Great expectations

The invention of the vacuum cleaner, the washing machine and the dishwasher were supposed to free housewives from drudgery. Instead, commonly accepted standards for housekeeping became higher with each new 'labor-saving' device, while the time spent on housework remained the same or even increased. New technologies create new expectations, and electronic publishing is unlikely to be an exception.

In the last decade, many journals have taken advantage of the opportunities presented by the internet. Two-thirds of all journals are now available both online and in print. Over 1000 peer-reviewed journals are published solely on the web, approximately 20% of them in the life sciences^1,2. Despite this rapid evolutionary change, online publication has recently become a lightning rod for a variety of more revolutionary ideas about scientific publishing³. However, many of the concerns discussed in this context, such as library journal pricing, restrictions on access to the literature and streamlining the peer review system, are only loosely related to one another. Separating these arguments would improve the clarity of the debate.

One component of this discussion is the belief that the web should greatly reduce the costs of publishing a research article. However, it cannot be taken for granted that switching from paper to electronic publishing alone would reduce publishers' expenses. For most journals, the main cost of publication under the present system is not paper or distribution, but skilled labor. The amount of editorial input varies considerably among journals, but in general, high-profile journals spend more on the peer review process, including the costs of considering papers that are ultimately rejected. Such filtering adds value to the highly selected papers that appear in prestigious journals, but having papers reviewed at multiple journals also increases the overall cost of scientific publication. Accepted papers then must be copy edited and formatted for publication, and this process is largely independent of whether the content eventually appears on paper or as a pdf file on a computer screen. Printing and distribution are a surprisingly low percentage of the cost of publication¹. Electronic submission can save money on administrative labor and the costs of shipping manuscripts to referees, but these expenses are typically even lower than the cost of printing. Because most of the expenses of producing a journal article do not scale with the number of copies produced, a journal's circulation is a strong determinant of the publisher's cost per copy. However, revenues (from both subscriptions and advertising) do increase as more copies are sold, explaining why many prestigious journals are cheaper than their more specialized counterparts.

Web publication also brings with it substantial new expenses for establishing and maintaining electronic archives, along with any additional searching or linking services that the publisher may choose to provide. In the future, sophisticated search options will be greatly facilitated if electronic articles can be tagged with metadata according to broadly accepted standards. Current web mark-up languages, such as HTML, label words according to their position and appearance on the page, but newer languages such as XML permit tagging for content, so that search engines can identify a string of words as “article title” or “gene” or “keyword”. Taking advantage of these new capabilities should improve the usefulness of scientific journals, but it is not likely to reduce their cost.

Reducing publishers' costs significantly would require saving money on labor. One possible solution to this problem would involve major changes to the current system of peer review. Some electronic journals have experimented with automated peer review, in which the author supplies key words and a database program randomly assigns referees who are described with similar key words. More radical proposals include posting unreviewed papers on a web site and allowing readers to comment online, or selecting papers based on the number of hits they receive from browsing scientists. Such approaches would undoubtedly result in substantial cost savings, but they also run the risk of forfeiting quality control.

Perhaps greater efficiency can be realized without loss of quality. Another effect of electronic publishing is that it has become easier to test new business models, because the web lowers barriers to entry into the publishing business. New electronic-only journals such as Journal of Vision and the Biomed Central (BMC) journals are peer reviewed and provide free access to primary research articles. J. Vision charges a processing fee to authors, and the BMC journals plan to do so from January 2002. A more radical attempt at quality control is the 'Faculty of 1000' initiative⁴, a soon-to-be-launched subscription service that will separate the filtering from the publishing function, by rating individual articles from other publishers based on the recommendations of well-known scientists. It remains to be seen which of these approaches will prove economically viable.

The current system of scientific publication has evolved over time in response to a variety of selective pressures, and like a biological system, aspects of it are undoubtedly shaped by history rather than function. Because there is no central authority to impose a solution, even revolutionary ideas will have to compete with other models for acceptance among authors, readers, referees, librarians and publishers. In a time of rapid change, perhaps the best strategy will be to occupy as many ecological niches as possible. Whether the future evolution of scientific publishing will be gradual or catastrophic remains to be seen, but it seems likely that the changes will include more competition among a larger number of publishers, which should be good news for scientists.