I recently wanted to read one conference paper and one journal article. The conference paper is named Division by invariant integers using multiplication. By clicking on the "All 9 versions" link on Google Scholar, I found the full paper on gmplib.org.

The journal article is named improved division by invariant integers. It is also available at several sources as PDF.

Why do publishers allow such redistribution of papers / preprints? I mean, if everyone is doing the same as I'm doing, the main revenue source of publishers will go away. Is it because publishers obtain their main revenue from university libraries? If I can't find a PDF preprint of an article, the next thing I'm going to do is to try to find the article through my university library.


Although this is speculation since I don't work for a publisher, there are a couple of reasons I can think of:

  • Due to pressure from funders, who want the research they funded to be publicly available.
  • To take away the focus from gold open access. Publishing preprints is known as green open access, and means that authors are allowed to post their work elsewhere. Often, however, this is with restrictions: only the non-peer reviewed version is allowed, and only after a certain period. This discourages academic libraries from cancelling their subscriptions, and appeases the funders who might otherwise call for the published version to be publicly available immediately. (Note: I think this was mainly a motivation before the publishers figured out they could heavily charge funders for the latter requirements.)
  • Because it might be hard to argue in a legal case for publishers to restrict access to research that they haven't been involved with at all (i.e. not yet been peer reviewed, layout done by the researcher themself, etc.). Yes, they often coerce authors in transferring copyrights to the publishers, but they might rather not test the strength of that argument in court.
  • As a courtesy, to maintain goodwill. It's clear now that it doesn't really threaten their business model (no subscriptions appear to have been cancelled in response to preprints being available), so they can easily do this.
| improve this answer | |

If a publisher doesn't allow distributors of preprints, they'll never get a submission from me. Why would I needlessly restrict the distribution of my own work? The publisher doesn't pay me anything in exchange for these rights.

| improve this answer | |

Because it will happen anyway, so they might as well appear magnanimous by permitting it.

Authors want their work to be read. The publishers effectively have three options:

  • ignore it, which makes them look as though they're not in control
  • sue their authors, which would probably be futile and would certainly reduce submissions to their journals
  • say that preprints are allowed, and focus their efforts on trying to convince people that it's worth paying for the "version of record" anyway.
| improve this answer | |

That'll depend on the field. In computer science, where your example comes from, there is some competition between publishers for conference proceedings and also, to an extent, competition between conferences for quality submissions and high-profile researchers on the committee.

Some conferences are moving or considering to move to open access proceedings, and this makes commercial publishers try and offer better deals, which includes better attitude towards authors. PLDI in your example is ACM's own conference and won't just go to another publisher, but still, harassing authors with copyright restrictions is going to repel authors and committee members. I know, universities and grants in Europe often require to upload publications to open access archives, but I don't know whether this would prevent authors from publishing in a venue that does not allow to publish preprints, since this problem does not normally arise.

I imagine that this indeed hurts publishers. For example, at some point in the past I did have access to Springer's LNCS in my institution, and these days there's some story unfolding in France with universities not renewing their Springer subscriptions.

I'm also curious to know where this will lead. Perhaps, publishers will soon start tightening their copyright agreements (thus confronting the existing open access policies). Or perhaps, they will raise publication costs. Participation in a CS conference is already expensive, usually somewhere between 500-1000 Euro just for being able to present (and then there are travel costs), so it may not make a big difference if a larger portion of this money goes to a publisher.

| improve this answer | |

It's a sign of goodwill, mostly.

Publishers, librarians and academics all rely on one another, and none of them can function without the other two. It's true that by allowing preprints, publishers are risking subscription revenue. University libraries are the main source of subscriptions, and they are highly incentivized not to subscribe if everything is available via Google Scholar! However prohibiting the author from uploading preprints will also generate hostility among academics, who might refuse to submit to the publisher (or worse - see The Cost of Knowledge boycott). It's a balancing act which most publishers err on the side of not prohibiting preprints.

Here's something related: can a university library completely discontinue journal subscriptions and ask its academics to use resources like Google Scholar or Sci-Hub? Yes and no. The library can do it and probably save some money instantly, but in the long run, the publishers will shut down and the journals too, and there'd be no one to handle peer review. Like it or not, all three of publishers, librarians and academics are stuck with one another.

| improve this answer | |
  • 5
    Commercial publishers are not crucial to peer review: they aren't paying for it to begin with. Several high-quality OA journals are doing just fine in this aspect. I suspect the real reason why libraries are reluctant to cut down on subscriptions is that this money is earmarked and they won't get to use it for anything else (but correct me if I'm wrong). – darij grinberg Apr 3 '18 at 3:22
  • 3
    I guess I was ambiguous somewhere I don't see. I am implying that a number of high-quality OA journals (Algebraic Combinatorics, Electronic Journal of Combinatorics, to name two) conduct peer review no worse than Elsevier's, without charging authors or readers. – darij grinberg Apr 3 '18 at 4:18
  • 3
    These journals have no OA APCs to begin with. The only relevant item on the budget for a modern OA journal seems to be the editors, and this isn't that large a cost. Peer review has never been part of the cashflow, so it cannot be a concern. – darij grinberg Apr 3 '18 at 4:30
  • 3
    Well, if you're in the combinatorics research community, you can see for yourself. I don't have a personal experience with the backends, but I know of the EMSs of Electronic Journal of Combinatorics (OA, founded in 1994, not as selective as some but respected and high quality) and Algebraic Combinatorics (former Springer journal that recently declared independence, top tier). The former is bare-bones but functional in all cases I've encountered. The latter is modern and ... – darij grinberg Apr 3 '18 at 5:33
  • 3
    ... feels better to me as a referee than Elsevier's (which seems to be from the 90s judged by the web design). Either way, an EMS is mostly a one-time cost, and can be shared across journals (this is what the Centre Mersenne is for). Sure, not 100% of authors know LaTeX well enough to publish without editorial assistance, but a journal can afford dismissing the 5% that don't. The fact that Elsevier and Springer are (probably) sinking a lot of money in these processes doesn't mean that they cannot work on a leaner budget. Your mileage may vary depending on what science you're in. – darij grinberg Apr 3 '18 at 5:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.