Josh from The Winnower (thewinnower.com) here. I wanted to reach out and see if users of mathOverflow would be interested in permanently archiving and assigning DOIs to top threads with The Winnower. We’ve begun to offer DOIs and permanent archival to blogs, scholarly reddit AMAs, and we think mathOverflow is equally deserving of such services, services that are typically only afforded to traditional scholarly publishers. In short we’d love to make these great Q&As citable in the scholarly literature and count on users CVs for credit in the workplace/academia. But of course, we’d like your feedback before we do anything. We’ve met with some great people at the Stack Exchange offices and based upon your feedback they are willing to help. So…

  1. Do you think top threads in mathoverflow should be assigned a DOI and archived permanently via The Winnower?
  2. If so, what threshold would you set
  3. If not, why?

For background here is some more info on why we are offering DOIs to new media and how we’re doing it.

What is a DOI?

Why we assign DOIs and archive scholarly reddit AMAs

And for those curious we archive content via Portico, the same method used for many leading scholarly journals.

Josh, founder of The Winnower (josh@thewinnower.com)

  • 9
    $\begingroup$ FYI: I expect that this will get a poor reception. I think that there is great skepticism within the mathematical community of allowing organizations (especially for-profit organizations) that are not run by academic mathematicians to play a role in scholarly publishing. Of course, many do for historical reasons (e.g. Springer and Elsevier), but few people like this state of affairs (though there are diverse opinions about what to do about it). Every for MathOverflow, there was serious skepticism during the transition to the current StackExchange software (which in effect made us part of $\endgroup$ Sep 15, 2015 at 2:20
  • 3
    $\begingroup$ their "community"; before, we viewed them as just a company whose services we were purchasing, though we never did end up paying them anything). Before we made that transition, we insisted on getting signed agreements that they would give us the data we need to switch to a different platform if we ever decided that we disagreed with their decisions. $\endgroup$ Sep 15, 2015 at 2:23
  • 8
    $\begingroup$ I think well-thought out essays that arose in answer to questions deserve a bit more linkability/permanence that they currently have. For instance, some of Thurston's answers wouldn't be out of place in generalist society journals such as the EMS Newsletter or the AMS Notices. Giving those DOIs would be a reasonable move, apart from the fact I don't know how the legalities work (aside from the Creative Commons license under which contributions here sit). $\endgroup$ Sep 15, 2015 at 2:59
  • 8
    $\begingroup$ (IANAL, but) Note that since answers here are CC-BY-SA, one could legitimately republish them at The Winnower and add a DOI, with appropriate attribution to the author and MO. $\endgroup$ Sep 15, 2015 at 3:01
  • 3
    $\begingroup$ MathOverflow's content is already being archived on archive.org ( archive.org/details/stackexchange , with periodical updates twice a year or so), although not directly in a human-readable format (but not hard to parse either, I believe). While a DOI for MathOverflow altogether sounds like a good idea, I don't see a good reason for DOIs on top threads unless we are doing DOIs on every thread. Most threads on MO should be about equally likely to be cited in research papers. $\endgroup$ Sep 16, 2015 at 20:23
  • 3
    $\begingroup$ @darijgrinberg "Most threads on MO should be about equally likely to be cited in research papers." I agree, and a much better source than "private communication", "manuscript" or "talk in the 1970s that few people alive can remember". $\endgroup$ Sep 17, 2015 at 1:17

2 Answers 2


DOIs are in some sense just like URLs, but with some differences. In particular, a DOI can be transformed into a URL by prepending http://dx.doi.org/.

  1. A DOI is intended to be a permanent name, that will not change meaning over time. On the other hand, a URL may become obsolete (for example, if MathOverflow was forced to move to a different host name, or changes to the software required us to change the structure of URLs).
  2. A DOI is perhaps accepted in some places, and not others. For example, because of the permanence issue, some forms of scholarly communication may object to using a URL to cite a reference, but allow a DOI.

