Sharing Published Short Academic Works in Institutional Repositories After Six Months: The Implementation of the Article 25fa (Taverne Amendment) in the Dutch Copyright Act

The ambition of the Netherlands, laid down in the National Plan Open Science


Introduction
The Dutch government has set itself the goal that by the end of 2020 all publicly funded academic publications must be open access (van Wezenbeek et al., 2017, p. 9).Within that context, the Dutch Research Council (NWO), the Netherlands Organisation for Health Research and Development (ZonMw), the Association of Universities in the Netherlands (VSNU), the Dutch Federation of University Medical Centers (NFU) and the Royal Academy of Arts and Sciences (KNAW) have tightened their open access policy in recent years.In the last five years the VSNU and the NFU have invested in open access via, so-called, read and publish (R&P) deals whereby articles are immediately made available on a publisher's platform and/or journal with an open licence, in this case preferably CC-BY.Negotiations are conducted with the larger legacy publishers with the aim of publishing all academic articles with a corresponding author from a Dutch university in open access.Reading access to all other closed articles is secured without additional costs (Heijne & van Wezenbeek, 2018).So far, eighteen of such deals have been concluded with publishers (openaccess.nl, n.d. a).The open access articles published under these deals are often still in journals that also publish closed access articles, hence the term "hybrid journals".
Only a very small number of full open access journals are part of some of the current R&P deals (e.g., the agreement with Cambridge University Press offers a list of full open access journals as well as the latest agreement with Elsevier) leaving aside that there are no agreements with full open access publishers yet.Almost all the APCs for full open access are now being paid decentrally by institutions.Despite all the efforts and good progress, it is to be expected -the figures of 2020 are published after the summer of the subsequent year -that for the year 2020 between 70% and 75% of the peer reviewed academic articles will be open access.(Bosman et al., 2021;VSNU, 2020a).With only such deals the target of 100% open access will be very difficult to achieve, not to say almost impossible.Not unimportantly, co-authors are not included in these arrangements.The long tail of publishers in the remaining 25% to 30% range is simply too big to cover with R&P deals.
Furthermore, this route of R&P deals does not cover all publication types.To create more diversity in options for researchers and not to depend entirely on R&P deals, it is therefore important to, in addition to the golden route, also invest in the green route, in which researchers deposit their own research output in a local or disciplinary trusted repository and make it widely available.This route has the advantage that it can and already does cover more publications and publication types, such as articles from smaller publishers and disciplines, that fall outside the current R&P deals.And it is an easy solution for articles with a co-author working at a Dutch research institution.The latter supports the broader goal of the National Programme Open Science (NPOS) to have open access for all publications, i.e. for co-authored articles as well (van Wezenbeek et al., 2017, p. 21).Unfortunately, the green route, in many cases, does not provide immediate access and is often limited to specific versions.Embargo periods are imposed, sometimes lasting several years, which is undesirable.Publishers usually determine these embargo periods, which has led to a negative trend in terms of the length of these embargoes (Jubb et al., 2017, p. 17).In twelve years there has been a dramatic increase in restrictions on what, how, where and when publications can be shared via the green route (Gadd & Troll Covey, 2019).
Besides researchers being unfamiliar with the possibilities of this legal provision, also the vague formulation in the legal text (e.g., what is considered a short scientific work and what is a reasonable period of time?) caused initial reluctance among researchers to make use of this right.For university libraries it was also difficult to use Article 25fa effectively because of the legal uncertainties (e.g., what exactly is considered to be a 'reasonable period of time'?).To increase the impact of Article 25fa, it is crucial that researchers are encouraged and supported by their organisations and that Article 25fa is translated into joint guidelines.As a result of doing this at the level of the NPOS, principles were created.It also gained strong governmental support, since the implementation was included in the Sector Agreement for Academic Education 2018 of the Ministry of Education, Culture and Science and the VSNU (Rijksoverheid, 2018).

