The current repository landscape

As described in the first edition, a repository is, at its core, a ‘digital assets management system of some kind or a network of systems that allows for the deposit and subsequent distribution of digital files over the Internet’ (ibid.: 198). But beyond this very basic description, repositories vary wildly in terms of content, access, management, communities and purpose; though, for the purposes of this discussion, I will be limiting my description of the landscape to those repositories that have some relationship to scholarly publishing and communications.

Conventional repositories have focused on content, often in the form of scholarly papers, generated by researchers either within a specific institution or set of institutions (institutional repositories) or for specific disciplines (disciplinary repositories). The OpenDOAR directory (http://www.opendoar.org/) lists over 2200 institutional and disciplinary repositories. But the landscape is much more diverse than even four years ago. Repositories focused on content, such as those for data, code, even scientific protocols, are now more common; these are sometimes quite closely linked to publishers. There are also services such as Mendeley and Academia.edu, whose primary focus may not be their repository role but who do allow researchers to upload their research and share it (often just to specific groups). These changes have made it more difficult to define and count what a repository is. While OpenDOAR and the Registry of Open Access Registries (ROAR; see http://roar.eprints.org/) can both be used to explore numbers and characteristics of conventional repositories worldwide, they do not list all repositories, particularly those that may not explicitly describe themselves as such.

Content within repositories can include: some version of published journal articles, book chapters and books; conference papers and posters; theses and dissertations; video and audio files of talks and lectures; original undergraduate research; technical reports, white papers, and other forms of grey literature; research data in all formats; and code, scripts and software. The repository, IDEALS (http://www.ideals.illinois.edu/), for which I am responsible at the University of Illinois at Urbana-Champaign, contains examples of all of these. For many institutional and disciplinary repositories, the major focus continues to be document-like objects, whether those are grey literature, theses and dissertations, or published literature. There is generally no peer or editorial review of the material deposited in the repository; for an institutional repository, the author’s affiliation combined with a repository’s collection policy is usually sufficient for accepting a deposit. Disciplinary repositories have a variety of methods to vet deposits – arXiv, for example, may require that a potential depositor be endorsed by a current contributor (see http://arxiv.org/help/endorsement). More repositories are beginning to include means for commenting or otherwise interacting with a contribution, as well as measures of impact; for example, figshare (http://www.figshare.com/), a repository that allows researchers to share and make available material supplementary to an article, allows comments, ‘shares’ via Twitter and other social tools, and downloads.

Most repositories provide open access to the majority of the content within them; where there are restrictions they are often in the form of limited embargoes or restrictions to a specific community (for example, the campus in the case of an institutional repository). Generally, the descriptive information, or metadata, is openly available for indexing via general and specialized search engines such as Google or Google Scholar; some are also open for harvesting via protocols such as the Open Archives Initiative Protocol for Metadata Harvesting (OAI-PMH) or via specific application programming interfaces (APIs). While the content and metadata may be openly available, for the most part, most content in repositories – particularly within institutional and disciplinary repositories – is not free from restrictions on use; in most cases, the authors or publishers still exert full copyright protections over the content. This is less true in some of the more specialized repositories, particularly those connected to the Open Science movement, defined by Peter Suber as ‘combining OA texts, open data, and open-source software and providing these sorts of openness at every stage of a research project, not just at the end in reporting results’ (Suber, 2012: xi); these repositories often use Creative Commons licences (http://www.creativecommons.org/) in order to explicitly allow reuse of code, data and research outputs.

Repositories also have varying commitments to the long-term preservation of the content within them. In their 2011 study of 72 research libraries with institutional repositories, Li and Banach found that 97.4 per cent included preservation within the mission of the repository, although how well preservation activities are actually integrated into the repository has been questioned (Li and Banach, 2011; Rieh et al., 2008). In some well-established disciplinary and content repositories such as arXiv (http://arxiv.org/) or ICPSR (http://www.icpsr.umich.edu/) there is an explicit commitment to perpetual, persistent access to the materials deposited, but this can be more difficult to assess in others where the primary focus is access and networking between researchers. As repositories take on roles of providing access to supplementary materials to journal articles (such as data or code), the lack of attention and clarity concerning preservation strategies and infrastructure becomes of increasing concern.

