Abstracting and reviewing in the digital era Andrew Odlyzko AT&T Labs - Research amo@research.att.com May 12, 1999 The creation of electronic versions of the main print journals is proceeding rapidly, much faster than was generally thought possible a few years ago. In a recent paper, "Competition and cooperation: Libraries and publishers in the transition to electronic scholarly journals" (available at ), I predicted that these electronic versions will soon become the main ones. Print editions will either be eliminated completely, or relegated to secondary roles. This shift be driven by changing usage patterns, and even more by economics. The central feature of the existing scholarly information system is that internal library costs are at least twice as high as the budgets for acquisitions. Thus a shift towards electronic journals offers publishers a chance to lower the costs of the entire system by disintermediating the libraries, while preserving their own revenues and profits. The transition towards electronic versions of journals is likely to be accompanied by growth of bundling, in which packages of journals are offered only as a whole, without the option of purchasing individual ones. Such a move will help publishers obtain more revenue and also make it harder to cancel subscriptions. This will be made easier by consolidation among publishers, either through outright mergers or alliances. The growth of publisher oligopolies is likely to be countered by growth of library consortia, creating buyers' cartels. The net effect is likely to be much wider availability of information. With marginal costs low, it will be in the publishers' interests to make their materials easily accessible. In the long run, though, free preprints from authors are likely to undermine this system, unless publishers move on to provide more value. The only certainty is that both libraries and publishers will have to go through significant transformations to survive. What about reviewing and abstracting publications? I expect that they will also have to change to prosper, but their evolution is likely to be easier than that of libraries and primary publishers. There are two main reasons for this. One is that abstracting and reviewing publications are much less expensive than the primary journals. Thus it is even possible for departments to pay for them out of discretionary budgets, which is certainly not feasible for the main journals. The other reason for optimism about the future of reviewing and abstracting services is that they provide a more substantial intellectual contribution. The problem for the primary publishers is that their functions of production and distribution are becoming superfluous, and scholars have growing incentives to distribute their preprints for free on the Internet. Peer review is done largely by unpaid scholars. Even editing has an incentive to move towards the preprint stage, as preprints become the main communications mode among researchers active in an area. On the other hand, abstracting and reviewing publications provide services that either involve additional skilled judgement, or are too tedious to be done by volunteer scholars. Although I am optimistic about the future of abstracting and reviewing, they will also have to evolve. To justify employing people, organizations have to provide services that require human judgement. As the primary literature migrates to the Web, basic indexing and citation linking will increasingly be automated, especially as the spread of metadata makes this easier. Search engines are often claimed not to be scalable, and even today, not one covers more than two thirds of the 400 million Web pages that are thought to exist. However, the entire corpus of STM literature appears to be well under 50 million articles, and grows by at most 2 million per year. Thus current technology is already adequate to provide basic search tools for all the scholarly literature, once that literature is placed online. As entire papers become accessible with a click, the value of abstracts taken from those papers diminishes. It will therefore be necessary to continue moving up the value chain, towards more evaluative products. Other challenges for abstracting and reviewing publications include deciding what to cover. The flexibility of the Net means that we will depart from the tradition of a single definitive print version, and will have to cope with multiple versions of a work. (This will be a reversion to the era before Gutenberg, when transcribers' errors caused scholars not to trust any single version they dealt with.) As publications evolve towards multimedia formats (a process that is likely to be slow), it will also be necessary to cope with novel formats. The Information Age offers great growth opportunities, but exploiting them will require adapting to changing requirements.