Preprints: revolution and resistance
2 August 2018
How long does it take you to publish your article in a high-profile journal? Assuming a few rejections on the way (probably due to that mean Reviewer 3), easily a year or two pass before you can finally see your work published in a journal. There is no doubt that the peer-review process significantly improves your manuscript and increases its scientific merit, however your main message usually remains the same. Thus, why not share your findings with the community once your work is complete and your manuscript is up to scientific standards? If your answer is yes, then preprint servers are for you. Publishing your work as a preprint comes with numerous advantages which I, as a person who submits his work to these servers and regularly reads others’ work from them, will try to summarise in this article.
What is a preprint?
A preprint is a complete manuscript that authors submit to non-profit public servers. These servers do not peer-review the manuscript, yet screen for basic quality-control (e.g. if a manuscript is scientific in nature but not, let’s say, metaphysical). The preprint should include your findings as well as methodology so that the community can immediately make use of your data and protocol. Preprint servers allow readers to comment on the manuscripts, creating a community-based peer-reviewing system. To submit your paper to a peer-reviewed journal, there are several options: (1) preprint servers have an agreement with most journals so that you can transfer your preprint to the journal directly via the server; (2) some journals give you the option during the submission process to upload your manuscript to a preprint server simultaneously; (3) or you can do two submissions separately.
Let us come back to the question of how long it takes to publish an article, and this time let’s look at some numbers. It takes 173 days (median) from article submission to acceptance for Nature and this is similar for Cell. It is 210 days for Nature Biotechnology and 245 days for Nature Cell Biology. Even after acceptance, it takes up to 10 weeks to see your paper online in Cell. It usually takes a year or two to produce data for a “high impact” paper and if you are in the third year of a PhD or postdoc position trying to publish your data so that you can apply for postdoc or PI positions, 300 days from submission to publication is quite a long time. In particular, if your results matter for human health, this is a long time before you can share your data with the community and potentially start influencing healthcare decisions. Again, preprints are not there to circumvent the peer-review system, which is absolutely essential for scientific integrity. However, the community can see your work and already give their opinions a year before your article is published. Yet, some fields are still fairly resistant to the concept. Bioinformatics and neuroscience are fields that have the highest number of preprints (see Fig. 1 for BioRxiv preprints as of June 2018), while immunology, cancer biology and clinical trials are well below average. This shows that for more translational/medical articles, the number of preprints is significantly less.
What are the concerns about publishing a preprint?
The main concern one might have for publishing a preprint is the possibility of being scooped. However, preprints so far have not caused a notable problem in scooping, according to ASAPBio, a non-profit organisation promoting open and transparent science. Of course, it may be too soon to conclude this about biology preprint servers since they are fairly new (BioRxiv, the most commonly used preprint server for biology, has existed since 2013; PeerJ preprints since 2013; preprints.org since 2015; F1000Research since 2013). However, ArXiv, the preprint server for physics which is host to more than a million preprints, has been functioning since 1991; and its existence has contributed immensely to the physical sciences. Individuals in the community can see, review, comment on, and respect others’ work. This community-based review process makes the subsequent peer-review process of the same article extremely easy. In addition to that, contrary to scooping, preprints can actually bring groups working on similar topics together, and this cooperation potentially enhances both groups’ chances to publish, preventing unintentional scooping.
The second concern for publishing a preprint is that the data has not been peer-reviewed, and therefore it cannot be guaranteed that the findings are completely sound; this could ultimately be misleading for the field. This is a fair concern. However, peer-review by 2-3 referees cannot eliminate all the flaws in a scientific study. This is one reason why reproducibility is the biggest crisis in experimental sciences. On the contrary, preprints can help eliminate experimental flaws; the wider community is the best reviewer because others can subsequently perform the experiments described in a preprint and quickly pinpoint the flaws if there are any. This cannot be done with peer-review since referees do not perform the experiments to see if results are reproducible. Furthermore, the response to the preprint by the community (e.g. in the form of comments on the website of the preprint) may even be used to improve the paper before it is subsequently submitted to a journal. Ultimately it is the community that makes science self-correcting and progressive.
One last concern is that a journal’s policy may not allow an article to be posted as a preprint before submission. Although there was resistance when preprint servers first started - mostly by the for-profit journals - nowadays almost all journals (all non-profit, community journals, Elsevier, Springer-Nature etc) accept preprints with a few exceptions (such as the Journal of American Medical Association).
What are the advantages of publishing preprints?
I personally publish all my papers as preprints before submission to a peer-reviewed journal. So far, it has been extremely positive for me. As a consequence of publishing preprints I have been contacted by the editors of journals, persuading me to publish in their journals (examples include Biophysical Journal and Journal of Immunology, whose editors usually screen preprints for potential papers). Some journals even have “preprint editors” dedicated to this job such as PLOS Genetics or Proceedings of the Royal Society B.
Of course, this does not guarantee their subsequent acceptance, but shows that the editors are interested in your work. Also, the metrics of preprints (see Fig. 3), such as how many times your paper has been downloaded, the number of views it has had, and readers’ comments, all of which are tracked by the preprint servers, are good indicators of how the community perceives your paper, which can be used to convince the editors that your work is interesting. Furthermore, I get feedback from scientists working in the same field. Since their sole purpose is to help me make this paper better, the comments are usually extremely constructive and useful. I also list preprints in my CV as most funding bodies now treat preprints as publications for grant applications.
Another advantage of sharing your work as a preprint as soon as possible is that it changes how fast it gets cited when it is subsequently published in a journal. One of my papers got cited a few times in the first month it got published because the community had already been made aware of it via preprint form. Since preprint servers also state which journal your paper is eventually published in, it is easy to find and cite the peer-reviewed published work. Furthermore, since a preprint has a digital object identifier (DOI), this can also be cited if preferred.
There are ongoing arguments on the flaws of the scientific publishing system: the extreme dependence on impact factor; dominance of for-profit journals and their paywalls; long waiting times from submission to publication; numerous iterations of reviews with endless new experiments required each round; and of course politics of the editorial/peer review process… I am sure preprint servers will not be able to magically fix all of these problems, but they at least partially overcome some of these issues by making your science open as soon as possible and democratising science by letting the community judge your work. Together with the other efforts of community-based journals (e.g. EMBO journals, The Company of Biologists, Elife etc.), such as scooping protection, transparent review process, editorial communication to reduce the unnecessary revisions, and open data sharing, preprint servers will contribute to better, more open and more reproducible scientific publishing. I strongly advise you to explore preprint servers both as authors and readers. If you have no idea where to start, I recommend signing up for article alerts and you will get daily alerts on the new preprints in your field.
So how do we stand in preprints?
As of June 2018, Oxford posted 1400 articles (of which 920 list an Oxford-affiliated corresponding author) whilst the University of Cambridge posted 1500 articles (of which 810 list a Cambridge-affiliated corresponding author). Sixty papers were posted with a scientist with an MRC WIMM affiliation.
More opinions on preprint servers can be found in this article written by editors of major community journals.
This article was edited by Jessica Davies