Read our COVID-19 research and news.

Scientific journal articles will become freely available, thanks to new policies at major U.S. science agencies.

Scientific journal articles will become freely available, thanks to new policies at major U.S. science agencies.

Nationale Bank van Belgie - Banque nationale de Belgique/Flickr (CC BY-ND 2.0)

NSF unveils plan to make scientific papers free

The National Science Foundation (NSF) today released a long-anticipated policy that will require its grantees to make their peer-reviewed research papers freely available within 12 months of publication in a journal. The agency is not creating its own public archive of full-text papers, but instead will send those searching for papers to publishers’ own websites.

Although that’s what most observers expected, it’s not what open-access advocates hoped for. “I’m disappointed,” says Heather Joseph, executive director of the Scholarly Publishing and Academic Resources Coalition (SPARC), a Washington, D.C.–based group which represents academic libraries. But scientific publishers who worry that full-text archives will harm journal revenues praised the plan. “This is a very good way to do things because it minimizes the cost to taxpayers without having to duplicate existing infrastructure,” says Frederick Dylla, CEO of the American Institute of Physics and a board member of a coalition of publishers that runs CHORUS (Clearinghouse for the Open Research of the United States), a system for providing links to papers on journal’s sites. (The coalition includes AAAS, which publishes ScienceInsider.)

Despite some grumbling, today’s NSF announcement marks a milestone: It means that essentially all of the major U.S. federal science agencies now have a public-access policy. That reflects a push starting in the late 1990s by some scientists and activists to make the results of taxpayer-funded research freely available to the public. Since 2008, the National Institutes of Health (NIH) has required its grantees to submit their accepted manuscripts to its PubMed Central repository, which posts full-text manuscripts online within 12 months of publication. And in February 2013, the White House Office of Science and Technology Policy ordered science agencies to come up with similar policies.

However, the White House did not specify if agencies should create their own full-text paper archives or find other solutions. Federal officials were heeding concerns from many publishers that PubMed Central infringes on their copyright and that advertising and other revenues would drop if readers were not directed to journals’ own websites.

NSF does not plan to build its own version of PubMed Central. Instead, the agency will work with the Department of Energy (DOE) to create a repository for NSF-funded papers by using an existing DOE system called PAGES (Public Access Gateway for Energy and Science). The NSF repository will contain abstracts, authors, the journal issue, and other metadata. No more than 12 months after a paper’s publication, the repository will provide a link to the full-text paper on the publisher’s website or, if that is not available, to a PDF of the final manuscript hosted in a separate, full-text DOE archive. It will be a “dark archive,” however—it will be invisible to the public and exist only to preserve articles long term.

In the future, NSF may allow access to papers through other repositories, such as PubMed Central and those run by universities. But for now, the system will be quite similar to DOE’s own public-access plan announced last August.

The problem, Joseph says, is that providing access through publishers’ websites means that text and data mining across articles will be hindered by different journal policies and formats. “I just worry about having another subset of federally funded articles resident on publishers’ websites where we rely on those websites for any kind of text or data mining, then a PDF in a dark archive that we can’t do anything with,” Joseph says.

But Dylla says it’s a better system, because it points readers to the most authoritative version of a paper. Publishers are working on a way of allowing full-text and data mining across all of their journals, but there still aren’t many researchers doing that kind of study, he says. (An NSF representative says the agency “is exploring ways to enable these kinds of operations” while still “protecting the integrity of the systems and collections.”)

NSF plans to launch the repository by the end of 2015 for voluntary submissions. In January 2016, the agency’s policy will become a requirement for papers resulting from proposals submitted after that date. After they deposit their papers in the DOE-run archive—either the final accepted manuscript or the published paper—researchers must include the paper’s unique identifier when they submit research proposals and reports to their program officers, or else the paper won’t count. (The policy also covers conference proceedings.)

NSF says that because the policy is not retroactive and NSF grants run for several years, it may be 5 years before all active awards are covered.

Several other science agencies are planning full-text archives along the lines of PubMed Central. NASA is working with NIH to create a “NASA-branded” version of PubMed Central for the space agency. And the U.S. Department of Agriculture is building a full-text paper archive called PubAg. The Department of Defense, which announced its draft public-access policy this week, is creating a public repository that will include all full-text manuscripts, as well as links to articles on publishers’ websites. (Other agency policies are listed here.)

A few science agencies haven’t yet announced their public-access policies, including the National Oceanic and Atmospheric Administration and the U.S. Geological Survey.

Meanwhile, SPARC is cheering the reintroduction in the House of Representatives and Senate today week of a bill, known as the FASTR (Fair Access to Science and Technology Research) Act, which would shorten the required embargo period for sharing federally funded research papers from 12 months to just 6 months.

*Clarification, 19 March, 10:58 a.m.: A statement about why it will take 5 years for all NSF awards to be covered by the policy has been clarified.