I think it's very reasonable for MathOverflow pages to have DOIs, with two essential caveats:

  1. The URL that the DOI resolves to must be under the mathoverflow.net domain, not at a third party site (e.g. the Winnower).
  2. The URL should resolve to a page that reflects a snapshot in time of the MathOverflow question.

At present there's just no good way to achieve the second requirement, because there's no functionality for snapshots. (There are plans afoot, however.)

In any case, if someone wanted to assign DOIs of the form 10.XXXX/<prefix>/QQQQ/TTTT (where QQQQ is the question number, and TTTT is a timestamp), which for now just resolved to https://mathoverflow.net/questions/QQQQ, and could later be adjusted to resolve to http://archive.mathoverflow.net/questions/QQQQ/TTTT, that would be fabulous.

  • 1
    $\begingroup$ Great points! We are only here to help. If Stack Exchange assigns DOIs themselves, that is great! However, it seems unlikely as they will need to first get an ISSN, then pay 50,000 a year to assign DOIs (crossref.org/02publishers/20pub_fees.html) and 81,960 for archival (portico.org/digital-preservation/join-portico/for-publishers). We are offering this free of service to the community as it fits our ethos and mission to publish "grey literature." We'd only implement a system after feedback from the community. Just as we are doing now. $\endgroup$
    – Josh
    Sep 17, 2015 at 18:25
  • 2
    $\begingroup$ @Josh: Please read the webpages you link to! -- The fee for assigning DOI's would not be 50000 dollars per year, but just something like 275 -- and the fee for archiving would not be 81960 dollars per year, but just something like 250. -- Or do you suppose MathOverflow's annual total publishing revenue would be more than 500000000, respectively, more than 200000000 dollars?? $\endgroup$
    – Stefan Kohl Mod
    Sep 17, 2015 at 19:11
  • $\begingroup$ @StefanKohl Thanks, but I don't think that is correct. The revenues refer to the parent company (Stack Exchange). I'm not sure what SE makes per year in revenue but based on how big their office is and where it is, it is probably at the high end of the spectrum for CrossRef and Portico. $\endgroup$
    – Josh
    Sep 17, 2015 at 23:22
  • $\begingroup$ to follow up. From the website: "*Notes on definition of "Total Publishing Revenue": Total publishing revenue includes all publishing revenue from all the divisions of an organization (primary and secondary) for all types of activities (advertising, books, journals, databases, article charges, etc). For membership organizations, member dues allocated to subscriptions will be included in total publishing revenue." - See more at: crossref.org/02publishers/20pub_fees.html#background" $\endgroup$
    – Josh
    Sep 17, 2015 at 23:26
  • 12
    $\begingroup$ MathOverflow actually has a complicated independent identity (we're a registered non-profit in the US). I suspect @StefanKohl is thinking about our (infinitesimal) revenue. $\endgroup$ Sep 17, 2015 at 23:27

Long-term archiving is a valid and important issue, but I do not think it should be done via thewinnower.com. This is a problem I feel we can and should solve in collaboration with libraries or archive.org, without the need to involve an additional party (especially an unknown one). Our current coverage by archive.org crawls is good enough for most purposes, although not perfect, and getting a library to agree to take in periodic data dumps and archive them via LOCKSS or the equivalent could make it even better.

Whatever system we eventually end up with, I don't think it should be done piecemeal according to "top threads" and I wouldn't want to seem to be claiming that DOIs make content more valuable or citable.

  • 2
    $\begingroup$ Sorry, as new company we are relatively unknown. We are exploring new areas/services in scholarly publishing that major publishers are ignoring. From the looks of it, it appears as if you think such services are valuable. However, not with the help of The Winnower. Based on the feedback received it looks as if we will likely not be the ones to assign DOIs or archive the content. I do sincerely hope that a library does step up to do this as you mention if we are not to be the ones. $\endgroup$
    – Josh
    Sep 14, 2015 at 21:23
  • 3
    $\begingroup$ Sorry if we gave the impression that DOIs are some sort of magical tool--they are not. They do have value though and I thought it would be a nice step in the evolution of scholarly communication if Stack Exchange threads had them. $\endgroup$
    – Josh
    Sep 14, 2015 at 21:23

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .