Current State of : Briefing Paper for Universities Australia (UA) Deputy Vice-Chancellors Research Committee

The purpose of this paper is to describe the current state of open access (OA) publishing as it relates to Australia, to help inform a potential UA position on OA. This paper has been drafted by representatives from the Council of Australian University Librarians (CAUL) and the Australasian Open Access Strategy Group (AOASG).

Background In the fifteen years since the Budapest Open Access Declaration defined OA as free access wi th associated use and reuse rights1, OA has evolved from a fringe publishing movement with uncertain business models and piecemeal infrastructure to a global movement. Increasingly this movement aims to not just make research outputs (including publications, datasets and code) more available but to exploit new modes of scholarly output and technology2 to increase the use, reuse and interoperability of research. Ultimately, OA initiatives and the move to increased openness in government, education and research sectors have the potential to maximise the utility and value of research outputs for the benefit of academia, education and industry, and to improve the integrity and reproducibility of scholarly work.

A diversity of models aiming to implement the change to OA exist, ranging from those offered by the largest publishers, to small start-ups driven by researchers (see appendix A). Internationally, a number of approaches have been agreed to at a national or consortial level (see appendix B). Although increasing diversity is likely to be the future of publishing (and should be welcomed), this diversity of OA models, differing funder conditions and strategies for compliance, as well as varying publisher requirements make for a complex publishing landscape.

Although there has been considerable work in Australia on OA from funders, individual institutions, libraries and advocacy groups, the Australian higher education sector is yet to determine a preferred position in relation to recent developments in the industry, and agree on a strategy to optimise the discoverability and impact of Australia’s research. A clear position on preferred strategies for OA could simplify publication and archiving choices for researchers. A clear position would also assist universities and funders to make evidence-based decisions on how to invest in digital infrastructure and publishing initiatives that optimise discoverability and access, in order to drive research translation and innovation in the Australian digital economy.

1 http://www.budapestopenaccessinitiative.org/read 2 https://101innovations.wordpress.com/

The Scholarly Journal Publishing Paradigm The findings of government and university funded research are primarily disseminated through scholarly publications. Much of the authorship, reviewing and editing of these publications is provided by university researchers. Currently, the majority of these publications are owned by publishers, mostly commercial, which charge subscription fees so that universities can access the content generated by their researchers as well as publications from elsewhere. There are also an increasing number of OA publishers, some of which charge an Article Processing Charge (APC) for research to be made OA. APCs may be paid by authors, their funders or their institutions. Some publishers also offer hybrid subscription/OA models with an option to pay an APC to make a particular article open within an otherwise subscription-based journal.

The primacy of the current publication paradigm is embedded within international ranking and national assessment exercises which are of central importance to universities. For example, the Academic Ranking of World Universities (ARWU) rewards indexed publications and those in Nature and Science3. These ranking systems perpetuate some of the current challenges with publishing. Universities pay premium subscription prices for access to these high prestige journals, and/or pay APCs to enable OA to articles. Pressure to publish in journals such as Nature, or in the top 20% journals (by Impact Factor) is also incentivised by internal, institutional reward systems. In response, publishers have expanded their operations to produce metric and aggregation platforms which are used in the rankings systems. Universities also pay to access these services.

To put the scale of the scholarly publishing endeavour in perspective, in 2016 , the world’s largest scholarly publisher, reported revenue of £2.32 billion and a profit margin of 37% 4. It is of note that this profit margin has been relatively stable over many years, even as publishing has evolved. In Australia, there has simultaneously been a growth in the payment of APCs to make publications open without any corresponding reduction in subscription costs. The total cost of journal subscriptions for Australian university libraries was AU$261,131,466 in 2016, a 40% increase in comparison to 2009 and a 17% increase since 2014. 5 This increase is largely due to significant, higher than CPI subscription price increases each year, and is further impacted by a weaker AUD (most subscriptions are paid in USD, with an exchange rate of 0.77 in 2016, 0.81 in 2014 and 0.92 in 2009). Although the majority of contemporary research libraries’ journal holdings are now electronic only, many subscription rates are tied to libraries’ historic print journal holdings, often dating back to the mid-1990s. APCs to enable OA to electronic articles are paid in addition to these existing journal subscription costs.

