Why upload to academic preprint sites like arXiv?

Question

The recent question about the legality of uploading to arXiv made me realize that I don't have any understanding of why one would upload to a preprint site. (I don't work in a field where they seem to be common.)

Why would one be concerned about getting a manuscript on arXiv, if it's already under review with a journal?

Is it because of the long publishing timeline of journals? (That could explain why computer science — which publishes much more in conferences — doesn't seem to do much with them.)

Answer 1

In mathematics, there are several reasons to post to the arXiv:

1. It provides free access to papers that might otherwise be hidden behind paywalls. Of course you could achieve this by posting on your web page, but your web page may move or disappear, while the arXiv is far more stable. This means it's better for archival purposes and it's better suited for citations. [Note that this is not just about preprints: it continues to be relevant long after publication.]

2. It attracts readers. Many people pay close attention to arXiv postings in their area, in a way that doesn't happen as much with journal tables of contents. I don't have statistics, but my impression is that a substantial fraction of the people who learn about my papers do so through the arXiv. If you can get noticeably more attention for your research with almost no additional effort, why wouldn't you?

3. It establishes priority. Submitting to a journal does not: if you submit to a journal without circulating your work publicly, then you may still end up sharing credit with anyone who makes the same discovery before your paper is made public. The way to establish priority is to distribute your work so widely that any competitors cannot credibly claim to have been unaware of it. Submitting to the arXiv is the easiest way to show that you have done so: it's widely read, and it preserves all versions of the article with time stamps.

4. It's conventional. This varies somewhat between subfields, but once arXiv use reaches critical mass in a given area, it becomes the standard way of announcing to the world that you have completed a paper. At that point, not posting to the arXiv looks strange.

Compared with other fields, mathematics and physics are anomalous in making heavy use of the arXiv, with physics even more so than mathematics. As I understand it, in high energy particle physics the arXiv replaced an elaborate system of paper preprint distribution that was used in the late 80's (and it rapidly became popular since it was obviously better than the paper infrastructure). Usage spread to other areas of physics, from physics to adjacent areas of mathematics, and then further into mathematics.

Things are trickier in other fields, because many of the advantages depend on network effects. If nobody in your field pays attention to a server, it's not conventional, and it counts for little or nothing regarding priority, then there's less reason to post to it (although it still has some value). I'm not surprised that it catches on only slowly and spreads primarily to adjacent fields, but I expect green open access servers like the arXiv will become more popular over time.

Answer 2

As I see it, the main reasons to use arXiv and similar preprint servers are:

• To disseminate your paper without waiting for the peer review and publishing process. This is a serious issue - in mathematics, for example, peer review often takes a year or more, and it can be several more years before your paper gets to the front of the queue to actually appear in print.

• To make the paper permanently available to readers who don't have a subscription to the journal where the paper is published.

• In some cases, to disseminate and solicit comments on a paper that may not quite be ready for publication, while at the same time establishing priority.

Incidentally, from the computer scientists here, I've gathered the impression that CS uses arXiv very heavily, to the point that it practically supersedes journals as a way to distribute papers. Perhaps I'm mistaken, or it varies by subfield?

Answer 3

Let me add one point not mentioned by other answers so far.

In fast moving fields (e.g., most of computer science), a great benefit of publishing technical reports in recognised/citable series (at least an ISSN) and/or arXiv is to get a citable reference before publication in a more serious venue, such as a conference, or a journal.

So the workflow is as follows:

1. write and polish a paper
2. before a submission to a serious venue, publish it as a TR/arXiv preprint (unless this clashes with a double-blind, or copyright policies of the target venue)
3. submit to the target venue
4. right the next day you can write a next paper referring the work you published as a preprint. It can be either an incremental work, or something else using the result you achieved recently, etc. All this you can do way earlier than the paper really appears in proceedings, or a journal issue.
5. if the work was accepted at the target venue, good for you
6. if it wasn't, you still can refer to the work, while improving it and submitting next time/elsewhere/etc.

As you see, the turn-around time is what this is all about. Of course one should be later careful regarding replacing references to TRs with references to the really published stuff in CR versions - if there's time/space for it.

Answer 4

Some researchers (supposedly from respectful schools) will use preprints as a way to hack the double-blind process and to try to impress possibly hesitating reviewers with their name or the name of their institutions. Preprints usually appear in Google Scholar if you work in that field. So, by the time the paper gets to the reviewer he or she already knows who wrote it. In my opinion that violates the double-blind process, but the official position is that it does not, at least in conferences where I submit. Is anyone aware of different policies?

Answer 5

If papers were posted on the arXiv before submission then I argue it would be easier to branch out. I also think it would increase the quality of submissions and the speed of progress. Overall, it should benefit your community and so, albeit indirectly, yourself.

I quote from this blog post an illustration.

The current system reinforces the partition of research into (sub)areas, making it hard for an outsider to leave their own. Of course, it is good to have a domain of expertise and produce deep results in it. Still, I think it would be better if it was a little easier to work in different areas.

To illustrate the difficulty, suppose you want to start working in new, hot area X. To learn the background, typically you have to read papers. However, for every paper that you read, it is not uncommon that there is another one which is or was under submission. Indeed, the community is producing great results the majority of which is rejected due to capacity constraints. So unless these works are on electronic archives such as the arXiv, you don’t have access to them.

Who does? The experts of area X, to whom these papers are sent so that they can be properly evaluated. But it may be hard for reviewers to ignore submissions until publication. Suppose for example you have been working on problem Y for months and now you are asked to review a paper that solves Y. Are you going to ignore this information and keep working on Y despite knowing that you will be beaten? Also, when the paper does come out you’ve had a long time to internalize its implications.

The edge currently given to an insider over an outsider is months if the paper is accepted right away; it may be years otherwise.