Librarians at Cornell University want more external funding to support their popular arXiv preprint server because the running costs are now “beyond a single institution’s resources”.

arXiv has become the most widely used preprint server among academics in the physical sciences. It received more than 60,000 new submissions in 2009, has about 400,000 registered users and provides 2.5 million article downloads per month. Its rate of expansion is so rapid that staff expect its budget – which covers personnel as well as operating expenses – to increase from $400,000 in 2010 to $500,000 in 2012. Most of the top 25 institutional users have already made financial commitments, but to meet the budget demands Cornell Library would like several hundred others to pledge support too.

According to a white paper on the arXiv website, Cornell is looking to introduce a three-tiered model in which the top 100 institutions, based on the previous year’s download activity, will be obliged to pay Cornell Library $4000 a year. For those ranking between 101 and 200 in downloads the suggested fee will be $3200 a year, while for those below 200 it will be $2300.

“Keeping an open-access resource like arXiv sustainable means not only covering its costs, but also continuing to enhance its value, and that kind of financial commitment is beyond a single institution’s resources,” says Oya Rieger, a Cornell University librarian who manages content on the server. “If a case can be made for any repository being community-supported, arXiv has to be at the top of the list.”

Unabated growth

It’s difficult for the library to maintain an indefinite commitment to unilateral support of a resource that provides so much benefit outside of the university Paul Ginsparg, Cornell University

Paul Ginsparg, the Cornell physicist who created arXiv as at the Los Alamos National Laboratory back in 1991, and who won a MacArthur Fellowship for his efforts, shares the same view. “It’s difficult for the library to maintain an indefinite commitment to unilateral support of a resource that provides so much benefit outside of the university,” he told “The usage levels, submission rates, external enthusiasm, visibility, impact, etc continue to grow unabated, so it’d be useful to assess whether this can be made a community-supported resource in the long term.”

Indeed, Cornell Library has long-term plans for even more community input. Along with consistent underwriting from the library’s own budget, it would like to seek support from other libraries, societies, endowments or funding agencies, such as the US National Science Foundation. It is also looking to strengthen ties with collaborations – such as the INSPIRE project by the particle-accelerator labs CERN, SLAC, DESY and Fermilab for collating high-energy-physics literature – to improve its service.

arXiv has become such a useful resource to researchers in many areas of physics, [so] it doesn’t seem at all unreasonable that the burden of financing it should be spread a little more evenly among its users,” says Alison Wright, editor of the journal Nature Physics.

Important resource

Gene Sprouse, editor-in-chief of the American Physical Society, agrees. “This important resource for the community of scientists costs money to run even though everyone, including me, would like information to be free,” he says. He points out that the APS’s 10-year-old open-access journal Physical Review Special Topics: Accelerators and Beams uses a similar funding model to that proposed by Cornell Library in which cash comes from an international consortium of accelerator labs. “The primary users of arXiv form a similar small community with a common interest, so there is a good chance for success, as with our journal,” he says.

However, the “institutions pay” model is not the only way to fund open access. The Institute of Physics (which owns and the Deutsche Physikalische Gesellschaft publish New Journal of Physics, which offers free access to all papers by charging authors publication fees. Tim Smith, who is the senior publisher in charge of New Journal of Physics, suggests that arXiv could consider charging a small submission fee to authors to cover costs. “It would be very interesting to see how the physics community would respond to a submission fee on arXiv,” says Smith.

One benefit of this approach is that income would be proportional to the number of papers processed – and processing papers is often the most costly part of an online publishing operation. According to arXiv’s own figures, about $7 per submission would cover the 2010 budget.