the URL ... I've changed it. But I like "last changed" better than "accessed" for some reason. Somehow it just seems weird to me that two people who looked at exactly the same thing should end up with different citation information. I don't feel too strongly about this, but I'd like to be more convinced before changing it.


    I agree with Anton that "last changed" seems better here. I know "accessed" is more-or-less standard language when citing web pages, but I think that came to be the case only because the "last changed" time is not typically available. When the "last changed" time is available (and is considered reliable), it's definitely the more relevant piece of information.