CORE Metadata, citation and similar papers at core.ac.uk

Provided by Frontiers - Publisher Connector Opinion Article published: 30 August 2012 COMPUTATIONAL NEUROSCIENCE doi: 10.3389/fncom.2012.00063 Post-publication : opening up scientific conversation

Jane Hunter*

Faculty of 1000, Ltd., London, UK *Correspondence: [email protected] Edited by: Diana Deca, Technical University Munich, Germany Reviewed by: Diana Deca, Technical University Munich, Germany

Conventional peer review: rights referees will often be doing work that has is) to work with named experts to identify and wrongs been done already on an article that was and recommend the most interesting papers Peer review is broken. We have all heard written months ago. published across 24 different subject areas in that phrase many times in recent years. It’s If standard peer review is intended to . In 2006 F1000 Medicine joined it – become a truism, a shorthand complaint help ensure that an article is good enough with the same aim, more experts and cover- about the status quo that rarely extends to be published, is it working? And in this age of 20 medical specialties. We merged the into a proposal for change. And even those context, what does “good enough” even two services in 2010, and biology and medi- who do not believe standard peer review is mean? Since most papers will eventually cine are now both covered at F1000.com. beyond repair acknowledge that there are be published, cascading until they find a Since then, we have launched F1000 problems; everyone can see the cracks. journal, that means that most papers are Posters, an repository for post- So what’s wrong? From an author’s good enough for someone and peer review’s ers and presentations – again in biology and point of view, a lot. Peer review is slow; supposed qualitative gatekeeper role is not medicine – and we are now in the early stages it delays publication. It’s almost always supportable. The impact of peer review on of launching our new open access, post-pub- secret; authors do not know who is the publication of an article is not so much lication peer review journal, F1000 Research. reviewing their work – perhaps an ally a question of yes or no, it’s more likely to be Faculty of 1000 practices two forms but, equally, perhaps a competitor. It can a question of when and where. of post-publication peer review: primary, block ingenuity; think of the classic case Yet even acknowledging the flaws, redun- open refereeing of articles after they are of Lynn Margulis and the 15 or so journals dancies, and costs of the conventional peer published in F1000 Research, and secondary that rejected her ground-breaking article review system, it is clear that we need peer review of the best already-refereed arti- “On the origin of mitosing cells” (Sagan, peer review. The more specialized science cles, published in any biology or medicine 1967) before it was finally accepted by The becomes the more we must rely on experts journal, at F1000.com. Both are illustrations Journal of Theoretical Biology. And there’s a to help us navigate the multiplicity of sub- of Clay Shirky’s “publish then filter” model lot wrong for reviewers too: what propor- ject areas we are not expert in ourselves. Peer (Shirky, 2008) and each adds value to sci- tion of referee reports are second, third, or reviewers are those experts and we depend entific discourse in its own way. even fourth round reviews? A referee’s hard on the refereeing process to protect us from I will describe our secondary post-pub- work may be contributing nothing new to sloppy work and invalid conclusions. lication review process first. an author who would rather take his or her So peer review is important but the way chances with another journal than do the it happens is problematic Secondary post-publication peer extra work suggested by reviewers for jour- At F1000, we believe that most of the review nals one to three. weaknesses of standard peer review can be The F1000 article recommendation service Does conventional peer review work for linked to two core issues, first that it is con- applies a layer of positive filtering on top publishers? Well, yes and no. Yes, at top-flight ducted pre-publication and second that it of traditionally peer reviewed literature; journals like Nature or NEJM peer review is is secret. Pre-publication peer review allows we review already-published biology and a gate keeper that helps guarantee publica- journals and reviewers to delay, filter, and medicine in order to identify and promote tion of only the most interesting articles, interrupt the essential conversation of sci- the best work. Our 10,000 named Faculty and yes, in theory, it helps guard against the ence, and secrecy makes these problems Members and their Associates select articles publication of flawed work, but it’s expen- impossible to resolve. that impress them, regardless of source, and sive – even though reviewers work for free write brief recommendations explaining – and it’s time-consuming. Nature or NEJM Post-publication peer review: two what makes the work significant and put- review thousands of papers each year that models from faculty of 1000 ting the science in perspective. These rec- would not make it into their journals; for A little background: faculty of 1000 began in ommendations and comments, along with third-, fourth-, or fifth-tier journals, some- 2002 with a post-publication review service links to the original articles, are published where further down the inevitable cascade, called F1000 Biology. Its remit was (and still on F1000.com.

