Completed on 20 Mar 2017 by Ira J Blader. Sourced from http://biorxiv.org/content/early/2017/03/15/110858.
Login to endorse this review.
Thank you for stimulating the much-needed discussion on the role that preprints may have in publishing in microbiology and likely other biological fields. As you and others have noted, scientific publishing is at a breaking point due to a variety of factors, and there is excitement (that I share) of the potential that reprint servers have in effecting important changes in this process. Three of these factors are the peer-review process, paywalls, and a paper’s fit with a journal’s perceived impact. Pre-prints have been proposed to help with the key goal of disseminating scientific knowledge as quickly as possible while also offering the ability to obtain reader feedback. Dr. Schloss’ Perspective does an outstanding job of discussing how pre-print servers do this. However, I have several questions and concerns about preprints:
1. The premise that a paper can be improved by reader comments is certainly valid. In my own anecdotal experience as an editor at the ASM journal mSphere, the first, and so far only, paper that I handled that had been posted on bioXriv and had gone through comments and revisions was perhaps the paper that was the best received by my selected peer reviewers. However, I am concerned that as we become more invested in preprint servers that reader weariness and complacency will become an issue. Time is one of our most cherished resources and the continued posting of work that requires colleagues’ feedback will necessitate readers to continuously check for updates on each paper in their field. I understand that journal peer review isn’t perfect, but at the end of the process it results largely in the same manuscript as would have been developed through bioRxiv or other preprint servers.
2. A key advantage to preprint servers has been proposed to be the near immediate dissemination of scientific knowledge without paywall restrictions. While this is true, the danger with this is that inherently incorrect information/data/conclusions can be posted and in many cases the initial version is the one that is remembered and often is difficult to reconcile. The autism-vaccine link paper is probably the best example of this. While that paper had gone through established peer review at a highly regarded journal, it did so at a time when fewer avenues of disseminating scientific information were available. The ever-increasing use of social media coupled with today’s political environment can lead to devastating consequences, and I am not aware of any mechanism in place in which a paper can be taken off of bioRxiv for scientific inaccuracies. Over the past decade, publishers have learned that that have an obligation and vested interest in ensuring that their journals publish reputable data as well as in notifying the public and removing papers whose scientific basis becomes suspect.
3. The ability for the community to identify data manipulation and scientific misconduct in a preprint is important, and I agree is a major advantage of preprint servers. However, important issues need to be resolved. First, accuser anonymity is an important protection when a suspected fraud concern is raised, and publishers act as a firewall to maintain this. Second, it is unclear how whistleblower protection laws will apply to raising concerns of papers posted online. Third, a scan of the Retraction Watch website reveals that many instances of scientific misconduct are caught long after a paper is published, and it is unclear whether bioRxiv can improve this situation.
4. One of the key complaints that many scientists have with peer review is that journals make acceptance decisions based how a paper’s perceived impact and significance fits with a journal’s stated standard. Often these decisions are made regardless of the paper’s scientific soundness and sometimes are rendered only after a number of rounds of peer review. The need to publish in high impact journals is well documented and preprint servers will likely not change this unless publishing models change or how the scientific community weighs the importance of where a paper is published. As Dr. Schloss notes here, journals using preprint servers as clearing houses to identify papers that they’d like to handle may aid in alleviating this choke point in publishing.
In summary, I appreciate Dr. Schloss’ perspective article and how it will hopefully trigger continued discussion on how preprint servers can improve scientific publishing. While I am a current skeptic of them, I am optimistic that change is coming for the better and perhaps as they evolve preprint servers will play an important role. I will be glad to be converted from my skepticism.