84

Is it possible to implement trackbacks from SE sites to the arXiv.org?

arXiv.org -- open access to over 800 000 e-prints

arXiv.org is nowadays the leading archive for scientific preprints. It is used by a majority of scientists in physics, mathematics, and other sciences. See e.g. this Stat.SE post. Scientists typically upload their preprints to the arXiv.org before submitting to journals in order to get their works exposed, and to receive fast feed-back.

Several SE science sites, such as, e.g., Math.SE, Phys.SE, and a handful of other SE sister sites, would benefit enormously (not least in terms of PR value) from trackbacks to the arXiv.

In this way SE would become visible to many scientists, experts, academics and scholars, with knowledge to share, who potentially could become new high-value users for the SE science sites.

Mathoverflow.SE already has trackbacks implemented (mainly thanks to Scott Morrison). Can this be widen to other SE sites as well, in particular SE science sites?


Other related Meta.SE posts: 1; 2.

  • 20
    I'm going to look into this, I can't see how it would be a bad idea, and it's being a good netizen for going the extra mile to indicate that a paper was cited. The PR value is nice, but doing the 'right thing' is a fine motivation alone. – Tim Post Oct 4 '14 at 15:49
  • 2
    +1, have wanted this for a long while for Physics. – Manishearth Oct 4 '14 at 17:01
  • 1
    great idea! only (small?) sticky point is that some se sites do not really like all the arxiv preprints because there is very low screening and there are many so-called "crank" papers.... – vzn Oct 5 '14 at 1:02
  • 17
    1. The issue of crank paper is hardly relevant for this meta question. A SE post that cites a crank paper as it is a reliable source, has a problem by itself, independently of a trackback or not. 2. Although not peer-reviewed, the arXiv has installed various anti-crackpot-countermeasures. In practice, the arXiv is a formidable source of knowledge. – Qmechanic Oct 5 '14 at 1:30
  • @vzn I'm not disputing that this is site-dependent, but in many branches of physics, at least, arXiv has a reputation approaching that of a low-level published journal. Actual crank papers are quite rare. – David Z Oct 5 '14 at 18:24
  • 1
    this can get a bit complicated, suggest some further discussion in chat, did not mean to imply it is my personal sticking pt above. my personal feeling is that arxiv is a great/ topnotch resource to include in questions as a ref or the topic of the question itself (one of the best on the web). what "anti crackpot measures" do you refer to? do not really like that idea much. fyi this is an example of one sites attitude on the subj & some can of worms: is it ok to ask about correctness of preprints on crank-friendly topics. – vzn Oct 5 '14 at 18:29
  • 3
    This request is not about what is on-topic or off-topic on SE, or trying to change that. E.g. on Phys.SE we close a question if it proposes a non-mainstream/crank theory. The request is about the subset of SE posts that the SE community has deemed worthy and on-topic, and that contains an arXiv preprint number. The request is that trackbacks on such SE posts should be sent to the arXiv. – Qmechanic Nov 19 '14 at 19:33
  • Another meta question about arXiv.org. – Qmechanic Dec 12 '14 at 13:04
  • 2
    @TimPost an issue can be that papers might get mentioned or discussed in quite negative ways. (This is not a hypothetical.) – quid Apr 25 '15 at 23:57
  • 5
    @quid certainly this can happen, but I don't see how that constitutes a reason not to implement trackbacks. Unless the people arXiv only want to post trackbacks to posts which provide positive commentary on papers, but I highly doubt that is the case. – David Z Apr 26 '15 at 13:06
  • 1
    @DavidZ thank you for the clarification. Your second sentence touches on a critical issue. If I had to decide on a per-site basis I likely would not accept the track-backs from this site. – quid Apr 26 '15 at 14:03
  • 4
    @TimPost Did this make it onto a development to-do list? Would some form of poll of the technical sites help push the balance towards putting this in place? – E.P. Oct 20 '15 at 22:41
  • 5
    @TimPost What's the status of your looking into it? I think this is a very good proposal and would have benefits for both SE users and SE as a company. On top of that is the obvious benefit to the rest of the internet. – Mast Feb 11 '16 at 21:21
  • 4
    @TimPost Can you comment on the status of this request? – E.P. Jul 4 '16 at 21:53
  • 1
    If this is status-six-to-eight-months, wasn't the sixteen-month timeline a good time for this? Reminder: It's 6 to 8 weeks, not months. – user8397947 Jul 6 '16 at 1:17
14
+50

This really should be implemented: it is good for the sites, it is good netizenship, people want this, and an implementation already exists. As far as I can tell - from any publicly available information - all that's needed is (in essence) for SE to flip a switch.

Just to sum up some responses, questions, and their resolution

  • I'm going to look into this, I can't see how it would be a bad idea, and it's being a good netizen for going the extra mile to indicate that a paper was cited. The PR value is nice, but doing the 'right thing' is a fine motivation alone.

    -- Tim Post, Director of Stack Overflow Communities

  • only (small?) sticky point is that some se sites do not really like all the arxiv preprints because there is very low screening and there are many so-called "crank" papers. (comment)

    This isn't really a problem. The arXiv does contain a nonzero fraction of 'crank' papers but generally speaking this is very low. The arXiv is not peer-reviewed but it is not anyone-can-post, and there are various measures to prevent 'crackpot' papers from getting in. Overall, the arXiv is a formidable source of knowledge and it is the standard go-to place for a large number of disciplines.

    In addition to that, if a post refers to a 'bad' arXiv eprint in a positive light, then the post likely has bigger problems. If a post treats a bad eprint like it's canon, then it should really be the target for some righteous community-moderation fire. On the other hand, if a post refers to a 'bad' eprint to criticize it...