Frontiers in Computational Neuroscience www.frontiersin.org August 2012 | Volume 6 | Article 63 | 1 Hunter Post-publication peer review

Why is this a useful thing to do? It’s publish revised versions on the site. All ver- to respond, to present their case. And useful because the vast volume of material sions are separately citable. All articles and because everything is happening in the published each year (or each day) makes it all versions are clearly marked with their open, interested scientists can, for the difficult for researchers to stay up to date referee status and articles that have not first time, read the back-and-forth and with their own specialized fields, let alone yet been refereed are labeled as “Awaiting make up their own minds. with peripheral fields – all those other sub- Review.” ject areas they should be keeping an eye on. The strengths of this model are that F1000 Research’s version of “publish Sure, you can search for articles and find, it’s fast, all good science can be published then filter” is an innovation in life-science more or less, what you are looking for, but immediately and become part of the record publishing and no doubt additional con- it’s helpful to have access to expert opinion to the benefit of scientists and others world- cerns will arise as we fine-tune our model. for timely guidance on what’s especially wide; it’s fair, publication cannot be blocked However, it’s clear to us that the research significant and why. The fact that F1000’s or slowed by the refereeing process; and it’s community as a whole is more than ready reviewers are named puts their opinions in open, and openness discourages bias. to contemplate and, we believe, support real perspective. No one has ever suggested that We do not see many weaknesses or risks change. Complaints about conventional, our F1000 Faculty Members should con- with this model ourselves – standard peer pre-publication, closed peer review systems duct this form of post-publication review review has few fans and is overdue for are mounting and the risks associated with anonymously. change – but then you might expect us to say our “publish first/referee openly later” sys- that. We do understand though that there tem seem relatively trivial when compared Primary post-publication peer are concerns. These include: with the increasing expense and frustration review associated with the status quo. F1000 Research, F1000’s new primary open – Is there a risk that F1000 Research will We were the inventors of and origi- access publishing program in biology and publish junk?: No, there is not. It will nal advocates for open access. We created medicine, publishes immediately, and offers publish good science and let the com- Biomed Central, helped set up PubMed fully open, post-publication peer review. We munity decide what the ultimate value Central, and fought the publishing estab- published our first articles in mid-July and of a specific piece of work is. As an aside, lishment for years to prove that open are planning for a full launch at the end of we expect that less junk – however one access can work, that it can be a profitable this year. might define that term in science – will alternative to standard subscription mod- Articles submitted to F1000 Research be submitted to F1000 Research than to els. F1000 Research and its novel publish- are first processed through an in-house conventional journals because few peo- ing model take openness to the next level. sanity check and then, assuming they pass, ple will want to see a severely negative Open access removes barriers for readers. published immediately. Post-publication review of their work become part of the Open, post-publication refereeing removes they are subjected to formal peer review. public record. Because F1000 Research barriers for readers and authors alike, and Referees’ reports are published on the site will publish immediately then review it refocuses the role of peer review from, and all referees are named. openly, sloppy work will be publicly at its worst, a behind-the-scenes variety The most important task for our referees described as such. of censorship to, at its best, the process of is to tell us immediately whether or not an – OK, if not junk then uninteresting science: expert criticism and advice that has always article is good science. We do not need to Maybe, maybe not. Uninteresting been its core and upon which the progress know if it’s exciting, or novel, or ground- science is still science, and we believe of science depends. breaking, we simply want to know that it’s it should be published. There is a rea- valid; that it’s sensible work, carefully done. son for top-line journals to sharply References We expect the vast majority of submissions Sagan, L. (1967). On the origin of mitosing cells. J. Theor. restrict what they publish, that’s how Biol. 14, 225–193. to be approved as good science. If it is good they create and maintain their identi- Shirky, C. (2008). Here Comes Everybody: The Power of science, an article will be marked as such. If ties and Impact Factors, but it’s hard to Organizing Without Organizations. Penguin Press. it’s not, or if it’s good science but the referee argue that such restrictions on scienti- has reservations, we require that the referee fic discourse are, overall, a good thing. Received: 03 August 2012; accepted: 07 August 2012; pub- lished online: 30 August 2012. add a report describing the problems and We believe they are not. Valid science Citation: Hunter J (2012) Post-publication peer review: – if applicable – suggesting improvements. should be published. opening up scientific conversation. Front. Comput. We encourage, but do not require, referees – No reviewer will want to be openly Neurosci. 6:63. doi: 10.3389/fncom.2012.00063 to add reports to articles they have approved negative about another scientist’s work: Copyright © 2012 Hunter. This is an open-access article as good science. Having now published our first articles distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and Authors have the opportunity to we are seeing in real time that this is reproduction in other forums, provided the original authors respond to a referee’s comments and are not the case. Referees are happy to cri- and source are credited and subject to any copyright notices encouraged to update their articles and ticize and authors are happy to be able concerning any third-party graphics etc.

Frontiers in Computational Neuroscience www.frontiersin.org August 2012 | Volume 6 | Article 63 | 2