Preparations
Firstly, the partnership of the Dutch University Libraries and The Royal Library of the Netherlands (UKB) did an analysis, including a legal consultation, to determine what the added value and the business case would be.
Secondly a proposal with principles and a project-based approach for implementation was drafted for the approval from the VSNU.This was followed by the adoption by NPOS.In 2018 the UKB and VSNU picked up the baton to work on the pilot and set out concrete outlines for Article 25fa, so it would be useful for researchers.This has been put to the test in the VSNU pilot project 'You Share, We Take Care!' in 2019 (openaccess.nl, n.d. b).The pilot was necessary to flesh out the definitions, implications and possibilities of Article 25fa.Principles and implementation guidelines had to be defined, which eventually were set at a six-month embargo period for either the version of record (VoR) with no distinction between disciplines.After gaining legal advice on the legal text, the conclusion was that there is legal ground for VoRs to be made open access.Finally, alignment between universities had to be secured in the event that publishers would challenge the principles as agreed on by the VSNU and researchers would receive takedown notices.It was expected that researchers needed to feel 100% comfortable with making use of this right.

Guiding Principles and Pilot Conditions
For the pilot the following set of guiding principles were defined, being: 1) The work will be shared in its definitive, published version (VoR).
2) The university libraries will only make the publication available through their repository.3) One of the authors has a current employment contract with an institution participating in the pilot (after the pilot this has been extended to emeriti professors and external PhD candidates).4) A reasonable period of time will be uniformly interpreted as six months, regardless of discipline.5) 'First publication' is the date on which the published version first becomes available online.6) Short scientific work includes journal articles, as well as conference papers and individual chapters in 'edited collections'.It does not include monographs or chapters of monographs.7) The amendment applies regardless of a publisher appealing to their respective national laws and/or of the researcher being a co-author whose work has been (partly) funded by Dutch public funds.The inclusion of a researcher as co-author is prima facie evidence that his or her contribution is of substantial value to the final product.
It is obvious that this vision and set of guiding principles does not always correspond with the guidelines of publishers.However, there was a need for clarification and unification on how to engage with researchers and gain their trust in making use of Article 25fa, how the workflow processes could be made as efficient as possible and not unimportantly also about the risks of receiving possible claims from publishers.By not immediately implementing the vision broadly but by organising a small but representative pilot phase, a testing ground was facilitated and as a result any impact of imposed adjustments was reduced.
One of the ways in which the institutions acted jointly was the agreement for a "double guarantee" scheme: participating authors are supported by their institution in legal cases and institutions share knowledge and possible legal costs if such cases arise.Collaborating this way also offered organisations the opportunity to investigate and learn how workflow processes could be efficiently set up and whether local guidelines should be (further) developed into an easily accessible manner for the benefit of the local research community.Another encouraging example of leadership endorsing the project was the joint statement by the rectors of all Dutch universities in which they announced they would use Article 25fa to share all their (older) academic articles open (openaccess.nl,2019a).
A national project team provided full coordination and was responsible for overall planning and management, legal advice and communication support for all universities.Each university had its own smaller project team.
During the pilot phase knowledge sharing between the national project team and the institutional teams was very important.At the end of the pilot phase, advice from the national project team was given to the NPOS stakeholders on the (re)calibration of the national principles and the way in which this could be widely implemented at both research organisations and funders within NPOS.

