Introduction

The relationship between research funding organizations and scholarly publishing seems to have entered a new, more active phase of engagement in the age of OA. Researchers’ ability to choose where to publish their results has long been taken to be a matter of fundamental academic freedom (American Association of University Professors [AAUP], 1940;United Nations Education, Science and Cultural Organization [UNESCO], 1997). Funders in the second half of the 20th century certainly required acknowledgment of their funding in publications, disseminated commissioned studies via publication offices, and sometimes supported the payment of “colour charges” (Hall & Bembridge, 1986, p. 273) and “page charges” (National Science Foundation, 1977). Moreover, by setting expectations for publication records, they did indirectly influence publications patterns. However, they seem largely to have avoided policy prescriptions regarding where or how their fundees should publish, and to have avoided direct intervention in the manner of research institutions and researchers’ membership organizations such as scholarly societies and national academies, which often directly operated publication initiatives (e.g., journals, serials, and presses).

This has changed. Since the rise of the OA agenda at the end of the last century, given urgent and compelling voice in the 2002 declaration of the Budapest Open Access Initiative (Chan et al., 2002), funders have taken an increasingly active interest in matters of publication. TheWellcome Trust (2018) states that “Transforming human health will take longer if research outputs—like publications, data, software and biological materials such as cell lines—aren’t managed, shared and used in ways that unleash their full value.” The Gates Foundation sees OA publishing of peer-reviewed research as holding the potential for researchers from diverse backgrounds to come together and accelerate the research process, in turn leading to new ways of making people’s lives longer, healthier, and more productive (Morgan, 2017). As a final example of funder motivations to actively support OA publishing, the Open Research Funders Group (ORFG;2018) has committed itself to the open sharing of research outputs, because it believes this will “benefit society by accelerating the pace of discovery, reducing information-sharing gaps, encouraging innovation, and promoting reproducibility” (ORFG, 2018). OA to publications means that research publications can be accessed online, free of charge by any user, with no technical obstacles. At the minimum, such publications can be read online, downloaded, and printed (ideally other rights to copy, distribute, remix, and mine would also be granted). Access can be either through author archiving in online repositories (“green OA”) or by publishing in OA journals (either full OA or hybrid) or other publication outlets (“gold OA”).

From the early 1990s, several initiatives have sought to harness the power of emergent digital networked technologies to provide access to research outputs. Often these have been driven by the research community, for example, the foundation of thearXiv.org preprint server in 1991 (Ginsparg, 2016). Several independent journals made their content freely available online, typically hosted by research institutes or departments. In the early 2000s, commercial (e.g., BioMed Central) and not-for-profit (e.g., Public Library of Science) publishers started to introduce and experiment with new OA business models, charging authors (rather than readers) for publication services. Observing these developments, and concerned both to increase access to their funded results and to find a solution to the spiraling costs of subscriptions in the early 2000s (the so-calledserials crisis) (Kiley & Terry, 2006), funders worldwide began to implement measures to support a transition toward OA.

To this end, since the early to mid-2000s, major funders have increasingly introduced policies or mandates to encourage or prescribe OA for publications deriving from their research funding (Vincent-Lamarre, Boivin, Gargouri, Larivière, & Harnad, 2016;Xia et al., 2012). For example, each of the 30 Science Europe member institutions now either have OA policies or are in the process of implementing one (Crowfoot, 2017). Perhaps mindful of the fact that such measures can be argued to infringe upon the academic freedom of researchers to choose where to publish (Johnston, 2017), funders remain keen to emphasize choice. Hence, funder OA policies, at least in the Global North, follow a broadly similar approach: they allow a mixture of green and gold OA options, fund article-processing charges, and impose restrictions on the maximum length of embargo periods (the publisher-prescribed length of time from publication until author-archived versions can be made openly available) for green OA. However, the nuances of these policies are often complex, with different legal, financial, disciplinary, and cultural contexts affecting factors like the extent to which gold or green is preferred and levels of funding for Article Processing Charges (APCs) (Science Europe, 2016). In Europe, for example, although many countries favor green OA, or a balanced approach, there is a preference for gold OA in the United Kingdom, the Netherlands, and Austria.

OA to publications is now a mainstream policy among major research funding organizations. Funders such as the EC have recently targeted that all European research articles should be available via OA from 2020 onward (Enserink, 2016;EC, 2017,2018b).

But this commitment brings an increasing need for funders to engage with the economics and politics of the provision of awareness-raising and support measures, publication funds, and repository infrastructures.

The barriers to OA are diverse, but top-line factors include lack of funding for APC gold publications, perceptions of lower quality of OA journals, and the complexities of embargo and licensing policies (Dallmeier-Tiessen et al., 2011). Other potential barriers include insufficient training, copyright/licensing challenges related to third-party content, and lack of incentives within organizations and research communities to move away from publication in traditional, restricted access journals.

Availability of financial support for APCs is hence a major driver for OA in many countries (Dallmeier-Tiessen et al., 2011;Jubb, Goldstein, Amin, & Pinfield, 2015). However, as shown by a recent survey among former grantees in the context of the EC’s FP7 post-grant OA pilot, many report difficulties in accessing funds for OA publication charges. On average, less than a fifth reported having access to an institutional publication fund (out of about 300 responses), while this share was particularly low for respondents from Eastern Europe (0%) and Northern Europe (5%). More common was that respondents used or had access to research grants (about 50%), personal funds (about 45%), and/or institutional or departmental funds (less than 30%) (Johnson, Fosci, Chiarelli, Pinfield, & Jubb, 2017).

Hence, many funders support the costs of APCs, either by making them eligible grant costs or by making available earmarked funds. This constitutes a considerable new financial burden for funders, who obviously have an interest in keeping costs down. However, controlling costs can be in conflict with the aim of increasing uptake of OA. The APC market is still emergent, with unresolved questions about what costs are reasonable, most obviously with regard to so-called “hybrid” OA, where the market has been branded “dysfunctional” (Björk & Solomon, 2014). This is exacerbated by a lack of transparency on the actual costs of publishing, and a perceived “price of prestige”—where APCs in more prestigious journals tend to be higher for similar levels of service in cheaper, less prestigious venues (Van Noorden, 2013).

Data from the Open APC initiative from 2005 to 2018 showed that across all 158 participating research performing institutions and research funders (mainly from Germany, the United Kingdom, Norway, Sweden, and Austria), the average APC for fully OA journals was €1,481 (Mdn = €1,407), but substantially higher for hybrid journals (average €2,490,Mdn = €2,443) (data as of May 6, 2018) (“OpenAPC Dataset,”n.d.). There is hence concern that hybrid APCs often reflect traditional publishers’ concern to maintain existing profit margins and market position rather than the true costs of publishing (Laakso & Björk, 2016). Currently a large share of APC expenditure goes to hybrid OA. For example, over the period 2013-2016, the Wellcome Trust spent around just a fifth (about €1.8 million) of its total APC expenditure on articles in fully OA journals (1,015 articles,M costs = €1,756,SD = €819,Mdn = €1,604) and over €7.1 million on articles in hybrid journals (2,767 articles,M costs = €2,572,SD = €893,Mdn = €2,565) (data as of May 6, 2018) (“OpenAPC Dataset,”n.d.). Exacerbating this, publishers have been accused of “double dipping” through hybrid OA (Björk & Solomon, 2014), gaining extra income by charging APCs and subscription fees for the same content. Given this situation, it has been plausibly claimed that subscription journals lack incentives to move toward OA (Johnson et al., 2017).

Some funders have reacted by capping the levels of APCs they will pay or refusing to pay for hybrid publications (De Castro, 2015). In other cases, costs for (hybrid) OA are included in big deal negotiations, for example, in the Netherlands (Heijne & van Wezenbeek, 2018). However, non-disclosure clauses often make it impossible to assess the true financial implications of such agreements. A study from 2013 targeted 10 biomedical research funders and investigated their approaches to the implementation of OA policies and related cost issues. Several of funders expressed worries about escalating costs as gold OA becomes more mainstream. In this context, they hoped or expected that OA would increasingly play a role in researchers’ decision-making processes about where to publish. Interviewees pointed out that researchers might currently be too insulated from the costs of publishing, and that an increased author awareness of costs would be a desirable outcome of the move toward OA. In addition, one interviewee believed that costs may play out as a factor when choosing between less prestigious journals (Collins, 2013).

To date, these decision-making processes have not been studied in detail, and it must be noted that OA continues to play a secondary role when it comes to the selection of where to publish. To a certain extent, it can be expected that awareness of OA publication costs is higher in projects where researchers have to cover these costs out of their own project budgets to secure compliance with a funder mandate. In turn, researchers will be less aware if these costs are directly covered by funders or institutions, or if deals with publishers are in place.

A number of efforts have been made to research the effect of “flipping” non-OA journals to OA (Solomon, Laakso, & Björk, 2016). There have been a few research institution–led initiatives to convert journals to OA at no cost to the author. A discipline specific initiative is SCOAP3 (Sponsoring Consortium for Open Access Publishing in Particle Physics), which involves redirecting subscription fees and instead paying for OA from a central fund (Romeu et al., 2014). At a much larger scale, the OA2020 initiative has been launched, led by the Max Planck Society based on a 2015 white paper (Schimmer, Geschuhn, & Vogler, 2015). It has many European national funders committed to a model of redirecting existing subscription fees into OA funds, at a large cross-disciplinary scale, with the aim of disrupting the existing subscription system. OA2020 has, however, been criticized for seeking to reproduce the current dependency on a very few large commercial publishers who have proven themselves to be expensive and resistant to change. Confederation of Open Access Repositories (COAR) and UNESCO (2016), in a joint statement, indirectly referred to the OA2020 initiative and pointed out that a number of issues need to be addressed during the large-scale transition to OA; in particular, such a system needs to provide support researchers from institutions with smaller budgets or developing countries may not be able to meet the fees; further concentration of the publishing market needs to be avoided and mechanisms should be developed to ensure cost reductions (Fecher, Friesike, Peters, & Wagner, 2017).

Among researchers, positive sentiments toward OA have yet to fully reflect publishing choices. Researchers are very aware of OA, and the vast majority believe it beneficial (Ruiz-Perez, 2017). However, this does not seem to translate into practice. Dallmeier-Tiessen et al.’s study from 2011 for the SOAP project (http://project-soap.eu/) found that although almost 90% of respondents reported positive attitudes toward OA, only 52% had actually published via that route (Dallmeier-Tiessen et al., 2011). The lesson here is that researchers value OA in the abstract but are more reticent to put it into practice. This can be attributed to a continuing lack of structural incentives to choose OA, especially in institutional promotion and tenure procedures (Xia, 2010), as well as lack of awareness about green OA “self-archiving” options, recurrent skepticism about the quality of OA journals, difficulties in accessing funds for OA publications, and general unease with novel workflows (Dallmeier-Tiessen et al., 2011). Hence, despite OA mandates, progress in OA transition to date has been relatively slow. A very recent study, for instance, estimates 28% of the scholarly literature to be OA (either green or gold) as of 2017 (Piwowar et al., 2018). Other studies reach different conclusions, depending on methodology and OA definition, but reflect the general conclusion of relatively low uptake across the piece. Jubb calculated that 16.6% of all articles are published in gold OA (Jubb et al., 2015). An Organisation for Economic Co-Operation and Development (OECD) report byBoselli and Galindo-Rueda (2016) estimated, meanwhile, that around 30% of publications are OA, with around 20% of closed articles later made available via green OA. In total, Boselli and Galindo-Rueda believe “approximately 50-55% of documents are openly available 3-4 years after publication.” A report by Science Metrix (Archambault et al., 2014) for the EC found that as of April 2014, more than 50% of the scientific papers published between 2007 and 2012 could be downloaded for free on the Internet, although this included a large proportion of articles which do not meet the definition of OA (being of dubious copyright or hosted on proprietary platforms like ResearchGate). What is more, growth in the OA market seems to be slowing, or at least no longer accelerating (Johnson et al., 2017).