Thus, despite advances in technology, the current system of scholarly communication is tied to ranking and assessment at an institutional and individual level, and continues to be tied to outdated hardcopy journal subscription deals. With the costs of accessing scholarly information continuing to escalate, funding agencies, universities and researchers are questioning whether the current approach is fit for purpose in the digital age and the current fiscal environment. The pricing and access/publication model is in nee d of reform.

3 http://www.shanghairanking.com/ARWU-Methodology-2017.html 4 https://www.thebookseller.com/news/elsevier-profits-3-despite-steeper-print-declines-493781 5 Statistics provided by the Council of Australian University Librarians 2

Current Australian Approach to Open Access There have been a number of key Australian developments relating to OA: ● ARC6 and NHMRC7 OA policies require funded research to be made openly accessible within a twelve month period from the date of publication. Currently, there is no systematic monitoring of compliance for ARC and NHMRC OA policies, or penalties for non-compliance. Some international funding bodies, such as Wellcome Trust, have sanctions for non-compliance (e.g. withholding a percentage of grant funding) 8. In terms of measurement, as part of ERA 2015, institutions were required to indicate whether a submitted research output had been made available in an OA repository with findings reported on 9. In the recent ARC Engagement and Impact Assessment Pilot, this data was utilised to provide a measure of engagement. ● In August 2017, the Australian government released its response to the Productivity Commission Inquiry into Intellectual Property Arrangements10. The Government supports the following recommendation: 16.1 The Australian, and State and Territory governments should implement an open access policy for publicly funded research. The policy should provide free and open access arrangements for all publications funded by governments, directly or through university funding, within 12 months of publication. The policy should minimise exemptions. ● In 2016, a working group was convened by Linda O’Brien, PVC Information Services, Griffith University (representing the CAUL Research Advisory Committee) with representation from UA, AOASG, Australian National Data Service, ARC, NHMRC, Department of Education and Training and the Department of Industry, Innovation and Science. As a result, the F.A.I.R. Policy Statement11 was established which has the overall aim of providing a pathway to improving the discoverability and impact of Australian research outputs and to improve community and industry access to research. The next steps are to establish a program to implement the F.A.I.R. Policy Statement. ● From 2007 to 2009 Australian universities were provided with ASHER funding to establish digital repositories for research outputs through an Australian Government investment of $25.5M under the Australian Scheme for Higher Education Repositories primarily to support the RQF and ERA 12. These repositories now contain over 1 million items, including OA versions of journal articles, theses and other university publications with more than half freely available for download (though most of these do not have associated OA licences, which specify their reuse rights, nor other consistent metadata). These items were used 36.5 million times in 201613. Many of these research outputs are Author Accepted (AAM) versions of published journal articles. Items in institutional repositories are discoverable via Google and NLA Trove. ● Currently, many Australian universities have a public OA policy or statement. The requirements and degree of compliance vary from institution to institution and university libraries and research offices have varying resource levels for advocacy, training and support and development for research management and repository workflow and platforms. Ten years on from the initial institutional repository impleme ntations, the question needs to be asked why Australia’s universities do not have a consistent OA policy position,

6 http://www.arc.gov.au/arc-open-access-policy 7 https://www.nhmrc.gov.au/grants-funding/policy/nhmrc-open-access-policy 8 https://wellcome.ac.uk/funding/managing-grant/complying-our-open-access-policy 9http://www.arc.gov.au/sites/default/files/filedepot/Public/ERA/ERA%202015%20Volume%202/ERA%20Volume%202%20Section%203.pdf 10https://www.industry.gov.au/innovation/Intellectual-Property/Documents/Government-Response-to-PC-Inquiry-into-IP.pdf 11 https://www.fair-access.net.au/fair-statement 12 https://industry.gov.au/science/ResearchInfrastructure/Pages/ASHERandIAP.aspx 13 Statistics provided by the Council of Australian University Librarians 3

despite the OA global movement and advocacy, and more recently, requirements and funding from the Australian government’s major research funding agencies.

