Since its launch, Nature Computational Science has implemented multiple policies and workflows with the goal of fostering research transparency and openness. For instance, our authors must adhere to specific guidelines regarding the reporting and reproducibility of published results, including the availability of relevant data and code. The journal also carries out code peer review1, mandates data and code citation2, and offers the option of transparent peer review3, in which reviewers’ comments, authors’ rebuttal and editorial decision letters are compiled and published alongside the manuscript if the authors opt in to this service. In addition, we also encourage our authors to deposit and share a preprint of the original submitted version of their manuscript prior to or during peer review, which represents another opportunity for increasing openness throughout the research process.

Preprints are certainly not new. As a matter of fact, they have had a long history in the physical sciences: arXiv, the popular open-access repository of preprints, launched back in 1991 primarily as a physics archive and then expanded to other areas such as mathematics, astronomy, computer science, quantitative biology, and statistics. Since then, many other areas have embraced the deposition of preprints — including, but not limited to, biological and medical sciences, chemistry, and social sciences — and a large number of archives have become available to researchers, with some repositories more widely used than others. Crossref — a non-profit organization that registers digital object identifiers (DOIs) for research objects — reported six years ago4 that the volume of preprints grew approximately 30% in 2016–2018, compared to an article growth of 2–3% for the same period.

But why have preprints increased in popularity? Why would researchers share preprints of their manuscripts?

Peer review can be a long process. Even though editors do their best to provide authors with a timely decision on their papers, there may be some issues along the way that delay the eventual publication of a paper. Preprints help minimize delays in communicating the research findings, circumventing the waiting period by allowing authors to publicly share and disseminate their articles as soon as they are ready to do so. As such, authors can lay claim to their findings and establish priority for the work that they have done. In addition, preprints are open access and, depending on the server, have associated DOIs, meaning that they are easy for other researchers to find and cite. Preprints can also supplement traditional peer review, as peers can discover the work and contact the authors with suggestions for improvements. Research5 has shown that preprints can improve early visibility and citations, and major funders, including the US National Institutes of Health, the Wellcome Trust, and the Simons Foundation, to name a few, have voiced their support for preprint sharing in recent years. Many universities have also encouraged listing publications on preprint servers as part of applications for faculty hiring and tenure/promotion.

It goes without saying that, because preprints can be posted without peer review, they should not be reported as established information and therefore caution must be taken when making use of the described results. Because preprints are not available only to scientists, but to the general public as well, there is a concern that potentially unreliable data can become part of the public discourse through traditional and social media — which actually happened6 during the COVID-19 pandemic. Servers such as bioRxiv and medRxiv, for instance, include a warning that preprints have not been peer-reviewed and that they should not be used to guide clinical practice. Peer review is still important, and preprints are not intended to replace scientific journals: after all, journals serve not only to manage the peer-review process but also to establish and uphold high standards and stringent policies for high-quality reporting and publishing of research findings. This does not mean, however, that preprints cannot be used responsibly for dissemination of research.

We at Nature Computational Science, together with other Nature Portfolio journals, recognize the important role of preprint posting in the process of open scientific discourse and actively encourage the posting of preprints for primary research papers, as described in our policy. Having posted a preprint does not compromise the novelty of the corresponding study, and therefore, does not jeopardize consideration of the study at our journal. In addition, the version that was originally submitted to the journal can be posted as a preprint at any time during the peer review process. Our authors can choose any preprint server of their liking; alternatively, at the time of submission, they can opt in to use In Review, a free journal-integrated preprint deposition service from Research Square. At the time of publication, Nature Computational Science offers authors the option to add a link from the published paper to its corresponding preprint, thus ensuring that the links are visible to all readers.

We believe that preprints provide a great vehicle for rapid dissemination of relevant findings. Provided that the scientific community and the media understand that the material might not have been vetted by peer review, the benefits largely outweigh the limitations.