Given these conditions, it is clear that achieving the transition to OA within a reasonable time period requires continued intervention from stakeholders interested in achieving that goal, including research funding organizations. Yet funders are increasingly aware that their interventions can influence market development in unexpected and potentially undesirable ways (Björk & Solomon, 2014). For example, current evidence suggests that generous funding for hybrid publications may lead to a steep increase in OA publications in the short-term but at the expense of a long-term increase in the level of average APCs (Johnson et al., 2017). In the United Kingdom, for instance, the 2013 gold OA-focused RCUK Open Access Policy and its provisions for APC Block Grants resulted in a large increase in hybrid expenditure. The result was that by 2015, U.K. institutions’ “use of OA in hybrid journals and of delayed OA journals is more than twice the world average in both cases, while its take-up of fully OA journals with no APC (Gold-no APC) is less than half the world average and falling” (Jubb et al., 2015).

In parallel to these direct investments in OA publishing, funders have a longer history of supporting publishing infrastructures and other supporting services to foster OA:

• Publishing services: The Public Knowledge Project (PKP), which develops and maintains the open source Open Journal Systems (OJS) is financially supported by the Canadian Foundation for Innovation, Canadian Internet Registration Authority, the Laura and John Arnold Foundation and the MacArthur Foundation. Another example of funders supporting publishing services is Collaborative Knowledge (Coko) Foundation, which is supported by Laura and John Arnold Foundation, the Gordon and Betty Moore Foundation, and the Shuttleworth Foundation.

• Preprint servers: Preprints are complete drafts of scientific documents, not yet peer-reviewed, that are made available online, often via dedicated repositories known as “pre-print servers” (Bourne, Polka, Vale, & Kiley, 2017). ArXiv, established in 1991, is by far the most used preprint server (for physics, mathematics, computer science, quantitative biology, quantitative finance, and statistics). Further preprint servers were established a few years later—for example, RePEc (Research Papers in Economics; which indexes several digital archives) and SSRN (the Social Sciences Research Network; which was acquired by Elsevier in 2016). Spurred by the creation of bioRxiv by Cold Spring Harbor Press in 2013, and the advocacy efforts of ASAPbio, a scientist-driven initiative to promote the use of preprints in the life sciences, interest in preprints has grown sharply in recent years (Luther, 2017). A host of new preprint servers have since begun to appear, including, but not limited to, many hosted by the Center for Open Science: SocArXiv (social sciences, since 2016), PsyArXiv (psychology, since 2016), PaleorXiv (paleontology, since 2017), EarthArXiv (geosciences, since 2017), and LawArxiv (law, since 2017). SciELO, the Scientific Electronic Library Online, which provides OA to more than 1,200 journals from Latin America, Spain, Portugal, and South Africa, has also announced that they will launch a preprint service in 2018 (Packer, Santos, & Meneghini, 2017). Funders have played a role in fostering these developments. arXiv lists among its supporting members the European Research Council, the Austrian Science Foundation (FWF), and the Simons Foundation (arXiv 2018); BioRxiv receives support from the Chan Zuckerberg Initiative (CZI) (Callaway, 2017); and the group of pre-print servers hosted on the Open Science Framework are supported by the Center for Open Science, in turn funded by the Arnold Foundation (Center for Open Science, 2018).

• Repositories: In 2000, the National Institutes of Health (NIH), through the National Library of Medicine (NLM), launched PubMed Central (PMC) as full-text journal article repository. From 2005 onward, it has become the designated repository for research articles in biomedical and life sciences funded by a number of U.S. government funders. In Europe, Wellcome Trust together with 27 other research funders supports EuropePMC, where research articles resulting from their funding are deposited in parallel to PMC (The Europe PMC Consortium, 2015).

