└── README.md /README.md: -------------------------------------------------------------------------------- 1 | **The Reviewer manifesto**, Nicolas P. Rougier, Oct. 2024 2 | This document is licensed under a Creative Commons Attribution 4.0 International License 3 | 4 | #### 1. I will only review articles for which a preprint is available 5 | 6 | > A review process can be extremely long and it is not rare to have a full year between the time of submission and the actual time of publication, mostly due to the submission/review/revision/editing pipeline. If you add up several unsuccessful submissions before an article is eventually published, this can further delay publication by several years, not counting the embargo nor the paywall that may prevent a large majority of people from reading the paper in a timely manner. This is the reason why I don't want to review an article that is not first accessible as a preprint, even though this breaks the double blind peer-review system and may ultimately bias the review. This is actually what eLife decided to enforce four years ago, moving from a "review, then publish" model to a ["publish, then review"](https://doi.org/10.7554/eLife.64910) model. This is hardly different from the model proposed earlier (2016) by [Peer Community In](https://peercommunityin.org/). Since a vast majority of journals now accept preprints, this should be made mandatory at a global scale and reviewers can help this move by refusing to review unpublished manuscripts. 7 | 8 | #### 2. I will grant myself a minimum of 30 days to review 9 | 10 | > When I'm asked to review an article, this generally comes with a deadline ranging from one week to two or three weeks. Some publishers have explained me there's no need to give more time since in most cases, the review will be written at the last minute. Yes, but still, I need a minimum delay of 30 days. This does not mean I will use these 30 days to write the review, it corresponds instead to the minimal time span I need in order to fit the review in my agenda, since (can you imagine?) I've also other work to do. Peer review is a service to the community and to be perfectly honest, it is generally not high priority in my agenda. Consequently, when a publisher ask me to review and gracefully grants me 5 days to do so (extreme case), my answer is a full no. Not counting the fact that some publishers literally sell [accelerated publication service](https://taylorandfrancis.com/partnership/commercial/accelerated-publication/) to authors and subsequently pressure reviewers to write a quick review. As a reminder for publishers: we're not paid, this is a service and you're not in a position where you can request or negotiate anything with us. 11 | 12 | #### 3. I will make my review useful to authors 13 | 14 | > I've never been trained to review such that I do not know what is really expected from me by the journal nor the editor, unless a review canvas is provided. What I know however, is the kind of comments I want to receive when I submit my own manuscript. I want to receive critical and useful comments that can help me improve my work: "you forgot this important result from the literature", "be careful, your statistics may be wrong / not adapted", "this paragraph is not totally clear to me", "experiment did not really answer your initial question", "For future work (only), you could run this new experiment to strengthen your claim", "I like very much this model", "that is a clever/brillant/amazing trick", etc. When I myself review and suggest changes, I also tell authors they're free to accept the changes or not (when the proposed change is relatively minor). Of course, if I think a result is plainly wrong, I'll try to explain authors why I think it is wrong (instead of just declaring it wrong) and they can in turn prove me wrong in their answer. This happened a few times during my career when I misunderstood an argument (nobody's perfect). Discussion with authors is not a fight with a winner and a loser, in the end we all want Science to progress and we need to cooperate towards that goal. 15 | 16 | #### 4. I will make my review open access under a CC-BY license 17 | 18 | > I've encountered a case where the publisher claims a copyright on my review, forbidden me to re-use it anywhere. I was stunned by the audacity of such claim that is baseless on a legal perspective. A simple trick is thus to license my review under a CC-BY or a CC0 license such I or authors can re-use it freely later (no need to add your name if you don't want to sign your review). Now, if you wonder where I can re-use a review that is tightly linked to a manuscript, do not forget that papers are not always accepted in a journal, and the next journal might as well solicit the same reviewer because the topic remains the same. If authors did not change their manuscript, I won't change my review either. Else, I'll certainly adapt it to some extent. Note that you could also decide to protect your review with a copyright notice, preventing publishers from forwarding it to an alternate title in their portfolio (cascade model). 19 | 20 | #### 5. I will not evaluate novelty, importance nor potential impact 21 | 22 | > It is not rare to have to answer a few questions at the end of the review and to grade the article in terms of novelty, importance or potential impact. But this does not make sense to me. There exist articles that have been published in top-notch journals and [never received a single citation](https://www.nature.com/articles/d41586-017-08404-0). On the other end, there exist some articles that have been published in confidential journals or as a simple preprint that nonetheless revolutionize a whole field. Science is uncertain and we must remain humble. This quest for novelty, importance or high impact is also the main cause for the huge publication bias in favor of positive results. There was no real place for negative results until the proposal of [the preregistration revolution](https://doi.org/10.1073/pnas.1708274114). This is also what led Konrad Hinsen and myself to create the [ReScience C journal](https://doi.org/10.7717/peerj-cs.142) in 2015 that is dedicated to the publication of replication in computational sciences. No other journal would have published a computational replication at that time. 23 | 24 | #### 6. I will not recommend acceptance nor rejection 25 | 26 | > Alongside the refusal of judging for novelty, importante or impact, I do not want either to recommend acceptance nor rejection. In that context, I fully agree with eLife new policy [Peer review without gatekeeping](https://doi.org/10.7554/eLife.83889). I won't paraphrased Eisen et al arguments in favor of this policy, but I will simply insist on the added value of a public review. 27 | 28 | #### 7. I will not answer requests from no-reply mail accounts 29 | 30 | > Finally, here is a final recommendation for publishers regarding their reviewing system. If you send me an email requesting a review but use a no-reply mail address, there is little chance I'll click on any links you may provide (if your email doesn't end up in my spam folder). The reason is simple. I make a huge difference between a mail from a colleague asking me for a review versus a mass, unpersonal and anonymous invitation where you feel like a disposable resource. Furthermore, if you ask me to register or navigate through a complex website, chances are that this will diminish your chances of me accepting your review invitation. Review is a free service from us to you, so please have some consideration beyond automatic acknowledgment and make our life easier. Else, your ratio of acceptance/invitation will soon reach 0. 31 | --------------------------------------------------------------------------------