# Selected Papers Network

As some (many?) of us are aware, there is a recent venture to enable some sort of "meta-discussion" interface for the articles in the form of the Selected Papers Network. For a quick introduction to what it is, one can see its help page. But basically, the website "collects links to discussions about specific papers" under one roof, potentially making it easier to find discussions and Q+A about a given paper.

Currently, the SPNetwork only indexes the Google+ discussions. Given that every now and then there are questions posted to MO asking specific questions about specific papers, I feel that maybe it will be useful for these kind of information to also be gathered by the SPNetwork.

Questions:

1. Should a feature request be made for the SPNetwork to also index MathOverflow?

2. If so, how should we implement things on our end?

• @FrançoisG.Dorais: two comments (a) can you post that as an answer and (b) the SPNetwork indexes more than just arXiv preprints. It is also capable of catching anything that has a DOI. The arXiv trackback feature presumably doesn't work for that :-). – Willie Wong Jul 4 '13 at 13:39
• Perhaps there should be an opt-in mechanism; I have some worries as the discussion on MO involving preprints and papers are quite varied in quality. It might be neither for this site nor for MO a good idea to have each and every question automatically showing up there (to some extent this also applies to arXiv trackbacks). – user9072 Jul 4 '13 at 14:20
• @quid: How would you make that compatible with the user contribution license? – François G. Dorais Jul 4 '13 at 16:49
• @FrançoisG.Dorais: I am sorry I do not understand your question, as I do not think my remark has anything to do with licenses, or to the extent it does I suggest that something more restrictive is done relative to the content here than what you suggest, so I cannot see why compatibilty with a license should be a problem (except if it is also for what you suggest). But anyway my remark was not a "per user" suggestion but a "per post" suggestion. I will clarify what I mean in an answer. – user9072 Jul 4 '13 at 19:13
• @quid: As far as I can tell, collecting citation data from posts is permitted by the user contribution license, are you suggesting it should be amended? – François G. Dorais Jul 5 '13 at 15:11
• @FrançoisG.Dorais: I don't think quid is arguing about whether it is permitted, but whether collecting all instances of mentions to arXiv postings would generate too much noise to make the function really useful. – Willie Wong Jul 5 '13 at 15:36
• Isn't that a problem for the spnetwork, Willie? Why should we have to deal with this? – François G. Dorais Jul 5 '13 at 15:42
• @FrançoisG.Dorais "Why should we..." We should not have to do anything. But implicit in my asking this meta question is whether we would want to do something to help. Naturally what we could do to help depends a lot on the spnetwork, but whether we as a community would help certainly does concern us. – Willie Wong Jul 5 '13 at 15:49
• I think @quid successfully derailed this discussion. Willie, the answer to whether we want to help is yes and we're already helping by providing the spnetwork convenient access to arXiv trackback data. – François G. Dorais Jul 5 '13 at 15:54
• @FrançoisG.Dorais: fair enough. Those do provide answers to my questions as they are phrased. And perhaps further discussion (either here or there) should wait until we see how the trackback data are used as is. – Willie Wong Jul 5 '13 at 15:55
• @FrançoisG.Dorais: you missed my point all the time. Now perhaps this is entriely my fault but then others understood it just fine, so this seems not so clear. – user9072 Jul 5 '13 at 16:12
• The main project site seems to be down. – Dmitri Zaitsev Nov 19 '17 at 21:07

We have been talking to Chris Lee and John Baez. There is a tentative plan to use the existing arXiv trackback mechanism for this purpose (once it is fixed).

We are also currently considering ways to improve the way MathOverflow deals with citations. Anyone with concrete ideas about this is welcome to contact the moderators.