Repositories and open access to the published literature

Most of the debate about repositories and their relationship to scholarly publishing has occurred within the sphere of the repository’s role in providing open access to the published literature. OA via self-archiving (i.e., ‘green’ OA) relies on repositories for the infrastructure to provide this access (though a similar result can be achieved simply by posting papers on a website) (Harnad et al., 2004). The research is (or may be) published through traditional means, but the author retains the right to make some version of that research available online. Generally, the version that may be made available is either the pre-print or the post-print. The pre-print can be defined as the version prior to submission or as submitted for publication, but not yet reviewed, edited and/or published; this would be the Author’s Original or Submitted Manuscript under Review under the definitions recommended by the National Information Standards Organization (NISO). The post-print is generally either the author’s final manuscript version (post peer review but before final formatting and editing) or the published version (with the formatting, pagination, and headers and footers of the journal); this would be the Accepted Version or the Version of Record under the NISO recommendations (NISO, 2008). As of January 2013, according to the SHERPA/RoMEO directory of publisher policies on self-archiving (http://www.sherpa.ac.uk/romeo/), some 68 per cent of the 1196 publishers included allow some version of an article to be self-archived; 60 per cent allow the Accepted Version to be deposited, and 25 per cent allow the Version of Record to be deposited. These numbers include very large publishers (such as Elsevier) and small publishers, university presses and scholarly societies of all sizes.

While these numbers may appear to be good news for repositories focused on open access to the published literature, a number of factors complicate the reality. Publishers’ policies, while technically allowing self-archiving, can include any number of conditions, such as embargoes and restrictions on the type of repository into which an author can deposit. Authors must understand these policies and, in some cases, negotiate for more specific rights to make their work openly available. Authors must be comfortable, in most cases, making available a version of the research which is not the Version of Record. Authors must be motivated to undertake the work of deposit in the first place. For many disciplines, OA is not an expectation, and early sharing of research is not common nor accepted outside of traditional conference venues. These factors are not trivial impediments; current estimates are that only between 11 and 15 per cent of total research output is freely self-archived (Björk et al., 2010; Suber, 2012: 58). Disciplines in which open sharing of research before publication is common (such as high energy physics and economics) may see a much higher percentage of deposit, but these appear to be the exception to the rule. This is despite some evidence that researchers support the principles of OA to academic research (Repository Support Project, 2011). Whatever the reason for low deposit rates – lack of direct incentives, difficulty in determining rights issues, little awareness of repositories – it seems clear that even with directed and concerted outreach most repositories are unlikely to affect the scholarly publishing system any time soon.

But what if researchers were required by their institutions or funders to make their research openly available? In early 2008, Harvard University’s Faculty of Arts and Sciences (FAS) voted unanimously to do just this; the faculty gave the university the rights to make their published articles available in Harvard’s institutional repository, DASH (http://dash.harvard.edu/). (For the text of the resolution see http://osc.hul.harvard.edu/hfaspolicy.) According to ROARMAP some 138 institutions and academic departments and colleges worldwide of all sizes and types have established such policies since 2008, including Duke University, Massachusetts Institute of Technology (MIT), Oberlin College, the University of Kansas, the National Institute of Oceanography, Queensland University of Technology, Trinity College Dublin and the World Bank. In the majority of cases, particularly those within academic institutions, these policies have been driven by the faculty themselves and are passed by faculty senates; this is of particular importance as the impetus for change is not driven by the library (although certainly the library may be the instigator) but by the producers, editors and peer reviewers of the research.