A Green-Preferred Approach for Australian OA? The development of institutional repositories has enabled Australian universities to primarily take a Green OA approach to date (see appendix A). It is timely to review whether a green-preferred approach should continue. There are some challenges with this approach: ● Compliance with OA policies in repositories is complex due to the assignment of copyright by authors to publishers. Publishers enact embargoes and policies at journal title level which require significant work to interpret14. The implementation of a university-wide licence where universities retain the rights to include publications by their authors in repositories despite publisher embargoes, as implemented at Harvard University15, and currently being explored in the UK16 could increase compliance with OA policies. ● Lack of visibility of, and interoperability between, institutional repositories. ● Lack of coordination of Australian university and funder OA policies, repository or discovery infrastructure and standards. ● Lack of automated processes for researcher deposit means that Green OA can be resource intensive. Some current infrastructure is old and not user friendly and hence requires more staff input to run. ● Lack of incentives for researchers to comply with ARC/NHMRC policies.

The alternative to Green OA via repository infrastructure is through payment of an up-front APC to enable either a Gold (APC only) or Hybrid (APC + subscription) approach. It is not possible to determine the volume of APCs currently being paid by Australian universities. APCs are generally not paid centrally and are difficult to track at an institutional level, especially when works are co-authored across multiple institutions. Requests for this information from traditional publishers, on any systematic level even confidentially, were declined 17. As the total cost of OA publishing is obscured, universities are unable to make informed, strategic decisions about how best to invest in initiatives to enhance research dissemination and impact. What is becoming clearer, however, is the increasing overall cost of APCs. While there is little accurate data for Australian Universities some sample data is below.

2013 2014 2015 2016

Small AU university - $46,713.80 $64,222.90 -

Medium AU university - $168,766 $173,125 -

Large AU university - $460,321 $718,754 -

Go8 large university $874,948 $1,081,295 $1,581,757 $1,721,558

14 https://unlockingresearch.blog.lib.cam.ac.uk/?p=1657 15 https://osc.hul.harvard.edu/policies/ 16https://www.slideshare.net/TorstenReimer/the-uk-scholarly-communications-licence-a-model-for-open-access-rights-retention 17 V Barbour, personal communication, 2016 4

Go8 large university $457,096 $475,182 $750,350 $722,42618

As it stands, for ‘Hybrid’ journals, universities are paying an APC to make an article openly available, and also paying for a subscription for access to the journal. This ‘Hybrid’ model, and the complexity of publisher’s OA archiving terms, leads to the much maligned practice of “double-dipping” (universities paying both for the subscription to the journal and to make the article ‘open’). Of course, whatever model of publishing occurs, there are additional hidden costs of researchers who participate in the drafting of papers, in peer review and membership of editorial boards.

Internationally some countries are aiming to “flip” the subscription model to an APC-only based model. At this point in time, CAUL is cautious about this as a sole approach, based on reports 19 that more money is required in the system to flip to an APC-based model and that doing so may perpetuate high costs and the current publisher control of the system. Based on the need for more money in the system a green-preferred model is favoured by CAUL.

What’s At Stake? Doing nothing has considerable risks. Publishers are taking advantage of a lack of coordination among universities and national negotiating bodies. If the status quo continues there will be significant additional recurrent expenditure with Gold OA and hybrid models continuing to provide a self-perpetuating profit strategy for publishers. If the majority of research outputs (current international research estimates indicate approximately 72% of published research outputs20) continue to be closed access, research cannot be efficiently translated into policy, practice and industry, and the Australian government will struggle to realise its ambition to create a knowledge driven, innovative economy.

Non-traditional research outputs, in particular research datasets, software and code are of increasing importance to the creation of new knowledge, and data publication is of increasing interest to large scale scholarly publishers. If Australian universities lose control of the rights to reuse their data and code, in the same way they have (in many cases) to access and reuse publicly funded scholarly works, the transparency and reproducibility of scientific research will decrease, the costs of conducting and supporting original research will increase, and the creation of new information and knowledge will be impeded.

It is essential that Australia’s research remains visible and discoverable and that critical stakeholders, including users who benefit from access to funded research outputs, such as government, industry, and professional practitioners, are not locked out due to paywalls and restrictive licences for access and reuse.