• Repository aggregators and abstracting/indexing services: Institutional repositories receive coordinational support via regional efforts like OpenAIRE (funded by the EC), SHARE (funded, in part, by the U.S. Institute of Museum and Library Services [IMLS] and the Alfred P. Sloan Foundation), and LA Referencia (funded by Latin American public science and technology agencies). Other services enable discovery of OA outputs by collecting, organizing, and systematizing information regarding OA publications from diverse platforms. Example services and activities include, for example, the OAPEN Library of OA books which provides a deposit service to the Wellcome Trust, the Austrian Science Foundation, and Knowledge Unlatched. In addition, OAPEN (Open Access Publishing in European Networks) is conducting projects with the Swiss National Science Foundation and the European Research Council (OAPEN, 2018).

• Other enabling services: In addition, funders have supported a range of awareness-raising and capacity-building activities by providing information on OA at various levels, from the general (what OA is, its aims and objectives) to the specific (e.g., individual journal OA policies, registries of entities). The former can be exemplified by OpenAIRE’s network of 33 National Open Access Desks and the FOSTER Open Science training initiative, while examples of the latter include the SHERPA services RoMEO (journal policies) and JULIET (funder OA policies), as well as OpenDOAR (OA repositories)—services supported via U.K. infrastructural funder JISC. Research funders have also supported several studies which investigated the relationship between OA policies and services, as well as the development of strategies for sustaining core services (Johnson & Fosci, 2015).

Funder OA Platforms

The Wellcome Trust, one of the world’s largest biomedical charitable foundations, in July 2016 announced its plan to launch an OA publishing platform to be titled Wellcome Open Research (henceforth WOR) (Butler, 2016). The announcement specified that management of the platform would be contracted to the OA publishing platform F1000Research and follow that platform’s publishing model. In the F1000Research model, following only an initial light “sanity check” by a professional editor, research outputs are immediately published and then openly peer-reviewed, with review reports and reviewer names published alongside in real time (F1000, 2018).

Wellcome has traditionally been at the forefront of debates about OA and data sharing. It has supported APC payments since 2003 and, in 2006, introduced a strict OA mandate that all publications must be made available within 6 months of publication via PMC (Kiley & Terry, 2006;Walport & Kiley, 2006). In 2012, together with the Howard Hughes Medical Institute and the Max Planck Society, Wellcome launched eLife, a peer-reviewed OA journal for biomedical and life sciences that aimed to compete with the most prestigious journals likeNature, Cell, andScience (Schekman, Patterson, Watt, & Weigel, 2012). In so doing, Wellcome took a step beyond merely supporting OA to take a direct interest in publishing. eLife remained editorially independent from its funders, however, committed to publishing all research based on merit regardless of funding organization. In 2017, Wellcome Trust was even revealed to have been among a group of investors who invested 52.6 million in ResearchGate, the academic social network (ResearchGate, 2017).

The 2016 announcement of WOR, however, was a step-change in engagement in publishing. It was welcomed as such by OA advocates like arXiv founder Paul Ginsparg, who said, “This really is a potential game changer for a major funder to be taking control of the research output” (Bohannon, 2016). Robert Kiley explained Wellcome’s motivation for the platform as stemming from a wish to increase speed, transparency, and reproducibility in scholarly communications, by offering a venue with no author-facing charges and relative cost-effectiveness for the funder, that would allow its researchers to publish all their research outputs (from articles and datasets to case reports and protocols, to null and negative results). All Wellcome researchers would be able to use the platform but could still publish wherever else they wished. The stated APCs charged per article were not significantly different than those charged by the F1000Research platform. The platform was opened for submissions in October 2016 and the first group of articles were published a month later. The next section gives an analysis of the outputs from WOR’s first year.