Pilot Results
More than 600 researchers, from 13 research universities, participated and more than 2,800 publications were deposited through the institutional repositories during this pilot (February 2019 to August 2019).In addition to peer-reviewed articles (75%), this included conference proceedings and book chapters from edited collections.The top three publishers were Elsevier (28% of publications), Wiley (12%) and Springer/Nature (10%).Many participating researchers wanted to share both recent publications and older material.
Researchers were overall positive about the extra opportunities to share work and the support from the university library.The major reason for researchers not to participate was uncertainty regarding the reactions of publishers and/ or their co-authors and feelings of uncertainty with regard to their position as for example being a journal editor if they deviated from the publisher's policy.In the course of the pilot interest from researchers grew substantially.
During the pilot there had been several meetings with the Dutch Publishers Association (Mediafederatie) where concerns about the implementation of Article 25fa were expressed by publishers.In particular, questions were asked about the six-month period for the Humanities and Social Sciences disciplines and the inclusion of book chapters.These concerns have been addressed by the VSNU but did not lead to changes in the guiding principles and set goals.Until now, there have been no formal requests for take-down or legal claims from publishers regarding the sharing of publications (openaccess.nl,2019b).
There have been very few occasions where publishers have threatened individual researchers with take down requests, but these were effectively dealt with by libraries and researchers by pointing the respective (society) publishers to the author's rights anchored in the Dutch Copyright Act. 1 The added value of Article 25fa as interpreted in the VSNU-pilot lies in: 1) providing a fall-back option in cases where other routes are not possible or less desirable for authors (e.g.reluctance in certain disciplines to share AAMs (Author Accepted Manuscripts) or embargoes of one year or longer); 2) facilitating open access for all short scholarly works, not just articles; and 3) facilitating retrospective open access.All three could support the advancement of open access in the Netherlands.Currently various institutions are considering how to effectively embed the principles of Article 25fa in their policy and daily practice.Whereas, for example Utrecht University has opted for a pragmatic, flexible model that can be scaled up easily when the need arises, Leiden University has set up an open access policy by which the principles of Article 25fa automatically apply to all short academic works affiliated with Leiden University, unless all Leiden authors decide to opt-out (Leiden University, 2020).We will go into more detail in the use-case section of this article.
Liber Quarterly Volume 31 2021 By order of the government, the Copyright Act including Article 25fa was evaluated late 2019 (van Gompel et al., 2020).For Article 25fa it is stated that the pilot 'You Share, We Take Care' has shown that it can make a significant contribution to the ambition of making as much academic work as possible available as open access.It is also interesting that the authors of the evaluation report indicate that universities could mandate their employees to make use of the provision (van Gompel et al., 2020, p. 58).This is a support for the opt-out approach described above and can be used by legal departments within the institutions to help support the implementation of such an opt-out.
Article 25fa is making the universities less dependent on publishers in the goal to achieve 100% open access.Increasing on the one hand pressure on negotiations with publishers for sustainable R&P deals and on the other hand when reaching APC caps within existing contracts, this both creates an increasing need for the use of Article 25fa.And last but not least, it functions as a safety net for publications that don't fall under a publishing agreement.In 2021 the VSNU published a feasibility study 'Towards 100% open access for Dutch Research Publications' (van der Graaf & Johnson, 2021) in which various recommendations are made to achieve the goal of 100% open access for all peer-reviewed research output -including articles, books and book chapters.One of the main take-aways is that a multi-track policy is needed.In addition to the recommendations for the gold (and diamond) route, Article 25fa is explicitly mentioned as a powerful tool to enhance the green route.However, it should be mentioned that not all publications made open under the Taverne conditions can count towards this 100%, given the fact that the national methodology of open access monitoring has its reference date in the first half of the year, looking back to the previous year.The dataset for monitoring articles in the previous year is downloaded from the CRIS in the first or early second quarter of the current year, at the latest.Hence, and because of the six-months embargo, articles which have been published in closed access in the months October till December won't be counted as open access in the following year when opened using Article 25fa (VSNU, 2017).
Article 25fa (i.e.empowering the green route) can be positioned as a safety net if no (new) agreements can be made with publishers.Further upscaling may lead to more organisations, both within and outside the academic world applying Article 25fa.

Retrospective Open Access
Another benefit of the Article 25fa is making it easier to achieve retrospective open access for publications from previous years, even before 2015 when the Article 25fa was adopted.This is an important route to explore.Libraries and individuals are still paying for access to back issues of journals.Some institutions can only keep access to a (rolling) window of back volumes under their license.Retrospective open access is only sometimes provided by publishers for publications older than a few years.Article 25fa has proved to be a very strong tool to provide open access to older publications, because everything older than six months can be opened instantly.However, it deserves exploring whether this could be scaled up at an institutional level in terms of capacity to collect, describe and store older short works.Still, it would be without an open license.