18 Figures are a “best estimate”. (1) Most, but not all publishing costs captured by the university’s financial system are OA APCs. Data may include other publishing costs such as page fees, image fees or editorial support. (2) Dataset only captures transactions where publishing costs have been correctly recorded. In some examples, it is likely there are APCs that have been paid from other university accounts, and or the transactions have not been correctly recorded as direct publishing costs. 19https://www.library.ucla.edu/sites/default/files/Ginny-Steel_open-letter_OA2020-PIF_October-2016_0.pdf and http://bjoern.brembs.net/2016/04/how-gold-open-access-may-make-things-worse/ 20 Piwowar et al. (2017). The State of OA: A large-scale analysis of the prevalence and impact of Open Access articles (). From https://peerj.com/preprints/3119. 5

An Ideal Future State for Australia’s Research Outputs Rapidly emerging innovation in technology and new thinking about publishing allow one to look far beyond the current state. There is an opportunity to move towards a fully interconnected, global scholarly ecosystem supported by a wide variety of open publishing models, with sophisticated linking of research outputs that are well described by metadata such as DOIs, and which are linked to data, software and to related other outputs, as well as being linked to individuals and institutions through identifiers, such as ORCiDs 21. Such a model would include traditional outputs such as research articles and monographs, but would also allow for emerging innovations such as and related models.

Ideally Universities would retain the rights to manage and communicate publicly funded research outputs, including scholarly manuscripts, research data and code. There would be: ● OA to all publicly funded Australian research publications (journal articles and monographs). ● OA to publicly funded Australian research datasets (except where access is restricted by the requirements of legislation, policy, ethical, contractual, or confidentiality obligations). ● Streamlined and simple ‘green preferred’ OA policy setting, with fully supported workflows and repository infrastructure. ● Informed, transparent, evidence-based decision making on OA publishing support models, including university and research funding, subscriptions, offsetting and bundled models with universities having a full understanding of the total cost of publishing and disseminating research. ● Less reliance on Journal Impact Factors and journal names as a proxy for quality.

Below are some suggested CAUL, UA and AOASG actions to achieve an ideal future state: ● Investigate the best mechanisms for Australian universities to retain non-exclusive rights to author manuscript versions of scholarly works (eg. Australian version of UK-Scholarly Communications Licence). If the sector did adopt a version of the UK-SCL model, university policies and procedures would need to be developed at the local level with the support of the DVC Research. ● Investigate the true cost of publishing and disseminating research by implementing consistent processes to collect and report university expenditure on Article Processing Charges (APCs) to CAUL for monitoring. ● Provide information for universities to make proactive, informed decisions on when it is appropriate and in their interests to invest in gold OA models. ● Improve Australian repository infrastructure with an emphasis on making Australian research more visible and discoverable through methods such as a national repositories network, collection and interoperability or standards. Investigate international networked initiatives such as COAR, OpenAIRE, SHARE, La Referencia22. ● Investigate researcher incentives and recognition for open research in internal and external recruitment, promotion and award exercises and collaborate with international initiatives such as DORA23. ● Continue to monitor emerging publishing and OA models, including OA2020 (see Appendix B).Emerging and established “hybrid” OA models, including compacts (bundling publishing and reading fees) should continue to be monitored. ● Liaise and collaborate with relevant international initiatives.

21 https://aaf.edu.au/orcid/ 22 https://www.coar-repositories.org/ 23 http://www.ascb.org/dora/ 6

We are a point in time where there is an opportunity to develop one coherent national approach on OA to ensure that the next phase is well supported by building appropriate infrastructure and developing well aligned policies and principles.

Authored by the Council of Australian University Librarians OA Working Group Virginia Barbour (Director, Australasian Open Access Strategy Group) Jill Benn (University Librarian, University of Western Australia) Mal Booth (University Librarian, University of Technology Sydney) David Groenewegen (Director, Research, Monash University Library) Kate Kelly (Director, Library Services, Southern Cross University) Stephen McVey (University Librarian, University of Notre Dame) Kate Stanton (Senior Manager, Copyright & Information Policy, University of Sydney) Harry Rolf (CAUL Communication and Policy Officer)

Appendix A: Open Access Models There are a number of models of OA operating now, with more emerging regularly, which can lead to confusion among researchers who have to make publishing decisions. Campaigns to educate authors on their publishing choices include Think, Check, Submit24 as well as information from university libraries and AOASG25. An explanation of the OA Spectrum is provided by SPARC, PLOS & OASPA26. Peter Suber from Harvard University has also written a comprehensive overview of OA27.