In this section, we report some findings based on the publication metadata and related events on the WOR publication platform, and consider all 192 publications (all versions included) which have been submitted between October 17, 2016, and November 17, 2017. A more detailed version of this analysis is available online (Schmidt, 2018a,2018b).

Over this period of 13 months, the submission rate to WOR was rather modest, with about 15 papers per month, and no acceleration of submissions could be observed (seeFigure 1).

Download

Open in new tab

Download in PowerPoint

Figure 1. Submissions to Wellcome Open Research by month (November 17, 2016-November 17, 2017).

Several article types can be published on WOR. So far, about three out of five of all articles are research articles (88 articles, 62%), followed by method articles (13 articles, 9%), study protocols (10 articles, 7%), and several smaller categories.

Of the 142 papers published on WOR by end of November 2017, 95 papers have only one version, 47 have two versions, and three papers have three versions. The rate of papers with only one version seems to be rather high. This might be partially due to the fact that for some papers, the review-revise process has not been closed yet.

Overall, 1,110 authors have been involved in the writing of 142 publications. In addition, seven consortia contributed to the writing of seven papers. On average, about eight authors were involved in each paper (M = 7.9,SD = 5.5, minimum = 1, maximum = 31).

Regarding duration between publication events, there was some variation depending on publication type (seeFigure 2). For research articles, the first review was typically received within about 43 days, and the second review within another 12 days. Indexing in PubMed and other bibliographic databases was accomplished by Day 65. The time until receiving the first review was somewhat longer for study protocols (Mdn = 57 days) and shortest for open letters (22 days) and data notes (28 days).

Download

Open in new tab

Download in PowerPoint

Figure 2. Time from submission to first review by article type.

As the information in the WOR dataset was incomplete regarding the review outcomes (variable “review status” with the following possible values: accepted, accepted with reservations, rejected), we only considered those articles who were already indexed by Europe PMC. It must be noted that publications are indexed only after they have “passed” peer review. A paper is considered to have passed peer review if it has received at least two approved referee reviews, or one approved plus two approved with reservations reviews (WOR, 2018). In consequence, the review ratings for papers on Europe PMC will naturally be somewhat skewed toward more positive reviews.

For WOR articles, review status information was parsed from the Europe PMC website. Information on 354 review reports was retrieved for all 111 WOR articles which are available on Europe PMC. In addition, there were 100 author responses. The distribution is strongly skewed toward positive review ratings: over three fourth approved (267 reviews, 75.4%), nearly one fourth approved with reservations (84 reviews, 23.7%), and less than 1% (three reviews, 0.9%) rejected the article under review.

This result is in line with the review ratings on the parent platform F1000Research. Based on a retrieval of all F1000Research research articles that have been indexed in Europe PMC, we consider 3,880 records of review reports which are related to 1,200 records of research articles. The distribution is very similar to the above: About three out of four reports approved (2,913 reviews, 75.1%), nearly one out of four reports approved with reservations (901 reviews, 23.2%), and only 1.7% (66 reviews) rejected the research article under review. Research articles have received between two and eight reviews, on average three reviews.

From this analysis, although no prior indication was given by Wellcome as to what would constitute success, it must be admitted that WOR cannot yet be regarded a full success. Operationally the processing of submitted papers seems to work well, but the overall uptake can be argued to be low compared with the investment made by the Wellcome Trust (although the total cost of the platform is unknown). The 142 publications on WOR amount to a share of about 2% of all WOR publications (estimate based on average number of publications indexed by Europe PMC in 2013-2016; overall over 27,000 publications). Kiley points out that WOR has been the fifth most popular publication venue during this first year of operation, afterScientific Reports, PLoS ONE, Nature Communications, andeLife (Kiley, 2017). It should also be noted that while information on APCs is publicly available, further information about cost breakdown of running WOR, also in comparison to F1000Research, is not publicly available, thus preventing further assessment of WOR from the point of view of cost-effectiveness.

