A new National Academies report envisions a future research enterprise designed around open science while acknowledging significant challenges in transitioning to such a system.
National Academies of Sciences, Engineering, and Medicine
A National Academies report released last week envisions a research enterprise in which open science is the “default approach.” The report identifies researcher adoption of open science practices as central to moving toward such a system, and calls for “persistent, coordinated action” from research institutions, funders, professional societies, and government, to provide needed support and incentives.
The committee that produced the report was not asked to evaluate whether moving toward open science is the right path. Rather, it was instructed to take for granted open science as a worthy aim and set out to explore how to overcome barriers to its adoption. Their report provides five main recommendations outlining steps research stakeholders can take to integrate open science into the research enterprise “by design.”
The sole sponsor of the study was the Laura and John Arnold Foundation, which also funds the Center for Open Science, a nonprofit organization that aims to "increase the openness, integrity, and reproducibility of scientific research." The chair of the study was Alexa McCray, a professor at Harvard Medical School.
Open science presented as historic milestone
An umbrella term, open science as defined in the report encompasses free public access to scientific publications as well as data and code that underlie them.
At the July 17 release event, National Academy of Sciences President Marcia McNutt placed the open science movement within the “vast arc of human history,” declaring it the fifth and latest notable milestone in the “purposeful sharing of information.” McNutt said the first was the development of language by early humans, the second was the development of written language, the third was the creation of libraries, and the fourth was the creation of the research journal to disseminate the results of scientific research. This model of publishing research journals first adopted in the 17th Century is still essentially the one used today, she said.
McCray described the open sharing of information as “fundamental to the progress of science and to the effective functioning of the research enterprise.” Listing the benefits to science, she pointed to greater rigor and reliability, faster and more inclusive dissemination of knowledge, and broader participation in research. McCray also observed the emergence of open science coincides with the revolution in the digitization of information. The report predicts that open science, in leveraging the digital revolution, will also revolutionize the practice of science, enabling the automation of search and analysis of linked articles and data to reveal patterns that would otherwise be undetectable.
The report notes that open science dovetails with civic goals as well, such as providing taxpayers access to the results of federally funded research and enabling anyone, including private citizens, to scrutinize, challenge, and build on scientific investigations.
Report calls for culture shift across research enterprise
The report’s first recommendation is for research stakeholders to work together to build a culture that is supportive of open science. For research institutions, it prescribes changes in researcher evaluation practices and the introduction of other incentives to better reward open science practices. McCray pointed out that the criteria currently used to determine eligibility for tenure at universities are primarily based on bibliometric factors such as Journal Impact Factor, and suggested other factors be given equal or greater weight. The report also praises efforts such as the San Francisco Declaration on Research Assessment that seek to achieve broad buy-in from stakeholders on alternative evaluation frameworks.
The report stresses the culture shift must take place at all levels, with researchers taking full use of opportunities for making their research products openly available, and funders rewarding those who host their data and products in publicly accessible databases. As an example, the report points to a recent policy change by the National Institutes of Health that encourages investigators to use and cite interim research products such as preprints when applying for funding.
McNutt urged all research stakeholders, including funders, universities, journals, and scientific societies, to “reward the right behavior.” She stressed,
We have to change the culture within the research community so that open science becomes the easy, the accepted, and the default path of least resistance.
Training, data stewardship, and ‘FAIR’ principles also key
The report’s second recommendation is for research institutions and professional societies to train students and other researchers in open science practices. For instance, the report says universities should integrate training in a range of open science best practices into graduate and postgraduate curriculums, and research funders should require training as a part of certain federally funded grants.
McCray said researchers can embrace open science during all phases of the research process, arguing,
Rather than something you do at the conclusion of your research, open science should be a byproduct of how you’ve done your research all along.
Other recommendations focus on scientific data, which the committee expects will become the “dominant resource” of the future ecosystem of science. The third recommendation calls for research institutions, scientific societies, and research funders to develop policies and procedures for preserving scientific data and other research artifacts for long-term public availability. Acknowledging that significant resources will be required to fund the necessary infrastructure, the committee suggests they come via “an ear-marked percentage of each funded grant.”
The fourth recommendation calls for the universal adoption of the “FAIR principles,” which hold that data ought to be Findable, Accessible, Interoperable, and Reusable. To this end, the report calls for scientific societies and research funders to “network and federate” existing open repositories to improve discoverability of scientific results and interoperability of data. It also calls on research funders to commission an independent evaluation of current university and federal data archives to assess the extent to which they currently adhere to FAIR principles.
The fifth recommendation underscores the importance of research community stakeholders each doing their parts to accelerate progress and realize the vision of open science by design. Among other actions, it calls on the federal government to update the White House Office of Science and Technology Policy’s 2013 and 2014 memoranda that laid out open science policies and requirements for federal agencies. Although the science agencies have all developed open access plans to comply with the memos, a 2017 analysis published by the Association of American Universities and Association of Public and Land-grant Universities called for more coordination and consistency across agencies.
Significant barriers to achieving open science acknowledged
The report acknowledges significant barriers and limitations to achieving open science, among them the costs of implementation and widespread infrastructure, the inertia in current business models of scholarly communication, differences in practices among scientific disciplines, and the imperative to keep sensitive personal and national security-related information confidential.
The committee finds open science has made steady but “uneven” progress in recent decades, concluding “the research enterprise remains some distance from achieving complete open science.” For instance, it notes a large percentage of the world’s scientific literature is still only available via subscription, and finds wide disparities in openness by discipline.
Stressing the significant differences in how open science will need to be implemented across disciplines, McCray highlighted the advantages of “community-driven” efforts in addition to top-down mandates and policies, and pointed out a number have already shown success. In particular, she said the early establishment of the arXiv pre-print repository at Cornell University in 1991 “has expanded access without disrupting scholarly communication in physics and astronomy.” The report also praises the Sloan Digital Sky Survey for making large amounts of data available to the public under open science principles, characterizing it as one of the “largest, most detailed, and most cited surveys in the history of astronomy.”
The Sponsoring Consortium for Open Access Publishing in Particle Physics (SCOAP3) is highlighted as an example of how stakeholders in a scientific field can arrive at a voluntary arrangement for open publishing. Under the arrangement, all high energy physics articles published by the American Physical Society, an AIP Member Society, are paid for centrally by the consortium and made openly available. In turn, APS reduces the journal subscription rate it charges libraries.
The report observes, though, that the most prestigious journals, such as Science and Nature, are still not open access, and that non-profit scientific societies depend on revenue from scholarly journals to fund a range of community activities, such as annual meetings, education and communications programs, and advocacy efforts. The report emphasizes that “careful planning of stakeholder buy-in” is needed to avoid unintended negative consequences and disruptions.
Particularly, the committee warns that an immediate move toward a “universal gold” open publication mandate, in which all papers are immediately available to readers at no cost upon publication, would have negative consequences. Instead, it suggests a positive and feasible short-term step might be a “universal green” mandate, in which authors are given permission to self-archive a version of their article in an open access repository as it is published.
Disclosure: AIP is supported in large part through revenues generated by AIP Publishing, an independent, nonprofit subsidiary of AIP that publishes scholarly journals in the physical and related sciences.