Repository-mediated (Green) OA Authors publish their work in a peer-reviewed journal of their choice, and then deposit a version of the work in either an institutional repository, or a central repository such as PubMed Central28 which makes it freely available. The version deposited will depend on the agreement signed with the publisher. University-run repositories sometimes provide long term preservation. One of the major challenges (to Green OA) is that scholars still prefer the final published version because it has all the edits, proper type-setting (e.g. for maths equations) and in many cases images and graphics that are not included in the Green archived versions due to other copyright or IP challenges. Some say their research doesn’t make the same sense without these features that publishers provide. Some do not trust that a Green version is definitive and properly checked.

Publisher-mediated (Gold) OA The author publishes their work in a journal that allows free and immediate access to articles via the publisher's website with an appropriate license that specifies reuse. Gold OA publishers use various business models, including the hybrid model; an article processing charge (APC) may be required to publish an article as OA. Approximately 70% of OA journals listed in the Directory of Open Access Journals (DOAJ)29 don't charge an

24 http://thinkchecksubmit.org/ 25 https://aoasg.org.au/ 26 https://sparcopen.org/wp-content/uploads/2016/01/hoii_guide_rev4_web.pdf 27 http://legacy.earlham.edu/~peters/fos/overview.htm 28 https://www.ncbi.nlm.nih.gov/pmc/ 29 https://doaj.org/ 7

APC. The major challenge to Gold OA surrounds the increasing costs, the lack of transparency in APC charges and questions in relation to sustainability30.

“Black” or Pirate OA Researcher frustration with both Green and Gold OA and the lack of access to research behind expensive paywalls has led to the establishment of illegal file sharing services for scholarly literature. The largest such service is Sci-Hub, based in Russia. A recent article was published on the impact of this phenomenon31.

Appendix B: International Developments Internationally there are many developments aimed at increasing the discoverability and access to research outputs. A few are highlighted below with others available from AOASG32.

United Kingdom In 2012 the UK Government announced that based on the recommendations of the Working Group on Expanding Access to Published Research Findings (Finch Report)33, a clear policy direction in the UK would be implemented to support for Gold OA publishing. This approach was funded by block grants administered via institutions. Subsequently at least GBP £16.7M was spent in 2014-15 financial year on APCs; however there has as yet been no significant reduction in subscription costs (GBP £180M). The Wellcome Trust tracks OA spending and in 2017 published their analysis34 of the spending by organisations that received a grant from the Charity Open Access Fund (COAF) between October 2015 and September 2016.

A group of UK institutions is exploring a “UK Scholarly Communications Licence” This approach is based on Harvard model OA policy35, which is used by over 60 institutions worldwide. Under this approach, institutions have a policy whereby authors retains copyright and grant their institution a (non-exclusive) right to post the author accepted manuscript in its repository under a non-commercial licence (CC-BY-NC). This approach supports OA policies through a green approach and would reduce the need for APC payments.

Open Access 2020 is an international initiative that aims to build a coalition to change the publishing model from one of subscriptions to OA and drive a fundamental change in the publishing system, by sharing information among partners on best practices and as a first step, converting journals from subscription to OA by re-directing the existing subscription spend into OA funds (via APCs). The Max Planck Institute modelled such a possible transition in 201536. However, there is much discussion about whether this model would necessarily result in a more sustainable situation in all countries.

European Union

30https://www.library.ucla.edu/sites/default/files/Ginny-Steel_open-letter_OA2020-PIF_October-2016_0.pdf and http://bjoern.brembs.net/2016/04/how-gold-open-access-may-make-things-worse/ 31 http://onlinelibrary.wiley.com/doi/10.1002/leap.1116/full 32 https://aoasg.org.au/news-updates/ 33 https://www.acu.ac.uk/research-information-network/finch-report 34 https://wellcome.ac.uk/funding/managing-grant/wellcome-and-coaf-open-access-spend-2015-16 35 https://osc.hul.harvard.edu/policies/ 36 http://pubman.mpdl.mpg.de/pubman/faces/viewItemOverviewPage.jsp?itemId=escidoc:2148961 8

The EU has articulated a vision for “The Three Os”: open innovation, and open to the world. This was most recently championed by the Netherlands during their time chairing the EU in 2016 and which culminated in the Amsterdam Call for Action on Open Science.37 There are a large number of previous and ongoing initiatives all of which support OA across the spectrum including, for example, of repositories through initiatives such as OpenAIRE38. There is also a developing European Open Science Cloud39 project.

