# How will you cite a discussion on this site in your paper?

Should we just provide the link or write down the names of users whose responses were most useful?

This has come up on MO, and they actually have a nifty cite link for questions and answers. A question gets cited like this:

@MISC {MO39428,
TITLE = {x-th moment method},
AUTHOR = {fkenter (mathoverflow.net/users/8894)},
HOWPUBLISHED = {MathOverflow},
NOTE = {URL: https://mathoverflow.net/questions/39428 (version: 2010-09-21)},
EPRINT = {https://mathoverflow.net/questions/39428},

URL = {https://mathoverflow.net/questions/39428},


}

and an answer gets cited like this:

@MISC {MO39450,
TITLE = {x-th moment method},
AUTHOR = {Mark Meckes (mathoverflow.net/users/1044)},
HOWPUBLISHED = {MathOverflow},
NOTE = {URL: https://mathoverflow.net/questions/39450 (version: 2010-09-21)},
EPRINT = {https://mathoverflow.net/questions/39450},

URL = {https://mathoverflow.net/questions/39450},


}

I think that's a reasonable way to do it, so the site gets mentioned. This can be done even if the mention is an acknowledgement: we can say "The authors thank so and So\cite{} for their comments leading to the proof of Lemma 2.1".

In other words, whether you choose to cite or acknowledge, the citation mechanism can be used, and I think that's appropriate. The difference between this and a coffee room discussion is that there's no permanent record of the latter to cite, and modulo linkrot, there's one here.

If you want to acknowledge multiple users, you can use their cstheory user page, or their actual page if it's linked there.

What we also need is a way to keep track of such references if they appear. It's too soon now, but maybe if anyone sees one they can create a new CW question and start listing examples.

• This is great, but for longer term citation, it might be useful to add a note about the version number. Unfortunately, unlike Wikipedia, the revision history does not provide easily accessible links for specific revisions. (The revisions could be linked to, but the interface does not make these links available.) Sep 22 '10 at 23:01
• I think we should add this to our FAQ. Sep 23 '10 at 14:47
• Sure. go ahead :) Sep 23 '10 at 16:55

The original question asks about citing a discussion, not an individual question or answer. I'd be tempted to use a style similar the one I use to cite Wikipedia articles:

@MISC{cstheory-meta-meta,
, author = "{CSTheory Stack Exchange contributors}",
, title = "How will you cite a discussion on this site in your paper?"
, howpublished = "\emph{Theoretical Computer Science Meta -- Stack Exchange}"
, url = "http://meta.cstheory.stackexchange.com/questions/352",
, note = "Updated September 25, 2010"
}


...although in some contexts it might be more appropriate to name the author of the question that sparked the discussion:

author = "Suresh Venkatasumbramanian and others"

• This deserves a separate question, but I really hope this site isn't always called "Theoretical Computer Science – Stack Exchange". Sep 25 '10 at 5:09
• I think we'll get a better name once we exit beta Sep 25 '10 at 5:56
• @Suresh: Isn't our name going to be CSTheory? Or do we get to choose a name that is different from our domain name? That would be interesting indeed. Sep 25 '10 at 14:44
• I don't think the 'TCS -SE' needs to be replaced with CSTheory. But I have no clue what will happen when we exit beta. We'll get a better idea in 4 days when the Web Applications SE2 site exits beta. Sep 25 '10 at 15:34
• @Suresh: The policy of Stack Overflow seems to be branding every Stack Exchange website differently, at least when the domain name is concerned. Guessing from this, I do not think that the name “Theoretical Computer Sceience - Stack Exchange” is acceptable. Sep 25 '10 at 15:54
• indeed. note that our domain name is set: meta.cstheory.stackexchange.com/questions/1/… Sep 25 '10 at 16:16
• @Suresh: What I was pointing out is that for all other SE sites, the name of the site ("superuser") and the domainname, ("superuser.com") are the same. In that vein, I guess our name will have to be "cstheory" since we've chosen our domain name to be "cstheory.org". Sep 25 '10 at 17:03
• Let's see. frankly I hope not. Sep 25 '10 at 17:58
• See the new meta discussion: meta.cstheory.stackexchange.com/questions/389/… . I propose we call it "STOC Exchange". Sep 25 '10 at 22:48

If it was someone else's answer to your question, I think it might be a better idea to add an acknowledgement, not a citation.

You don't add a citation if a coffee-room discussion with your colleague led to a proof idea; it is much more typical to thank your colleague in the acknowledgements (unless you decide to co-author the paper). Discussions on this site should be somewhat similar to that.

In the acknowledgement, you can mention our (temporary) domain name cstheory.stackexchange.com and the full names of the users with useful answers (answers that contributed something concrete towards preparing your manuscript).

• If the question is about acknowledging, I agree to your answer with one exception: I will not probably write the domain name if I know the real name of the person I acknowledge to. This is the same as the fact that I will not write the name of a conference if I get an idea from a coffee-break conversation during the conference. Sep 21 '10 at 15:07
• Sure, but I would very much like to see this site advertised wherever reasonably possible. :) Sep 21 '10 at 15:17
• I understand. I may write the domain name, too. :) Sep 21 '10 at 15:28
• You could always do the following: JF thanks Jukka Suomela and Tsuyoshi Ito for useful discussions on X, which were facilitated by the CS Theory Stack Exchange (cstheory.stackexchange.com). Sep 21 '10 at 16:02
• You could also leave out the HTTP link. I think mentioning "CS Theory Stack Exchange" is sufficient. Or once our name becomes cstheory.org, our name and domain will be the same. Sep 21 '10 at 22:29
• Better yet, include the http link as a hyperlink in the pdf of the paper (via \usepackage{hyperref}). Sep 25 '10 at 4:52

Edit: I would like to clarify that this answer only applies to the case where you really cite a post on this website. As Jukka and Joe say, if you want to mention a general helpful discussion on the website, acknowledgment seems more appropriate.

I do not think that anything is different from citing other webpages. I would do the following:

• I prefer to cite books, journals and proceedings to citing preprints or webpages, considering the long-term availability.
• I would try to make it clear what I am citing. Am I citing a question? Or an answer? Or a collection of answers?
• When I cite a question, I would cite the name of the author, the title, the URL and probably the date.
• When I cite a answer, I would cite the name of the author, the URL and probably the date. Also, I would probably write something like “An answer to ‘[the title of the question]’” in place of a title.
• If you are citing a result proved in an answer, perhaps you do the time honoured tradition of citing it as private correspondence with whoever gave the relevant answer and include the URL. However, personally, I would go for an acknowledgment instead. It seems a more appropriate place to reference informal discussion. Sep 21 '10 at 16:07
• @Joe: I think that whether to cite or acknowledge depends on the context. I gave this answer because the question was “how to cite,” not “whether to cite or acknowledge.” Sep 21 '10 at 17:17
• Oh, that wasn't meant as a criticism of your answer. I just meant for citation (say of a proof) personal correspondence may be the way to go. For general helpful discussion, on the other hand, an acknowledgment seems appropriate. Sep 21 '10 at 17:25
• @Joe: I agree about the difference between when to cite and when to acknowledge. Sep 21 '10 at 17:34

My 2 cents: acknowledge the person in the text and put the reference to the cstheory discussion in a footnote. My wife works in art history, and they proceed this way (however, they reference real life events and private communication more than q&a website ;) ).