• Do you intend to accompany this by a change of policy regarding discussion of papers and preprints on MO. At the moment it is rather discouraged. To on the one hand discourage it, and on the other hand help to give it visibility when it happens seems inconsistent. (Since I think some think I am against most every type of question, let me stress that I was rather not the one typically against reasonable discussions of preprints and papers; if the need should arise I can link old "tea" discussions to support this.) – user9072 Jul 5 '13 at 13:14
• @quid: I don't know how you find all those implications. All we do is send data to the arXiv & spnetwork. There are no plans to change anything about how MO works. – François G. Dorais Jul 5 '13 at 15:02
• I did not say you want to change anything about how MO works. By contrast I rather want to suggest that if there is the intent to send things to spnetwork, there likely should also be a change in policy. Otherwise the resulting situation strikes me as close to absurd. Ever so often it was said MO is not for discussing papers, asking about their correctness and so on, and now you want to enable it that it is also a frontend to a page whose purpose is just this. This is in my opinion inconsistent. – user9072 Jul 5 '13 at 15:14
• @quid: I still don't understand the issue here. Why can't things stay the same on MO? What's wrong with the way things work now? – François G. Dorais Jul 5 '13 at 15:18
• Let me try again. It is my understanding that discussion of preprints and questions on specific papers (such as their correctness) are rather discouraged here on MO. (This is my perception of how things are.) And some held the opinion that it could be good to have a seperate site for just this. Now this site exists. This is all fine. But now comes the problem. You want to enable people to basically use an MO post instead of a dedicated Google+ post to more or less achieve the same goal. This might lead to an increase of posts of this form on MO, which are considered as OT by some. – user9072 Jul 5 '13 at 15:26
• @quid: There are no plans to use MO in this way. All we do is send the citation data. – François G. Dorais Jul 5 '13 at 15:29

This is not really an answer to the question but more an eleaboration on my comment.

As far as I understand the way arXiv-trackbacks for MO are supposed to work is (almost) completely automatic, if there is a link to the arXiv it creates a trackback (except if it falls through the cracks for some reason).

To do the same for this network is in my opinion not a good idea, by contrast there should be an opt-in mechanism, by which I mean that the post only shows up there if in some way the post is "marked" in a special way (this could be some non-invasive text-pattern or special way to link, or still something else technically this should be not a problem).

The reason why I think it is not a good idea to send everything over there that happens to link to the arXiv is that this would flood that page with tangentially relevant entries, and in some cases also with problematic ones. There is simply a big difference between making a post that is on some paper, and a post happening to link to a paper. (This remains true if this is restricted to links in questions.)

This also applies to the arXiv trackback in a certain way. But first there are many many papers on the arXiv while there are not yet that many posts on this site, so it is in a relative sense not so much problem, besides these trachbacks on the arXiv being quite a side-aspect while on this site they are the main thing.

So, if we send them many low-quality entries (at least for the purpose at hand) this is seems like a problem.

(Final explication what I mean with "problematic": sometimes certain (not all) discussions of preprints here on MO are critizised as potentially offensive or embarassing for the author, still a track-back is created, as it is automatic, so that everybody looking at the arXiv might stumble over this potentially offensive discussion. This is a slightly strange state of affairs in my opinion. But then these trackbacks are not so visible, but for that other site this is a bit different. In addition, it could also make sense to inform people that what they say here on MO on a paper, will in some immediate way be linked to the paper. It is not clear to me if everybody is aware of this. Of course the license permits it being done, but not everything that is permitted needs to be done, or at least that it is done might be communicated more prominently. To be clear I have not personal problem with any of this, this is only something that crossed my mind on ocassion and also just now, and to explain what I meant.)

