Tuesday, May 19, 2015
Towards an open scientific publishing system for the digital era
The current system of scientific publishing has existed for nearly 350 years. It is based on peer-review and publishing in scholarly journals and books. The purpose of peer-review is to ensure minimum levels of scientific quality and relevance, whereas publishing companies bear the costs, normally recovered by means of subscriptions and book sales. This system has worked well, implementing rigorous examination of scientific work and fueling scientific and economic development through the efficient dissemination of results. It is unfortunately far from perfect. I won't repeat all criticisms here, more thorough analysis can be found almost everywhere in the internet as, for example in Donald Forsdyke's page and this article by Larry Wasserman. For the sake of this post, suffice it to say that there is concern among many scholars that the ills of the system have largely outweighed the benefits. One particular criticism is the case when journals abuse their position as the bearers of scientific message. Established journals may sometimes fall in the hands of large publishing companies that demand exorbitant sums of money for the subscriptions. The severity of this problem can be better understood when one thinks that peer-review is an unpaid service that every scientist does to the scientific community. Also note that in the case of books, authors usually get very little pay in comparison with the cost of each book. In every case, most of the money usually goes to the publishing companies. The cherry of the cake is that these companies usually demand the transfer of copyright from the authors! As a result of this system, many of the important journals in most fields are unaffordable for the general public and scholars not associated to large libraries at universities and research centers. In particular, libraries in poorer regions and countries can seldom afford journals and books, providing an evil positive feedback between poverty and ignorance. Another not unusual irony is that these subscriptions are paid with tax money, but the average person on the street cannot access the scientific literature she or he helped to pay! It is easy to see that this situation contributes to make the learned discourse elitist and hinders the development of society, in another ironic joke of history, as this was the prevalent situation at the start of the great scientific revolution after the Renaissance, which the practice of publishing and dissemination helped to eradicate. The old system of scientific publishing has served science well for a long time, but has now become outdated and possibly irrelevant in the modern era. Fortunately, a lot has happened in science publishing since the times of Galileo and Halley. Besides digital media and instantaneous worldwide information flows, we also have the concept of open science, a movement encouraging a more open scientific culture, in which there is a transparent development, dissemination, discussion and use of scientific results by society. In line with this philosophy, why not try to think what is needed for a publishing system in which: - anyone can submit a piece of scientific work, - anyone can access it, - anyone can peer-review, criticize, and comment it, and - anyone can use, modify and build upon it? Such a system would ensure unrestricted access to scientific results by everyone, making the process of scientific discussion more transparent, and enhancing development because everyone will be allowed to see it. At the same time, quality assessment would be made more significant and reliable because peer-review would be done in a massive scale, instead for the typical two or three reviewers of the traditional system. None of these ideas is really new and even the first and fourth are already part of the traditional system. Open Access journals implement the second, and in a few cases also the third (as The Cryosphere). ArXiv and Google Schoolar, among others, provide examples of how parts of this can be implemented technically. In fact, the requirements that such a system should pose are fairly minimal for today's technology: - at least one repository for the material, akin ArXiv; - user registration and login, with a unique identification, for submitting, reviewing, voting, commenting and downloading (enforces a minimum of etiquette, common to a vast repertoire of discussion forums); - some minimal layout, like the classical title, author(s) name(s), abstract, results, discussion and references, to make search easier; - unambiguous attribution of credits and rights to authors, providing some sort of legal protection like Creative Commons; - votes restricted to one (up or down) per user, perhaps with the possibility of regret and reverting the vote (StackExchange works this way); - and a good search engine! Obviously, these ideas pose a lot of questions, mostly practical, but also a bit philosophical. For example: Q: who will pay for such system to work? A: The examples mentioned above give some good answers. ArXiv is supported by a big donation, and Google Schoolar is maintained by Google. In addition, it is not too hard to imagine that such a system would be incomparably cheaper than the enormous sum of money spent in journal subscriptions. Q: who is going to "weed out" bad papers if there are no reviewers and editors? A: Crowdsourcing! People reading the posted articles can comment and vote, up or down, giving grounds for evaluating the quality of the scientific contribution made. It is not difficult to imagine how this elementary procedure can provide a basis for much better measures of quality and productivity for individual researchers than the so-much enshrined metrics in vogue today. I am also tempted to answer with another question: do we really need someone else to decide what we should read? After all, we are adults and, hopefully, as educated people, it is up to us to decide what is relevant and good for us. Q: there will be too many papers around! A: Yes, but in this case this is a good thing. Let's not forget that there are already many papers around right now and it is often difficult to find everything that is relevant. We only need good search engines! An open peer-review system, with voting and commenting provides convenient measures to select what you want to read. One could, for example, choose the most up-voted or, better still, the most commented papers in a given field. Compare this idea with with the current practice of selecting what to read among the plethora of papers around. Q: who would like to submit a paper to such an open repository, that has no "status" or "recognition" within my field? A: Me! and surely thousands of other scholars that already use ArXiv. Metrics such as "impact factor", which many people consider relevant, are nothing but the statistics of a social phenomenon within academia, namely that of attributing importance to a handful of journals, leading to many others to feel compelled to publish there. I do not see any problem in people being compelled to publish in a place with better philosophical grounds. It is just a matter of critical mass. Plos showed this is possible. I am confident that a system working along the lines described above would contribute to solve many of the ills of the current system, while making substantial progress towards a more open, democratic and free access to scientific results. In addition, problems like merit and precedence attribution would be more transparently handled. As Michael Nielsen wrote in his book, perhaps the hardest difficulties in starting to move in this direction are not technological, but cultural. In particular, fear, distrust, conservatism and other old cultural barriers amongst academicians are particularly hard to remove. The example of the ArXiv experience shows, nevertheless, that this is possible. NOTE: This post contains some of my meditations on this subject, but I make no claim to originality. I have been reading and thinking a lot about this for years, so I am very probably making a cocktail of ideas, most of which are almost certainly not mine. Please feel free to comment if I am missing an important citation!