A group of German institutions have collaborated on a country wide initiative - Project DEAL. Its stated aim is “to conclude nationwide licensing agreements for the entire portfolio of electronic journals (E-journals) from major academic publishers from the 2017 licence year. The intention is also to bring about significant change to the status quo in relation to negotiations, content and pricing in the process.” It is notable that they go on to state the power of national consortial negotiations “The effects of a consortium agreement at the national level should relieve the financial burden on individual institutions and bring wide-scale, lasting improvements in access to scholarly literature for academics. An open access component is also planned.”40

Netherlands has primarily opted for a Gold OA approach, which is supported by the Government. Funds have been made available for APCs. They note that they believe “this route provides the best guarantee that publications are immediately available. The Green route often means lengthy embargo periods.” The Association of Universities in the Netherlands (VSNU). Is tasked with managing the transition to OA. However, negotiations with publishers have not been easy. A list of publisher agreements is available41.For example, Elsevier has agreed to make just 30% of papers from Dutch authors freely available by 2018, but only after a significant (undisclosed) increase in the annual amount that libraries pay 42.

Finland has had an open science roadmap for several years now - the national Open Science and Research Roadmap 2014–201743. The Academy of Finland requires that “Academy-funded projects commit to open access publishing, which can be green or gold.” There have most recently been intensive negotiations with publishers to reach an acceptable deal with publishers; this is being managed by the library consortium, FinELib. Because of a lack of acceptable progress of negotiations, in June 2017, a boycott of one of the publishers, Elsevier, was launched44.

Canada Canada has an OA policy driven by its three funding agencies - the Tri-Agency Open Access Policy on Publications. The policy allows both green and gold OA. It is of note however, that rising prices of journals have led to a number of universities cancelling big subscription deals, including The University of Ottawa, University of Calgary and Memorial University of Newfoundland.

USA

37 http://openaccess.nl/sites/www.openaccess.nl/files/documenten/amsterdam-call-for-action-on-open-science.pdf 38 https://www.openaire.eu/ 39 https://ec.europa.eu/research/openscience/index.cfm?pg=open-science-cloud 40 https://www.projekt-deal.de/about-deal/ 41 http://openaccess.nl/en/in-the-netherlands/publisher-deals 42 http://openaccess.nl/en 43 http://julkaisut.valtioneuvosto.fi/handle/10024/75210 44 http://www.nodealnoreview.org/2017/06/15/no-deal-no-review/ 9

OA policy in the USA is currently driven by a directive, issued on February 22, 2013 by the White House Office of Science and Technology Policy (OSTP), which required all federal agencies with annual research and development budgets of $100 million or more to provide free and timely online access to articles and data that result from that research. SPARC tracks these policies45.

G7 Countries In September 2017, the G7 science ministers published a communique46 supporting openness in science and the building of supporting infrastructures, including the need to incentivise open practices.

Other Initiatives Some significant funds are being injected by charities such as the Wellcome Trust and the Bill and Melinda Gates Foundation into new OA publishing platforms47 that are managed for them by F1000 Research48. The Open Research Funders Group (ORFG), a partnership of highly-visible organizations committed to the open sharing of research outputs, recently released the HowOpenIsIt? Guide to Research Funder Policies 49 to help funders develop their own open policies.

Starting in Brazil and now expanded to 15 countries is the SciELO50 program that seeks to improve the scientific journals that indexes and publishes in OA. OA is particularly important for developing and emerging countries (who cannot afford access to many subscription based sources) and the evolution of OA publishing in South America is described by SPARC51.

SHARE portal52 by the Open Science Foundation is building a free, open, data set about research and scholarly activities across their life cycle.

Open Library of the Humanities53 is funded by an international consortium of libraries and is dedicated to publishing OA scholarship with no author-facing APCs. It is one of a number of emerging, academic led- initiatives in OA, which are contributing to a diverse landscape.

45 http://researchsharing.sparcopen.org 46 http://www.g7italy.it/sites/default/files/documents/G7%20Science%20Communiqué.pdf 47 https://www.nature.com/news/gates-foundation-announces-open-access-publishing-venture-1.21700 48 https://f1000research.com/ 49 https://f1000research.com/ 50 http://www.scielo.org/php/index.php 51 http://www.sparc.arl.org/news/open-access-latin-america-embraced-key-visibility-research-outputs 52 https://share.osf.io 53 https://www.openlibhums.org/site/about/ 10