  • an issue can be that papers might get mentioned or discussed in quite negative ways. (comment)

    If the paper is bad, and the post that discusses it says that it's bad, then that's precisely what the trackback system is meant to provide: a way to link from the arXiv page to the wider peer review going on in other places of the internet. If a paper has serious flaws, then by publishing its authors are assuming the risk that people will call out the paper's flaws in other public venues. That's how it goes, and this sort of stuff needs to be defended for the good of science.

    On the other hand, if a post is wrongly tearing on a good paper, then again it needs to be the target of some righteous community-moderation fire. A trackback that sends you to a rant about how the paper is wrong... and the rant is seriously downvoted and in pale gray? Not a particularly big problem. And, in any case, people can already send a trackback from a ranty blog with negative opinions of a good paper - there's always a 'reader beware' stamp on all trackbacks to begin with.

Some other points and concerns:

  • Stack Exchange communities like the arXiv. Like, a lot. Like, wildly-upvoting-the-arXiv-community-ads a lot. See, for instance, the 2016 community ads for Physics, Academia, Mathematics and Computer Science. See those nice, fat scores? See how they're at or near the very top of the list? Try some other years, too, and see how those went.

  • We link to the arXiv by droves. At the current SEDE count, there's currently

    on both posts and comments. That's a lot of trackbacks that are not there - a bunch of authors with fewer ways to know their papers are being discussed.

  • An implementation already exists, and it is currently live at MathOverflow - for some details, see here. This might need a little bit of up-scaling to deal with a bigger volume of posts, but again, it's doing fine at MO.

  • One potential concern is what happens with questions and answers that send a trackback and are then deleted. This is in general a concern with the trackback mechanism - it is always susceptible to link rot to begin with - but at the expense of some extra coding SE can do a lot to prevent this, either by contacting arXiv administrators with a way to let them know that a trackback target has been deleted, or by holding off on sending the trackback if the post gets downvoted or otherwise gets signals that it might get deleted.

  • Relevant sites where trackbacks are a good idea:

This feature request has been around for three years and the implementation already exists, and yet the only Stack Exchange response has been a vague "yeah, that's definitely a good idea" and no motion afterwards. So, if this is not going to get implemented, would the team care to comment on why it's actually very very very hard, or actually a very very very bad idea?

13
+100

The bounty asks (taking things a bit out of order)

Can the SE team comment on the status of the roll-out to other SE sites?

Alright, here goes.

If this is status-declined, can we have an official statement and some thoughts as to the reason?

First off, the good (or less bad) news: this isn't declined. As Tim said a while back, we do like this idea.

If this is status-six-to-eight-months, wasn't the sixteen-month timeline a good time for this?

The problem here really has nothing to do with the specific idea at all. It's the usual story: too many good ideas, not enough time to implement them all. And even if there weren't other feature requests and bugs to compete with, larger projects have been taking up a lot of the company's collective time for a while now, and probably will continue to do so for a while longer.

There are already working implementations of this.

This helps, from a proof of concept perspective, but—with all due respect to Scott's work—it's not just something we can copy-paste. This is a request for official SE implementation of a feature, which means a fair amount of our dev time needs to go into evaluation, design, construction, review, the whole deal. Not to speak too much for the dev team, but I have to think they would want to build hooks into the actual system, for example, rather than using an independent scraping script.

So, for the moment, the best I can offer you is continued limbo. We're not declining it because it's not a bad idea. It's not planned because there's a lot going on (as always) so we don't know when we can find the bandwidth. It's just... a good idea that the time isn't quite right for.

  • 6
    Thanks for the response, though of course I'm not happy that this seems to be status-indefinitely-deferred. Maybe we can convince you that this is something we really want by another show of votes, though - at least to put it on status-planned. I'll award the bounty (at the end of its period), not because I agree but because this does answer the questions it poses. – E.P. Jul 11 '16 at 16:45
  • 3
    You can use the last paragraph as a generic answer to about 90% of the feature requests... ;-) – Shadow The Princess Wizard Jul 11 '16 at 18:35
  • You wink, but it is an accurate explanation for a large number of requests (maybe not quite 90%). "An overflow of good ideas and not enough resources to implement them all" is a good problem to have, but unfortunately, a good problem to have is still a problem... by the way, we're hiring. – Pops Jul 11 '16 at 21:18
4

Reviving this post, as it is a highly desirable feature that could obviously attract new visitors and active community members for several SE communities, as is mentioned in one of the answers to this post.

To the aforementioned list, I would add Computational Science SE.

  • We have at least 255 arXiv mentions according to the query (as of June 4, 2019), which is a relatively large number for a relatively small 8000-question community.
  • The users already requested this feature back in 2012 on our Meta.
  • The usage of arXiv links would be encouraged by this feature, and will steadily grow.

Such feature as arXiv trackback could be a big driver for our community development.

1

I agree with this idea wholeheartedly.

I want to add two thoughts.

  1. I think only answers should trackback.

    Comments are not there to stay, and questions are unlikely to contain something a visitor of arXiv would be interested to read about, namely a citable use of the work.

    Answers, on the other hand, are more likely to use the reference in a productive way, hence are more in the spirit of citations.

  2. What if a reference to arXiv is edited away, or the whole post is deleted? Can we take back a trackback? Does arXiv automatically check for stale links?

  • 3
    I don't agree with tour opinion re trackbacks in question. As of today, 1626 questions link to arXiv in Physics.SE. Most of them request for clarification about the preprints, like this question. Trackbacks to these question would gather the attention of authors themselves and other experts in the field who can answer the questions. In this way, SE could attract high-quality users – Bosoneando Jul 12 '16 at 9:42
  • 2
    @Bosoneando Do we have information about what kind of trackbacks arXiv wants? We don't want to get blocked because of "spam". – Raphael Jul 12 '16 at 10:33

You must log in to answer this question.

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