Two case studies: Leiden University and Utrecht University
In this part we describe two case studies at two different universities, namely Leiden University (LEI) and Utrecht University (UU).The description of these cases mainly focuses on process, system and results.We conclude the case studies section with some overall conclusions regarding the given support for researchers.
The case at LEI, with its recently adopted opt-out, is an example of a more policy driven approach.The implementation and effectuation at UU has been mainly process driven: an opt-in strategy, which benefits from an optimisation of workflow, in order to achieve large quantities.

Pilot Phase
At the Leiden University Libraries (UBL), staff of the already existing project 'UBL Helps with Open Access' actively supported researchers to make their publications open access.During this project, researchers were asked by the project assistant to provide versions of their articles published in 2018, 2019 and 2020 that could be uploaded open access in the Leiden Repository.These files were uploaded by the project assistant.
During the 'You Share We Take Care' pilot the same project assistant uploaded, and in accordance with the guiding principles and conditions, the published version for a selected group of Taverne participants.The selection of the participants was done by the open access experts of the Centre for Digital Scholarship (CDS) at UBL together with research policy officers of the faculties.Most of the researchers who were asked to participate in the pilot project responded quickly and enthusiastically.The CDS took care of the License Agreements ("for making a Short Scientific Work Publicly available in the Institutional Repository under Article 25fa of the Dutch Copyright Act"), in which the researchers grant the university the right to deposit the short works in the institutional repository, to be signed by and returned to the participants.The secretarial office took care of the signature of the university librarian (on behalf of the university) and the archiving of these license agreements.At Leiden University, only publications from the current and the previous publication year were included (see Figure 1).The pilot only had the intention to explore the potential scope of Article 25fa by testing the principles (such as the six months embargo).Moreover, uploading has proven to be labour-intensive.UBL did not have enough staff to upload all the publications of the years before 2018.

After the Pilot Phase
At Leiden University, uploading was temporarily discontinued after the completion of the pilot.The university first investigated whether the labourintensive process, in which researchers give permission via a two-sided contract, can be arranged more efficiently.Legally it was possible to do this.Permission has been arranged by adding the Article 25fa terms and conditions into the new open access policy and the employment contract.Based on the policy, and the employment agreement, academics who do not want to make use of the terms and conditions of Article 25fa can always indicate that they do not wish to use Article 25fa for a particular work.By including the Article 25fa terms and conditions in the policy and employment contract at Leiden University the university made the transition from an opt in to an opt out policy.When the policy was adopted in December 2020, academics and support staff could upload publications under the Taverne terms and conditions, also retrospectively.In the Leiden University Current Research Information System (LUCRIS) the author or support staff can select '25fa Aw (Taverne)'.LUCRIS is an information system at Leiden University to register and manage metadata for the research output and activities conducted by researchers of Leiden University.LUCRIS is connected with the Leiden Institutional Repository.The repository staff checks the metadata of all publications.

Pilot Phase
The UU decided to run the 'You share, We take care' pilot as if it was already a service implemented by the Utrecht University Library (UUL).All subject specialists from the library were involved in recruiting researchers for participation.The entire UU Repository team was involved in processing (i.e.validating and uploading) the publications of all participants.The secretarial/ administration office took care of the license agreements to be signed by both the participant and the university librarian (on behalf of the university).
Involving the entire repository staff also made it possible to make articles and book chapters open access available retrospectively.Library staff were able to make all the publications of the participants available which were published during or as a result of their affiliation to UU.This was expected to be an incentive for researchers to participate.It would also enable the library to make older publications open access available, for publication years with less open access availability than the current and previous publication year, which was the initial focus of the national pilot.
During the pilot phase 1005 short works were made available and counted in scope for the pilot evaluation (i.e.eligible works were published six months before the start of the pilot, which was February 2019 and onwards).All the other articles are either opened after the pilot phase and/or aiming at retrospective open access as Figure 2 shows.

