Open Access and the evolution of scholarly communication Andrew Odlyzko University of Minnesota odlyzko@umn.edu The currently prominent issue of Open Access is just one aspect of the ongoing evolution of scholarly communication. It is not necessarily the most interesting one, but it is the most pressing, because the growth in Open Access mandates could lead to a disruptive change in journal publishing. At present, the momentum seems to be towards Gold Open Access, in which authors, or authors' institutions or funding agencies, pay existing publishers. While this may not be ideal, it may represent the most practical step towards a more effective and efficient scholarly publishing future. Over the last two decades, ever since electronic publishing became a subject of wide discussion, the scholarly communication system (interpreted broadly, and so going beyond just journals and books) has undergone extensive evolution. There has been growth in intensity, variety, speed, and nature of the exchanges that take place among researchers. The extent of collaborative work (as measured by the average number of authors on a paper, for example) and the volume of multidisciplinary research have continued growing, in a trend that predates the Internet (and is shared with other disciplines). Preprints are now widely distributed through email, personal web pages, and preprint archives, and are often located using search engines. Even more strikingly novel are the ongoing massively collaborative mathematics research projects, and the many efforts in genomics and other areas where much of the activity involves curating large datasets. (While not in what used to be considered core of mathematics, these efforts increasingly do involve mathematicians.) Developing mathematical software is yet another area that is growing in importance. Many of these activities cannot be accommodated, at least not easily, and often not at all, by the traditional journal publishing framework. Still, the one element of the scholarly communication scene that, at least on the surface, has not changed much in the last two decades, is the journal. Most results are still published there (with Perelman's proof of the Poincare conjecture one of the relatively few exceptions), and promotion, tenure, research grants, and the like all depend primarily on journal publications. There have been changes in the journal system. The one that is most noticeable is that most journals are available online, and paper copies are beginning to fade away. (Mathematical Reviews has finally, but only recently, stopped producing a print version, for example. Major savings, in production costs, and especially in library costs, will result when this step is taken by most publications. Two decades ago, in contrast, there was widespread skepticism as to whether electronic versions of journals would ever be significant!) What is less visible to most scholars is that the journal price escalation that was claimed already some decades ago to be unsustainable has been sustained. What is even less well known is that for most readers at higher education institutions, the availability of journals has greatly increased. The "Big Deal" packages from publishers, consortia licensing, special deals for underdeveloped countries, and the like (all enabled by the technological developments of the last few decades, which have lowered the marginal costs of distribution) have led to a greater fraction of current journals being subscribed to by university libraries [2]. Further, the digitization of print papers has made easily accessible (either freely, or through low-cost providers such as JSTOR) much of the older literature. Together with the other developments cited above, such as spread of preprints, preprint archives, and search engines, these developments have brought us closer to the ideal of a freely accessible online "World Library of Mathematics" that has everything relevant in it. However, we are still far from that ideal, and the barriers imposed by journals supported by subscription fees are a major hindrance. Why has the traditional journal continued to thrive? It is still the repository of the "publication of record," and it is the community's desire for traditional peer review that keeps it afloat, with all its unnecessary costs and encumbrances. Novel forms of peer review have been slow to emerge and even slower to be accepted. However, they appear bound to grow in importance, and the role of the traditional one, based on journals, to shrink. Not only is there an increasing range of activities that don't fit the journal publishing framework, but the defects and deficiencies of this framework are becoming ever more apparent. Peer review is indispensable, as otherwise the "noise" generated by a spectrum that goes from crackpots to careless scholars to those who are diligent yet make mistakes (and who does not?) would be overwhelming. But traditional peer review is not foolproof, as many studies have shown. (Most of the thorough studies have been in areas such as medicine, but almost surely reflect what happens in mathematics as well.) There is also wide perception that the problem is getting worse. To some extent this may be due to the pressure on scholars to publish, so that in the rush to write, they are less willing to referee carefully. But it probably also reflects the growing complexity of the research enterprise. Arguments increasingly often are not simple ones, but complex amalgams of results and techniques from a variety of areas so that no single individual understands everything. This changes the nature of what we accept as valid mathematics. There has been rigorous debate about the validity of computer-assisted proofs (which help cope with some aspects of the complexity of modern mathematics), such as those of the Four Color Theorem, or of Kepler's Conjecture. But there are also questions about the validity of the classification of finite simple groups, with some published results explicitly making the caveat that they depend on the correctness of this great achievement of mathematics. What examples such as these demonstrate is that we are moving away from the model where a result that is published in a reputable journal is regarded as trustworthy unless shown otherwise. Instead, we will have to work with a continuum of peer review, where everything is regarded with some suspicion, with the strength of the doubt dissipating with time as more people read it and apply it. This necessarily implies a declining role for the traditional journal. However, this evolution is proceedings at the glacial pace of most changes in academia, and will likely take decades to play out. In the meantime, mandates are likely to be the main impetus towards Open Access. Whichever form, or, more likely, mixture of forms, of Open Access, is adopted, is hard to foresee, but right now it appears that Gold Open Access will be the most important. One can argue that a more desirable path would have been through Green Open Access, and with libraries and researchers collaborating to establish new, lower-cost electronic-only journals. However, the usual inertia of academia has prevented this from happening on a large scale, and in the meantime publishers have moved faster. Recent developments in scholarly publishing are best seen as a competition between libraries and publishers for resources, and publishers have been winning this tussle. What helped them, more than anything else, is that most of the costs of the academic publishing sectors are not those of publishers, but are internal to libraries, and can be decreased with the move to digital information [1, 2]. In particular, the frequently asked question as to where departments can find money to pay for Gold Open Access fees has an easy answer, namely in library budgets. (Such a move is facilitated by the fact that the current journal system is unnecessarily expensive, and much of its complexity and cost can be eliminated.) In practice, of course, the answer is not all that easy, because of the convoluted money flows in universities. But universities do have incentives to support publications by their staff, so should be able to shift the funds around. Hence I expect there would be some disruptions if a sudden shift to Gold Open Access were to occur, but I do not expect it would last long. In the long run, whether we move through Gold or Green Open Access, it seems almost inevitable that Open Access will prevail, and will be just one phase of a more thorough change in scholarly communication, in which peer review itself is changed. [1] A. Odlyzko, Tragic loss or good riddance? The impending demise of traditional scholarly journals, Intern. J. Human-Computer Studies, 42 (1995), pp. 71-122. Preprint available at http://www.dtc.umn.edu/~odlyzko/doc/tragic.loss.long.pdf. [2] A. Odlyzko, Open Access, library and publisher competition, and the evolution of general commerce. Preprint available at http://ssrn.com/abstract=2211874.