The fact that the rejection rate on F1000Research is very low has been strongly criticized byVines (2013) for the very high rate of positive reviews (“approved” and “approved with reservations”), in comparison with a sample of papers from medical journals for which the average length of reviews was substantially longer (464 vs. 254 words) and only 42% were positive. Vines goes even so far to completely dismiss the reviews, that is, readers are advised to consider papers on F1000Research as if they have never been through peer review. Although this view seems somewhat exaggerated, it seems reasonable that in the case of positive review ratings, the motivation for authors to revise a paper may be lower. In addition, the label “not approved” is not to be confused with “rejected” (see WOR FAQs). The notion that journals advertise high rejection rates as a measure of prestige has been criticized by several authors, not the least because the most cited journal do not necessarily have the highest rejection rates (Schultz, 2010), and low rejection rates can actually be interpreted as a sign of self-regulation and high efficiency (Pöschl, 2012). Perhaps most importantly, when peer review is focused on assessing methodological quality rather than perceived importance of the reported research, rejection rates are expected to be lower as no artificial scarcity is created by selectivity.

Inspired by the Wellcome example, in March 2017, the Bill and Melinda Gates Foundation, another major philanthropic funder of biomedical research, announced it would also be launching a platform based on the F1000 platform (Butler, 2017). The first Gates Open Research articles were published in November 2017. As of March 1, 2018, a DOI was available for a subset of 25 records. Since then, the number of publications has doubled: according to Crossref, there were 53 articles with registered DOIs on the Gates Open Research platform as of May 10, 2018. Regarding submitted article types, about three fifth were research articles, followed by about one fifth study protocols and data notes, open letters, method articles, and systematic reviews ranging between 4% and 7%.

The time from submission to publication across all publication types was about 19 days (Mdn), ranging from 10 days for method articles and 52 days for data notes. The first review typically arrived after 31.5 days, again taking shortest for method articles (21 days) and longest for data notes (65 days). The second review was available after another 9 days. Publications were indexed after about 41 days. Overall, these durations were slightly shorter than for submissions to the WOR platform. However, it must be noted that the dataset only records the first 4 months of operation of the platform, and thus these findings are only indicative.

An increasingly long list of other funders, research organizations, and institutions have since followed the example of Wellcome and Gates, with F1000-powered publishing platforms announced by the HRB Ireland, the African Academy of Sciences, UCL Great Ormond Street Institute of Child Health, and the Montréal Neurological Institute and Hospital. These platforms remain in various states of development at the time of writing.

With the success of this model, in July 2017, F1000 announced Open Research Central, a centralized portal through which researchers will be able to submit work to any of these F1000-powered open research publishing platforms. This had been signaled in advance by Kiley on WOR’s announcement a year earlier, telling Nature “the expectation is that this, and other similar funder platforms that are expected to emerge, will ultimately combine into one central platform” (Butler, 2016). Of note here, however, is F1000’s stated intention to eventually transfer governance of this portal to the community:

The case of the HRB Ireland gives us some indication of the behind-the-scenes workings of these deals, as it is F1000’s first agreement with a public funder. The public tender report (Office of Government Procurement Ireland, 2017) advises that the sum of €400,000 had been made available to “establish a single operator framework for the provision of an Open Research Publishing Service” for a total of 4 years. The tender seemed implicitly targeted toward a narrow range of possible providers by stipulating that the “platform should provide users with immediate publication followed by invited, transparent, post-publication peer review.” Only one tender application was received and the contract was granted to F1000.

The EC in mid-2017 announced its intention to also provide such a platform for researchers funded via its framework program Horizon 2020 (Enserink, 2017). More details were given in an Information Note published in December 2017. That note made explicit that the Commission was following the example of Wellcome and Gates to raise the level of OA publications stemming from their funded research in a cost-effective manner. The note is also careful to emphasize the voluntary nature of the platform, which would be free to use for Horizon 2020 grantees. It foresaw the benefits of raising OA compliance rates, giving more flexibility to researchers, and demonstrating the EC’s position as a leader in Open Science implementation, as well as enabling competition though transparency regarding costs.

