14

Some journals advise authors to send their manuscript to arXiv before submission.

What is(are) the reason(s) behind this suggestion?

2
  • 15
    Can you give an example of a journal advising this?
    – TimRias
    Jan 11 at 11:04
  • 27
    I don't know why a journal would recommend this, but as a fellow mathematician I encourage this in the strongest possible terms. It is the best way to disseminate your work widely, and to draw attention to your achievements.
    – academic
    Jan 11 at 12:24

7 Answers 7

14

A rumor I heard suggests boosting a journal's impact factor by:

  • Delaying Publication: Publishing papers as late as possible, exploiting extended peer review and printing queues.
  • Early arXiv Deposits: Encouraging authors to share papers on arXiv well before formal publication.

Assuming a paper is submitted to arXiv in 2024 and published in 2028, the four-year gap accrues additional citations that are initially counted towards the published paper, elevating the journal's impact factor from 2029-2030.

This is a cynical notion, and reputable journals likely do not deliberately follow this strategy. Nonetheless, it underscores the importance for an editorial board and author to maximize a paper's visibility, possibly through early publications on platforms like arXiv. (The publisher might have different opinions of the arXiv.)

9
  • 12
    From the best of my knowledge, the only place that merges preprint citations with the actual journal publication entry is google scholar. I don't think this "boost" can really happen considering that IF is estimated from more standard databases such as web of science, where preprints are not considered.
    – The Doctor
    Jan 11 at 14:04
  • 6
    @TheDoctor I believe that is correct. Here's what Clarivate wrote when they added the Preprint Citation Index to Web of Science last year: "[Preprints] are clearly marked in search results and not included in any citation metrics including Times Cited Counts or the Journal Impact Factor" (first word edited for clarity).
    – Anyon
    Jan 11 at 14:54
  • 8
    Could you expand on how delaying increases the impact factor. I don't get it. The total number of citations will remain the same (or go down by lesser early visibility). So it must be about which year the citations are counted towards or something? Is it that "early" citations count more? (Where here early can even be before publication.)
    – Kvothe
    Jan 11 at 18:56
  • 8
    @Kvothe Impact factor is based on citations in the first two years post-publication. If the content has already had its profile boosted and made a splash as a pre-print, one would expect citations in those first two years to be higher than if it had first appeared “cold” in the journal only, due to the lag in when citations themselves appear in the published record. ie the paper will hit the ground running, rather than have to build momentum. Jan 11 at 21:46
  • 1
    @MichaelMacAskill Only up to a point. For most papers (other than the few that become major landmarks), most of the citations they generate are in a relatively short period after their appearance. So if the time lag between preprint and official publication is too long – relative to the norm for the area – then most of the citations to that paper will have already appeared before the paper is published and hence will not count for impact factor. Jan 11 at 22:44
23

A non-cynical answer is that mathematicians seem to value collaborative work very highly and to be willing to share that work so that it can be extended. The same is true to some extent in CS and maybe other fields. This is more than just establishing priority. We ("mathy" types) are happy when our work is accepted and extended. It implies that we make some contribution, hopefully significant, to the march of human progress.

Note that math also tends to use alphabetical listing of authors, again emphasizing that the work is truly collaborative.

Combine that with the long time it takes to verify some math results, leading to long time to formal publication, makes use of preprints natural in such fields. It is an evolving recognition, I think, of the value of collaboration and the value of enabling future work on a research thread, even if it is by others.

In some sense, then, this is just a recognition of the pressure and value of making results known early and generating analysis of those results.

Moreover, if an author submits to a preprint service and gets immediate negative feedback, it might prevent submission to a formal process, slowing down the process for other, more valuable, papers. It is an easy weed-out system for good journals, speeding up everything else.

3
  • How does a preprint get "negative feedback"? It seems to me that silence is the most natural response to a bad paper... Jan 12 at 18:47
  • 2
    Mathematicians talk to each other @DavidRaveh
    – Buffy
    Jan 12 at 19:08
  • 2
    @DavidRaveh It's not so much about "bad papers" (as the journals would want to reject these anyway), rather about finding the odd mistake or weakness and in this way improving already good papers., Jan 12 at 22:33
13

It's a good idea (for you) to put your paper on arXiv when you submit, for various reasons that include establishing priority and raising awareness of your results. Indeed, arXiv is the main way I (and probably many other mathematicians) become aware of new papers.

As to why a journal might advise it, I don't think I've seen this happen very often. The most obvious case of journals that do are what is called "arXiv overlay journals", where the final published paper will be hosted on arXiv but linked to from the journal and with their formatting added. This is a low-cost way of achieving "diamond open access", that is, completely free for authors and readers. A fairly prestigious example is Discrete Analysis.

In other cases, I suspect that it is not so much "advice" as the journal wanting to make it clear that they permit putting the preprint on arXiv.

13

I know from an editor that has followed the submissions to their journal for a certain period and correlated to when (and if) the corresponding manuscripts appeared on the arxiv, that there was a significant correlation between manuscript being posted on arxiv before submission, and whether they would eventually be accepted. This suggests that the manuscripts that appeared on arxiv were of higher quality. One possible explanation was that posting on arxiv indicates a certain confidence of the authors in the manuscript, since it will be publicly visible there for perpetuity. This seemingly filters out the category of authors that are willing to continually fling garbage at journals, in the hope that it will eventually stick somewhere.

This certainly correlates with my personal (anecdotal) experience in refereeing papers. Papers (in my field) that are not already on arxiv, appear to be almost exclusively garbage.

Journals may be advising this to help discourage the garbage submissions (that only cost them money).

Alternatively, I've pretty much stopped accepting referee requests for papers not already on the arxiv both because the general low quality, and because it avoids any future ethical issues regarding using results from papers that I referee. (I can simply cite the arxiv version, if I ever need to.) If others are doing the same, this may have stimulated journals to start recommending first posting to arxiv, since it will make it easier for them to find a referee.

2

I see three reasons why a journal might want to encourage this. The first two already got mentioned.

  1. It is a good idea.

  2. It serves as an implicit screening. The main forum to advertise algebraic geometry papers is the math.AG section of the arXiv. If you aren't willing to advertise your paper there, how could it be significant enough for [our prestigious journal]? This works in practice, as mentioned in other answers.

  3. There is also an ethical issue that is solved by this requirement. In principle, anyone could submit two similar papers to different journals (say, use the same methods to solve the case k = 19 and the case k = 20, but split the result up into two papers), and editors and referees would not be aware of the overlap (unless coincidentally it goes to the same referees). If the paper the journal receives for the case k = 19 is on the arXiv, the journal has some protection against this. (And yes, I have encountered seen versions of this in practice.)

1

With the growing concern around paper mills, use of AI and the growing numbers of retractions, journals are perhaps seeking to leverage the community review implicit to the preprint/preproof process. I can't see journals intentionally delaying publications for a long period- authors would stop submitting to them and this would damage their operational KPIs. Seems more likely any delays are most likely damaging to journals not helpful.

0

I believe that the citation number is not that important when putting the paper in arXiv as per say. For me, i see it as a way of protecting the idea(s) and the technical contribution. I had submitted a paper to a magazine paper where we talked about a certain theory to be leveraged within topic X. It got rejected and after 4 months the same core ideas where found in another magazine which was published. I meditated a lot and realized that arxiv is a way of protecting your work.

Nowadays, review process is not that ethical as editors and reviewers look at names of authors and whether they are IEEE fellows to give respect to the paper being reviewed.

You must log in to answer this question.

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