Scientific literature's open sesame?

How could you make the results of the world's original biomedical research freely available to anyone who wanted them? This question remained hypothetical until the arrival of the world wide web, which allows distribution of material at only a fraction of the cost of distribution on paper. But publishing peer reviewed original research has some costs that the internet cannot magic away. Recently, a way to meet those costs has become clear. The goal of original research being free to everybody everywhere could be very close.

Currently subscribers to journals, mostly academic libraries, pay for access to scientific information. In the new model, authors (or, more likely, those who employ them or fund their research) would pay the costs of peer reviewing and electronically disseminating their articles. This one off processing charge would ensure that the article was freely available to all, forever. Journals' estimates of how much this article processing charge would need to be to cover their costs vary between $500 (£314) and $1800 per article.¹ In disciplines where authors already pay submission and page charges the change would be small.

Experiments with the “author pays” model are already under way. Over the past two years, the publisher BioMed Central has set up 90 electronic journals adopting this model. The two new journals planned by the Public Library of Science (a pressure group set up to promote open access to the world's primary research literature) will be funded this way, and several existing journals are likely to begin experiments over the next year.² In these experiments peer review would occur exactly as it does now—the author pays model would not become a form of vanity publishing.

While authors and readers have most to gain from this change, it's the funders of research that could do most to bring it about. Under the current access model, the result of hundreds of thousands of pounds of research funding may be seen by only a small fraction of the intended audience, because it is published in journals that few institutions can afford to subscribe to. For example, an annual subscription to Brain Research costs $20 000. Poor distribution limits the possibility that research could change practice—of increasing concern to funders.

This could be remedied if funding bodies earmarked just a few per cent of their research grants to cover article processing charges, recognising the costs of dissemination as a legitimate component of the total costs of research. The point hasn't been lost on some of the world's biggest research funding agencies, which have signed up to an institutional membership programme at BioMed Central. Payment of a membership fee waives article processing charges for staff. Last month the NHS in England joined a list of members that included the US National Institutes of Health, the World Health Organization, France's Institut National de la Santé et de la Recherche Médicale (INSERM), and Cancer Research UK.³ The NHS's decision is consistent with a Department of Health report stating that research and development findings should be readily available, open to critical examination, and accessible to healthcare professionals, patients, carers and the wider public.⁴ Funders could go one step further and make open access publication a condition of funding.

Even if this switch from paying for access to paying for dissemination resulted in no financial savings within scientific publishing, it would be worth doing, as the end result would be availability to all, instead of only to those who can afford subscriptions. But charging for article processing makes the costs transparent, and once users can take price into account in decisions about where to submit, the prices should fall. A genuine market could develop among suppliers, competing on more than just impact factors—for example, speed and quality of peer review.

The biggest obstacle to change is therefore likely to be the scientific and medical publishers—the fastest growing subsector of the media industry and now worth $7bn a year. While its market leader, Elsevier, generates annual profits of £290m with margins of nearly 40% on its core journal business,⁵,⁶ many publishers achieve profit margins of at least half this. “Not for profit” publishers have also been cashing in on this bonanza, becoming cash cows for the scientific societies that own them. William Arms, professor of computer science at Cornell University, likens these arrangements to “gentlemen's clubs living off their bar profits.” Societies now have a dilemma: moving to the author pays model reduces their profits and the possibility of subsidising other good works, but it furthers the dissemination of knowledge in the area of science that they foster and promote (the reason they originally embarked on their scientific publishing programmes).⁷

Surprisingly, another obstacle to the acceptance of the new model is likely to be academic institutions. While sharing the funding agencies' goal of achieving the widest possible dissemination of research findings, their reliance on journal impact factors as a surrogate for the quality of research protects the status quo. It gives no incentive for high impact factor journals to reconsider their publishing model and discourages staff from submitting research to journals that have adopted the new model but which have low impact factors.

None of the supporters of the new model believes that transition will be easy. One way of arriving there in stages has been proposed by David Prosser, based on the model adopted by several entomology journals. Authors are presented with two options: either to pay the processing charge that makes the article available to all or not to pay the processing charge, which leaves the article available only to subscribers. The result would be a hybrid online journal in which access would be dependent on authors' willingness to pay.⁸

Further tweaks can be envisaged. While current experiments charge only for accepted articles, a schedule of charges could be drawn up—starting with a small submission fee and scaling up for increasingly detailed feedback. Charging for submission would mean that the authors of accepted papers would not be subsidising the costs arising from rejected papers. Once an article is accepted for publication, its technical editing could be an optional extra that authors pay for. Decisions about readability would therefore rest with authors. A consensus is already emerging that processing charges will be waived in cases of financial hardship—for example, for researchers from developing countries.

Two years ago, we wrote of “an Aladdin's cave of ideas” in connection with PubMed Central, another initiative to free up the world's biomedical literature.⁹ Since then, we've continued to make the full contents of the BMJ freely available from bmj.com and our original research articles freely available from PubMed Central (one of few “traditional” journals to do so). We've made the full contents of the BMJ Publishing Group's specialist journals freely available to people in the developing world and we've helped WHO to broker a similar arrangement with the world's largest scientific publishers.¹⁰ But most of the world's original biomedical research remains out of reach to most people who could use it. Widespread acceptance of an author pays model could be scientific publishing's open sesame moment.