Horizon 2020 allocates almost €80 billion of funding over 7 years from 2014 to 2020 (EC, 2018c). As a public funder, the Commission faces different constraints and considerations than private funders, including more scrutiny and regulations. Also, the range of subjects covered by its funding is much larger than the more targeted approach of the Wellcome Trust, the Gates Foundation, and HRB, which are explicitly addressed to health/life sciences. Hence, for the EC to enter this space will be a huge step in legitimizing such platforms. In all, €6.4 million will be allocated for a period of maximum 4 years for the EC platform—dwarfing the €400,000 allocated for the HRB platform for the same amount of time. The Open Research Europe tender was published by the EC on March 31, 2018 (EC, 2018a).

The platform is intended for Horizon 2020 beneficiaries to publish “scientific articles” in all major fields of scholarship, including Social Sciences and Humanities (SSH). The publication model specified diverged somewhat from the other funder platforms established until that time, in that it should offer two options: (a) a standard option in which manuscripts are peer reviewed before publication and (b) a model in which manuscripts are uploaded to a preprint server in advance of peer review. Peer review would in both cases be “open peer review,” although there were no exact specifications as to what aspects (Ross-Hellauer, 2017) of open peer review should be included, nor whether the publication of reviewer names or reports should happen after publication or in real time. Both preprints and peer-reviewed articles should be licensed either Creative Commons CC0 or CC-BY “or equivalent,” and text- and data-mining should be offered “in accordance with existing practices as they evolve over time.”

The contract notice explicitly stated that the EC is looking for customization of an existing publishing solution. The tender specifications hence included a number of criteria which seemed designed to ensure that only very established providers could tender, including needing to guarantee uptime of greater than 99.999%, having experience in IT publishing infrastructure in at least three European Union (EU) countries, and having an annual turnover of more than €1 million for the last 2 financial years. Such strict terms caught criticism from innovative noncommercial providers, such as Martin Eve of Open Library of Humanities (Eve, 2018) and Jean-Sébastien Caux from SciPost (Caux, 2018). The concerns of both were that these stringent conditions would prevent an innovative and truly open but budget-wise small solution from competing for the platform.

The platform architecture was not required to be open source, but there was a stipulation that it should be portable (not: forkable), and planned hand-over to the Commission (or party designated by the Commission) at the end of a 4-year period should be made possible. As part of this handover, the contractor would need to provide whatever is necessary for the Commission or a third designated party to run and maintain the entire platform infrastructure and, if necessary, redeploy it in a new environment. This would imply the transfer of both the content of the system and the workflows.

Processes, policies, and operational costs (including price-per-article) should be fully transparent to the public. The €6.4 million budget was broken down into €1 million for implementing and maintaining the platform infrastructure, communications, and sustainability (prepare for handover), with the remaining budget to be used for the production of peer reviewed articles, on a per-article cost basis (with preprints excluded from this budget calculation). The tender foresaw 5,600 peer-reviewed publications in 4 years, which would translate to an average publishing cost of €965 per article. A question mark should be raised about whether the platform will reach such levels of uptake, however. The projected 5,600 peer-reviewed publications in 4 years would represent 10% of projected number of Horizon 2020 publications. Given the Wellcome example, where the first year saw only 2% of Wellcome publications published via WOR, this could be a difficult target to achieve.

Finally, the tender contained stipulations on governance and sustainability. A scientific advisory board (whose role and mandate were not made explicit) should be selected by the contractor and approved by the Commission, while the contractor would also be responsible for developing a sustainability strategy to plan for operation of the platform beyond the initial 4 years, exploring potential synergies, business scenarios, funding models, and potential additional streams of revenue.

A Review of Roles and Motivations

Issues and Open Questions

Principles and Recommendations

Acknowledgements

Articles Citing this One: 0