In the 20th century, when I had a draft paper, I might make a few copies and share them with close colleagues, but the main action would be send it anonymously to a journal – in paper.
In the modern world, there are more things one can easily do with a draft paper to get comments from colleagues – one can send it by e-mail to dozens of colleagues, or upload it to a repository like Zenodo.org, or upload it to a commercial science networks like Academia.edu or ResearchGate.
But one still needs official publications, of course – these are not going away, even though they are not strictly speaking needed for dissemination anymore. Without official publication, a scholarly work is unlikely to be cited widely, and unlikely to provide the kind of recognition that we need for our careers. So we still need to submit our papers to journals (or publish them in edited volumes – something that continues to be widespread practice in linguistics).
Now three questions arise:
– Is it OK to submit a paper to a journal at the same time as uploading it to some other site?
– Is it OK to use commercial science networks, which exploit our data the way that other “gated communities” do, without being accountable to science?
– Is it even sufficient to have everything on Academia, so that I don’t even need my own website?
I’ve been wondering about the best answer to all these for quite a while, and here I’d like to share my preliminary conclusions. As circumstances change (and they tend to change quickly in the modern high-tech world), or as I hear new arguments, I might change my mind, of course.
First, is journal submission compatible with simultaneous uploading to some other (semi-)public site?
I think it should be – you don’t need to hide your paper until it has been accepted.
It is true that traditionally, peer review is supposed to be double-blind, and authors should be anonymous. Some journals say explicitly in their submission guidelines that authors should anonymize their articles, but in the modern world, true anonymity is almost impossible to achieve for most papers. Conference and colloquia programs are typically available online, so by searching the internet for the title of a paper, we will often be able to get a fairly good idea of who the author might be.
In my experience, reviewer anonymity is far more important, especially since many reviewers are junior and non-tenured, and by writing critical reviews of papers (co-)authored by influential colleagues, they could be afraid to jeopardize their careers. Moreover, even more important than journal paper reviews are grant reviews, and these are never anonymous. So if we accept non-anonymity for grants, it cannot be so terrible for papers. Moreover, uploading a paper at the same time as submitting it for review is something that is already being practiced widely anyway, and I have not seen anyone object to it strongly. Thus, I see no reason not to endorse this practice.
(Well, maybe there is one reason: Reviewers may be somewhat less motivated to read a paper carefully when they feel that the paper is already being discussed widely elsewhere. But if peer review becomes less informative for the authors while feedback comes from elsewhere, this may be an acceptable tradeoff.)
Second, are commercial science networks acceptable?
Some people say that they are evil, because they are parasitic on the publicly funded research infrastructure.
But commercial companies that serve scientists are active in all kinds of domains: From cleaning companies and equipment maintenance to logistics companies that ship our samples and airlines that transport our fieldworkers – we need service-providers that can do things that we cannot do ourselves.
Now it is true that I have argued that publication should be seen as part of science, not as something that can be outsourced, because the publication brands are used to measure our success, so they are the currency of our business. Publication brands should be scholar-owned, but the commercial science networks are adding something to this: They are telling us how many people are reading our work, who is citing us, and so on. These are the kinds of things we really care about, so if they provide a useful service to us, I don’t see why we shouldn’t pay them for it, either with our data, or even by signing up to a premium account.
As long as no publicly funded organization provides these services, I would not want to do without them. Scientific libraries often emphasize how important they are, but in fact, many of us by now find Academia.edu much more important than libraries.
Third, is it enough to have a ResearchGate account?
No.
Commercial science networks (of which there are two well-known ones, Academia.edu and ResearchGate) do not provide truly open access – one needs to become a member to access the papers. Moreover, and maybe even more importantly, there is no guarantee that the materials uploaded will be there permanently. If they run out of funding, they might simply close down without even sending you a warning.
For permanent accessibility of your papers, it’s best to upload them to a repository like Zenodo.org. This is safer than keeping them on your own website, because when your contract terminates, your university may switch off your website right away. Or if you pay your own provider, there may be changes in the terms, and users may lose access to all the materials. Big public repositories that provide uploaded materials with a DOI seem to be the best place for your work, whether unpublished or published (of course, if it’s published, it can be uploaded only if it’s permitted).
There’s been a similar discussion going on in Association for Computational Linguistics, but with ArXiv rather than Academia.edu (see a blog post here: https://chairs-blog.acl2017.org/2017/02/19/arxiv-and-the-future-of-double-blind-conference-reviewing/).
My personal stance is that double-blind reviewing is important to protect since it helps mitigate implicit bias against underrepresented minorities. There’s some good discussion & sources by Hanna Wallach here: http://dirichlet.net/pdf/wallach13benefits.pdf
Thanks for these links. I think that the issue of bias needs to be taken seriously, but I don’t see any mechanism by which double-blind peer review could be enforced. If I had said in my blog post that peer review must be double-blind, some conscientious readers might feel under moral pressure not to upload their work to a repository and thereby lose the possibility of early exposure. Others with fewer scruples would do it anyway, and would be at an advantage. There would thus be a bias in favour of those with fewer moral scruples, which is not a good situation either. So I think it’s best to evaluate abstracts and papers blindly, but not to go to great lengths to ensure author anonymity.