The Public Library of Science and the ongoing revolution in scholarly communication Andrew Odlyzko AT&T Labs - Research amo@research.att.com http://www.research.att.com/~amo I enthusiastically support the goal of making scholarly articles easily available on the Internet to everyone, without any fees or other barriers to use. Yet I have not signed the Public Library of Science (PLS) petition. My own contribution to the freeing of scholarly literature (aside from propagandizing in its favor through articles and lectures) has been to make all my recent preprints available for free on my home page (and on preprint servers in many cases). In copyright transfers to publishers I have been reserving the right to keep the preprints posted on the Web, and I have been urging others to follow the same policy. The difference between my policy and that of the PLS (which requires publishers to make articles available for free access from centralized servers within half a year of publication) is one of degree. Both courses of action produce improved access to scholarly publications. The improvement is especially dramatic for the general public, but is also dramatic for those fortunate to be at the few hundred institutions around the world that have good libraries. Both courses of action also serve to push scholars, publishers, and librarians towards the new era of learned discourse that is evolving. The reason I do not endorse the PLS petition is because it assumes a certain fixed model for scholarly publishing. By requiring free public access to published articles after six months, but not earlier, it implicitly says that publishers need some barriers to induce subscribers to pay. Yet why should it be six months, and not six days, or six weeks, or six years? Further, the petition requires posting the articles on centralized servers. Yet the choice between centralized versus distributed databases is by no means clearcut. I do not wish to commit myself not to publish in outlets that might wish to experiment with different policies. The PLS petition also fails to promote the free circulation of preprints. While published papers that are peer-reviewed might be of greater utility to the general public, for active researchers it is the early versions that matter the most. Yet some journals still adhere to the policy of refusing to consider for publication papers that have been widely circulated as preprints. That serves to impede progress in science, and should be discouraged. The one thing that is certain in these uncertain times is that there will be much experimentation. This is unavoidable, since nobody can be certain of how scholarly communication will evolve. We will be working our way free of the shackles imposed by Gutenberg's print technology and exploring the novel flexibility of the electronic medium. A prominent feature of the evolution that is unfolding is the speeding up of communication. A recent article (by A. M. Campbell in "Science" in April 2001) about a new high temperature superconductor noted that "...every superconductivity laboratory in the world immediately began to make measurements on this new material and dash into print. Fifty preprints had been posted on the Web by the end of February--before the original paper was even published." Some traditionalists bewail this hurried pace of research and publication, but that is how the world is evolving. No group that has embraced rapid electronic communication has been willing to give it up. The leisurely pace we have grown used to was forced on us by the print medium, and was not a result of an informed choice. The faster pace of communication, with a variety of different methods of communication, including preprints, as well as even less formal methods such as phone, fax, and email, is creating a continuum of publication. This will require a continuum of peer review as well. Some of the opposition to the PLS (or earlier to any kind of electronic publishing) was based on fears that the peer review system might collapse. I am not concerned about that danger, as displacement of traditional journals from their central role will not be too rapid, and in any case novel forms of peer review could be quickly set up. The one thing that modern communication does enable scholars to do is organize quickly. Furthermore, novel peer review schemes are not only desirable, but inevitable, and are growing rapidly. (Note that the fifty papers on the new superconductor did not get written at the same time by chance. Some form of informal peer review helped persuade the authors that this was a promising subject to investigate.) Some of the coming transformations may appear uncomfortable today. For example, the notion of a final definitive version of an article, which seems so basic to scholarly publishing, is likely to fade away. It did not exist before the economics of print forced it on us, and is likely to play a much reduced role in the future. As an example, could anyone propose a definitive version of the human genome database? It already is a living object, constantly enlarged, corrected, and updated. That is how an increasing fraction of scholarly communication is likely to be handled. While I do not endorse PLS fully, I do see it as a sign of an imminent transition in scholarly communication. The huge number of signers of the petition shows that scholars are waking up to the opportunities that free distribution of their works offers to them as authors and to society in general. The scholarly publishing area is full of complicated feedback loops and perverse economic incentives to the extent that I have often compared it to the American medical sector. Both fields are full of inefficiencies and resistance to change. Moreover, there are no magic solutions in either one. Many simple solutions (such as demanding lower prices from journal publishers) are doomed to fail, since they ignore not only the dynamics of the capitalist system, but also where the real costs of the system are (namely inside the libraries). There are also paradoxical phenomena, such as print sales increasing as a result of making a complete book or journal available for free downloads from the Web. All these factors make it impossible to plan the evolution of scholarly publishing. However, rapid evolutionary change is coming, especially since the authors, who have the greatest power, are slowly realizing this and beginning to accelerate the pace of change. Several previous attempts at boycotts (for example, of very expensive journals) have failed. The PLS may also fail. Yet change is coming. We are reaching the point where even in fields that have not traditionally relied on preprint distribution, there are demands for freer circulation of preprints and reprints. What used to be of interest only to a small group of early adopters, is now becoming part of the mainstream. We are entering the period when the new rapid communication technologies will begin to dominate in very visible ways. The transition to the new era will not be easy, but it does offer enough opportunities that it will accelerate. It is exciting to watch this evolution, even if the slow speed at which it happens is often frustrating. Andrew Odlyzko is a researcher and manager at AT&T Labs - Research. Aside from technical work on computational complexity, cryptography, number theory, combinatorics, coding theory, analysis, probability theory, and related fields, he has also been investigating electronic publishing, electronic commerce, and economics of data networks. His papers on electronic publishing and other subjects are available on his home page, .