Typically, these policies (which usually grant the institution non-exclusive, joint copyright in a work) allow an author to opt out if a publisher will not allow self-archiving and they do not specify where a researcher must publish. Indeed, a proposed OA resolution (not a policy, that is, non-binding) at the University of Maryland failed in part because it specifically encouraged faculty to publish in OA journals where practicable (Hackman, 2009). Institutions with OA policies provide support through the provision of addendums for publishers’ copyright transfer agreements, the harvesting of content directly from publisher sites (where possible), and negotiation with publishers to mitigate the need for the institution’s authors to negotiate to retain rights (Duranceau and Kriegsman, 2013). MIT, for example, has negotiated with Springer to allow articles by MIT-affiliated authors to be ‘archived and/or deposited in any repository, or used for any scholarly or educational purposes’.²

Data on the overall impact of these policies is not readily available; many institutions are still in the process of implementing these. Harvard’s FAS and MIT report that only 5 per cent and 4 per cent respectively of the articles have had an opt-out waiver issued (ibid.: 92). According to a press release from the MIT News Office, roughly 33 per cent of the approximately 7000 articles published by MIT faculty since 2009 have been made openly available in DSpace@MIT.³ Publishers have reacted in a variety of ways; MIT publicly tracks publisher co-operation with their OA policy and most publishers listed do co-operate, with the exception of some of the largest publishers such as Elsevier, Wiley and the American Chemical Society.⁴ Elsevier explicitly requires authors to opt out of any mandatory policies, despite allowing the deposit of the Accepted Version. Elsevier’s copyright policy reads that:

The reasoning behind this policy is that institutional OA policies represent the systematic distribution of the research, while voluntary deposit does not (one would guess because voluntary deposit is largely so uneven).

Despite the deliberate lack of co-operation by some publishers, many OA advocates have seized on the development of institutional OA policies as key to the future for ‘green’ OA. However, the rate of adoption does appear to be slowing. According to ROARMAP, since a peak in 2009 when more than 50 institutional policies were passed, in 2010 the number had dropped to approximately 30; 2011 and 2012 both saw fewer – between 20 and 30. This may be due to the difficulty in getting such policies passed through faculty senates, to interest in understanding the impact of such policies better, or because more attention has been focused on ‘gold’ OA.

An arguably more powerful driver of ‘green’ OA is funders. Funders have shown interest in OA in order to increase the impact of the funded research and, in the case of government-sponsored research, in order to allow the public access to research funded through tax monies. In late 2005, the Wellcome Trust, the largest private funder of medical research in the UK, required that its grantees make published research openly available through PubMed Central. Since then 56 funders worldwide, both public and private, have instituted similar requirements for their grantees (see ROARMAP http://roarmap.eprints.org/). Perhaps the largest of these has been the US National Institutes of Health (NIH) which instituted a public access policy in 2008 that requires published articles to be made openly available in PubMed Central a year after publication.

It is these funder mandates, particularly from government entities, that have attracted the most attention from publishers, researchers, politicians and the public. In the US, legislation that would expand the NIH policy to all federal agencies expending funds above a certain threshold (the Federal Research Public Access Act, or FRPAA) has been introduced several times in the House and the Senate since 2006, most recently in February 2012 (H.R. 4004 and S. 2096), but it has not yet seen success. New legislation that differs slightly from FRPAA was introduced into both the House and Senate in February 2013; in addition to open access to federally funded research, the Fair Access to Science and Technology Research Act (FASTR) (H.R. 708 and S. 350) would require that the research be in a form that is easily accessible for computational analysis, and that agencies consider requiring licensing terms that would allow reuse of research with attribution.