• SE has a mechanism for selecting which questions should be tweeted. It is probably takes some combination of votes and views to decide which question to tweet and which question not to tweet. See for example see cstheory twitter. Something similar might be used for restricting which questions to spnt and which questions not to spnt (in combination to a having the #spnt hash-tag inside the question). – Kaveh Jul 4 '13 at 20:20
• I agree that there are some issues to think about, but ideally the discussion should be informed by information about the frequency of different cases. Your answer seems to focus on the case where someone asks a question about a preprint. I suspect it is much more common for trackbacks to occur where someone says "you can find an answer in the following preprint". – Neil Strickland Jul 5 '13 at 14:32
• @NeilStrickland: this is likely true, and also why I consider the problem of tackbacks to arXiv as less of a problem, since a trackback of this form seems at least somewhat in the spirit of a trackback to the arXiv. However, for what is discussed now, AFAIK understand this new site, I fail to see the relevance of having, say, an MO answer that mentions the paper as proving something showing up there in many cases. I link to the arXiv in not few of my answers, but I would hardly ever consider them as relevant commentary on the paper, IMO they would be 'noise' on that other site. – user9072 Jul 5 '13 at 14:42
• Wouldn't it be better bring your concerns up on the spnetwork instead of here? We don't know what they plan to do with the data we will provide them. – François G. Dorais Jul 5 '13 at 15:09
• @FrançoisG.Dorais: I have no stake in spnetwork and no personal concern related to it, so I do not see why I should care to raise any concern there. I do have some stake in MO and am worried that if not done carefully the posts provided on spnetwork via MO will make look MO bad (either in a direct way, or just because by harvesting too much many things are sent there that do not have much anything to do with the intended purpose there). – user9072 Jul 5 '13 at 15:33
• Because of the user content license, we have very little control how others use MO content. If you have an issue, you have to bring it up with whoever uses MO content, there is no other way to do it. – François G. Dorais Jul 5 '13 at 15:37
• @FrançoisG.Dorais In addition the question (a discussion) asked if "we" should ask spn to do something. My answer would be, rather no. And if at all, than only for explcitly "marked" posts, which was my original comment. Since IMO this is better both for MO and spn. Perhaps Kaveh's comment communicates in a clearer way what I consider the better idea than just sending 'everything' there. And after all this site also does not scan Google+ either for papers AFAIK but only takes posts explictly tagged for showing up there. – user9072 Jul 5 '13 at 15:40
• @FrançoisG.Dorais: if I may jump in here. How much information will the arXiv trackbacks actually provide? My impression is that it essentially just says that "url ... has mentioned this arXiv posting". This is not how spnetwork currently scans Google+. I think what quid is saying is that maybe we should also use something like the spnetwork hash tags on Google+ to give them a suggestion of what we would prefer them to index and what not. – Willie Wong Jul 5 '13 at 15:41
• @FrançoisG.Dorais: regarding your reply which I had not seen. Yes I know. But the question was if there should be some activity on MOs site. And also if MO should help. To which I suggested a way that seems better to me than just sending everything. Now you can disagree with it being a better idea to introduce some "mark" on MO for sending things there, but to say it is not relevant here but I should suggest this there is a bit surprising. – user9072 Jul 5 '13 at 15:42
• @WillieWong: yes, exactly. Also I upvoted Kaveh's first comment here. – user9072 Jul 5 '13 at 15:44
• Still, quid and @Willie, wouldn't it be better to bring this up with the spnetwork and discuss it there? github.com/cjlee112/spnet/issues – François G. Dorais Jul 5 '13 at 15:48
• Re using vote counts: that's a terrible idea. – François G. Dorais Jul 5 '13 at 15:49
• @FrançoisG.Dorais: regarding you last comment, I heartily agree. Regarding your previous: if the community just says "meh, I am not going to bother with another set of social conventions", then there's little point in bringing it up in over there. (Question 1 is intended to gauge interest in this idea...) – Willie Wong Jul 5 '13 at 15:52
• @quid: Regarding using some mark to send things on the spnetwork, there is already one though, for the alpha phase, it's only scraped on G+; in later phases, it will be scraped from blogs, twitter, MO, etc. – François G. Dorais Jul 5 '13 at 16:24
• @FrançoisG.Dorais: thank you for the information. re usgae, I am not fully convinced this will work well without further quidelines, but perhaps this is a premature discussion and it is best to wait how things develop. – user9072 Jul 5 '13 at 17:54