After the Pilot Phase
After the pilot UUL continued the Article 25fa service, since it was already implemented as if it was a regular service.The UU made two major improvements after the pilot phase.The UU has incorporated Article 25fa (adding "Taverne") in the standard PURE installation (CRIS -Current Research Information System) and DSpace (repository) workflow of the library.The repository staff checks the metadata of all research publications that have been registered in Pure and adds the full text if available and makes them open access available depending on publisher policies.The UU has added a branch to this pipeline, in which new publications by Taverne participants are captured and made open access following the (joint) principles of Article 25fa.The UU has also introduced an online registration form to replace the signing of a paper copy of the license agreement.It safeguards the authentication of UU researchers by using a verification email that is being sent to the institutional email of the researcher and requires verification before the participant is registered.It also generates a PDF-copy of the license agreement for both the researcher and the institution.

Next Steps and Further Implementation
Universities have already started scaling up the use of Article 25fa in the course of 2020 (VSNU, 2020b).Some are considering moving from an opt-in to an opt-out, similar to what University Leiden has done.Another development is that the University Medical Centers embraced Article 25fa early 2021, which means that a significant amount of output in the biomedical field could potentially be opened under the same terms and conditions as described in this article.In a letter about the progress on open access in the Netherlands, sent in the fall of 2020 from the VSNU to the Minister of Education, Culture and Science, the following phrase is an explicit recommendation for Article 25fa: We need "a legally enshrined right to retain the right to use a work at all times to be shared immediately upon publication, if necessary in the author's version (requires sharpening of Article 25fa).This will also have to apply when copyright or an exclusive license has been transferred to a publisher."(VSNU, 2020a, p. 4).It is important to understand how Article 25fa relates to Dutch research funders' policies.As said, funders aim for full, immediate access, with an open licence.With Taverne it is not possible to meet those requirements yet.The VSNU letter also hints at a 0-month embargo with an open license, albeit the author accepted manuscript when necessary.Here an essential requirement to take into account is the retention of copyright.
Recently, the Association of European Research Libraries (LIBER) published their thought-provoking 'template model law' which can be seen as a practical tool that allows authors to retain rights and share their work without embargo or any restrictions on reuse (LIBER, 2021).
In the aforementioned VSNU feasibility study towards 100% the passage "partnerships with other institutions will greatly facilitate an optimal use of the Taverne clause in the Dutch Copyright Act (Green OA route)" (van der Graaf & Johnson, 2021) gives reason to think about how non-university research institutions can (start to) make use of Article 25fa.Further implementation and setting up support for researchers requires specialist knowledge.This was gained in the 'You Share We Take Care' pilot and the subsequent phase.However, it is important to further safeguard this in the academic organisations that make use of Article 25fa and the organisations that want to make use of this in the future.A good collaborative support infrastructure, where sharing of best practices and use of tools would be organised, is essential to make this successful.

European perspectives
Similar copyright legislation (the right to share academic work), although differing in scope, are already in place or seriously considered in other EU-member states, like in Germany, Belgium and France (Sondervan, 2020).It is clear that there is a place for national legal arrangements to foster and advance open access.Most remarkable differences when comparing the other national legislation/amendments with the Dutch legislation are disciplinary specific embargoes (e.g.SSH versus STEM disciplines, the first usually having longer embargo periods) and the eligible version (author accepted manuscripts versus VoR).The question remains how to promote and scale the use of these amendments, but the Dutch implementation (e.g., a national project team, project teams at institutional level and an investment in support for both researchers and administrators) could be seen as a proof of concept.Perhaps there lies a role for the European Commission to investigate to what extent it is necessary or useful to align these national amendments in the near future.Indeed, in its recent communication on the new European Research Area it was announced that the European Commission will start an analysis of "authors' rights to enable sharing of publicly funded peer-reviewed articles without restriction" (European Commission, 2020, p. 14).
The Dutch experiences can serve as an example, although it can also be concluded that just providing researchers with the legal right to share will not automatically lead to more papers being made open access.Researchers need support, guidance and reassurance.Libraries are well placed to provide them with that.

Fig. 1 :
Fig. 1: Number of articles by publication year of short works at University Leiden released under Article 25fa.