The FASTR Act comes on the heels of a very active year in terms of OA initiatives. In November 2011, the White House Office of Science and Technology Policy issued two public requests for information (RFI): one on Public Access to Peer-Reviewed Scholarly Publications Resulting From Federally Funded Research and the other on Public Access to Digital Data Resulting From Federally Funded Scientific Research.⁶ Three hundred and seventy-eight comments were submitted for the RFI on public access to publications, and 118 on the RFI on public access to data.⁷ In December 2011, the Research Works Act (H.R. 3699), legislation that would roll back the NIH policy and prohibit federal agencies from enacting such policies, was introduced in the US House by Representatives Darrell Issa (R-CA) and Carolyn Maloney (D-NY), and it had the support of the American Association of Publishers (AAP). This legislation met with a storm of protest from researchers, libraries, societies (including the Modern Language Association and the American Physical Society) and some publishers (including MIT Press, Nature Publishing Group and the American Association for the Advancement of Science [AAAS]) (Howard, 2012a). Opponents noted that both sponsors had received campaign contributions from Elsevier (Taylor, 2012). A website, The Cost of Knowledge (http://thecostofknowledge.com/), was established by Sir Timothy Gowers from Cambridge University in late January 2012 to protest against Elsevier’s business practices, against high prices for journals, and to provide support for acts such as the Research Works Act (as well as the Stop Online Piracy Act [SOPA] and the PROTECT IP Act [PIPA]); on this site, academics could pledge to not publish, edit or peer-review journals published by Elsevier. This garnered immediate support from academics, with over 7000 signatories by late February (Howard, 2012b). At this point, Elsevier explicitly withdrew its support for the Act, and the two sponsors stated that they would not pursue it. In May 2012, a petition to the White House to require free access over the Internet to scientific journal articles arising from taxpayer-funded research was begun;⁸ by February 2013 it had reached over 65,700 signatures.On 22 February 2013, the Office of Science and Technology Policy of the White House issued a memorandum directing federal agencies who expend over US$100 million in research and development to develop plans to implement OA policies for peer-reviewed journal articles, as well as management and sharing policies for research data.⁹ As of February 2014, these plans have not been released to the public.

So where do repositories fit into all of this? According to an analysis of OA policies from 48 funders conducted by SPARC Europe, in most cases funders mandate that researchers make their research openly available via repositories rather than the ‘gold’ route – publishing in an OA journal or utilizing the so-called hybrid model of publishing an OA article within a subscription-based journal.¹⁰ Of all funders, it is only Research Councils UK (RCUK), under the recommendations of the Finch Report, that would require OA via the ‘gold’ route and that would not consider direct deposit in an OA repository sufficient (although a journal that deposits research into an OA repository would meet the requirements). Funding for article processing charges (APCs), utilized by many OA journals and hybrid models, would come out of the RCUK budget, and there is an explicit expectation that institutions of higher education and research would set aside funds for APCs.¹¹ The dismay of many OA advocates at the Finch Report and the RCUK policy was based primarily on three factors: 1) they dismiss the infrastructure already in place for OA, a well-established network of institutional repositories, built over the last ten years; 2) they focus too much on assuaging publisher fears of OA via the ‘green’ model, and, thus, present a much more expensive model for OA; and 3) they are out of step with the majority of funder policies worldwide (SPARC Europe, 2012). The fears mentioned in the second point essentially centre on whether subscription rates to journals would fall (via libraries cancelling) by providing open access to published research via repositories (even if just the Accepted Version), forcing publishers to shut down journals or shift business models. There is little evidence to date either to prove or disprove this fear, although often the example of physics and the arXiv repository is used (as I did in the previous edition of this book) to illustrate that active ‘green’ OA can exist alongside subscription-based journals (Shreeves, 2009). While it is unclear whether the physics example can be generalized, the publishing industry itself provided some evidence that it is unlikely that libraries will cancel journals because of the availability of articles via OA repositories. In 2006, a survey commissioned by the Association of Learned and Professional Society Publishers (ALPSP) found that librarians were more likely to consider faculty needs, usage and price rather than OA when considering journal cancellations (Ware, 2006). With the introduction of the FASTR Act in the US which favours the use of repositories as well as the White House memorandum, this debate on how OA should be enacted is likely to continue for some time.

Further impact of repositories

Beyond repositories and OA, two additional areas should be mentioned when considering the impact of repositories on the future of academic journals. The first is that repositories have enabled the growth of infrastructure and expertise for publishing within libraries and other institutions. The second is the role that repositories are beginning to play vis-à-vis research data.