2002年2月14日，《布达佩斯开放获取倡议》(Budapest Open Access Initiative，BOAI)正式发布，这是开放获取运动的标志性事件之一。今年是BOAI发布20周年。BOAI在2022年3月15日再次发布新建议，其中有些提法比较有趣，值得大家琢磨。英文全文链接：https://www.budapestopenaccessinitiative.org/boai20/

这次好像没有中国学者参与(BOAI15有，中国科学院文献情报中心原主任张晓林参与了)，暂时也没有找到中文版。我先贴出来英文版，找几个有趣的点点评几句。

有意思的点：

1.长，但具体，实在。

2.再次明确：开放获取本身不是目的，最终的目的应该是科学研究的公平、质量、有用和可持续发展(equity, quality, usability, and sustainability of research)。

3.绿色开放获取依然是第一位的，这再次提醒大家，不用一说开放获取就是APC啥的。就绿色开放获取，直接呼吁高校和科研机构首先得出政策，至于是否自建机构知识库，大意是：小机构别折腾了，参加大的即可。这也是我一直呼吁的，中国需要机构知识库，但不是一哄而上一家一个，有几十个甚至十几个特别好用的即可。

4.提到OA期刊因为年轻，评估时吃亏。这个我一直比较认同，新刊，不管OA还是不OA，都需要一个成长过程，得给OA期刊时间，不要一棍子打死。涉及评估时，强调了一个其实应该是共识的东西：评估工作本身的质量，而不是评估这个工作发表的期刊如何如何。其实，就是全球都有的“以刊评文”现象，得改。

5.关于文章处理费APC的内容很多，偏批评吧，甚至用了Move一词。也明确说，希望绿色开放获取和钻石开放获取(不收APC的期刊)能够替代收取APC的开放获取出版。我个人保留一点意见，愿望固然是好的，可是绿色开放获取普及得有点慢，钻石开放获取期刊数量众多但发文量有限(真发文量大了需要不少费用的)，或许呼吁APC本身的透明、公平公正等，更现实？

6.点名批评转换协议，这让我有点意外。细看内容也可以理解。不过，转换协议本身就是临时的，转换后自然就没有了。全球来看，或许转换需要一个过程罢了。对中国，转换协议其实是一个独特的机会！

英文版原文：

THE BUDAPEST OPEN ACCESS INITIATIVE:
20TH ANNIVERSARY RECOMMENDATIONS

PREFACE

The Budapest Open Access Initiative celebrated its 20th anniversary on February 14, 2022.

To mark the anniversary, the BOAI steering committee is releasing a new set of recommendations based on its original principles, current circumstances, and input from colleagues in all academic fields and regions of the world.

In September 2021 we sought comments from the global OA community on 12 questions. In addition to gathering the email responses to our questions, we hosted a series of Zoom conversations with stakeholder groups and regional communities. The comments informed our discussions about the new recommendations and we thank all who participated.

We remain committed to the principles articulated in the original BOAI statement from 2002 and the 10 year anniversary statement from 2012. But the history of OA has continued to unfold, for example in the growth of the overall volume of OA literature, growth in the percentage of new research that is OA from birth, growth in the number of OA repositories, growth in the number of new OA journals, and growth in the number of non-OA journals converting to OA, and growth in the use and acceptance of OA preprints. It has also seen a proliferation of new OA policies from funders and universities, new services for implementing OA policies, new research assessment practices, new pieces of research infrastructure, new tools, new journal business models, new methods of peer review, new OA options for authors, new advocacy organizations, and new partnerships and alliances.

The same 20 years have sharpened our understanding of certain systemic problems. We know more today than we knew before about the harms caused by proprietary infrastructure, commercial control of research access, commercial control of research assessment indicators, journal-based research metrics, journal rankings, journal business models that exclude authors on economic grounds (just as subscription journals exclude readers on economic grounds), embargoes on repository OA, publisher exclusive rights, narrow fixation on a journal’s version of an article, and tenacious misunderstandings about different methods for providing OA itself. As our understanding improved, we saw the need to favor open infrastructure, academic or nonprofit control of research access and assessment indicators, policies to ensure unembargoed OA, assessment methods without perverse incentives, inclusive journal business models, and fundamental changes to research culture above and beyond changes to the associated technologies, policies, and economics.

We became increasingly clear that OA is not an end in itself, but a means to other ends, above all, to the equity, quality, usability, and sustainability of research. We must assess the growth of OA against the gains and losses for these further ends. We must pick strategies to grow OA that are consistent with these further ends and bring us steadily closer to their realization.

We deliberately do not aim at a comprehensive list of recommendations. We created a long list 10 years ago for the BOAI 10th anniversary and could write an even longer list today. But our experience is that shorter lists can be more effective than longer ones. They prevent high-priority recommendations from becoming buried in an avalanche of worthy but less-urgent recommendations. Moreover, since our 10-year anniversary statement, many other groups have produced excellent recommendations which we support, and which, taken together, cover the territory very well. Among the longer lists, we can single out the November 2021 UNESCO Recommendations on Open Science, because of their breadth and their approval from 193 nations. We urge all UNESCO member states to implement the principles in the new recommendations.

To keep our own list on the shorter side, we make only four high-level recommendations (though we admit with sometimes-lengthy sub-points and elaborations). We also focus on OA for research articles and their preprints, as the BOAI has always done. We strongly support open data, open metadata, open citations, open code, open protocols, open books, open theses and dissertations, open educational resources, open courseware, open digitization projects, open licenses, open standards, open peer review, and the many practices that make up open science. We also see their kinship in a larger ecosystem of open research and education. But to give recommendations on all these fronts would greatly lengthen the list and risk the avalanche problem.

We hope you’ll use the #BOAI20 tag in social-media discussions of these recommendations.

SUMMARY

Open access is not an end in itself, but a means to further ends. Above all, it is a means to the equity, quality, usability, and sustainability of research. Our four high-level recommendations address systemic problems that obstruct progress toward these ends.

1. Host OA research on open infrastructure. Host and publish OA texts, data, metadata, code, and other digital research outputs on open, community-controlled infrastructure. Use infrastructure that minimizes the risk of future access restrictions or control by commercial organizations. Where open infrastructure is not yet adequate for current needs, develop it further.

2. Reform research assessment and rewards to improve incentives. Adjust research assessment practices for funding decisions and university hiring, promotion, and tenure decisions. Eliminate disincentives for OA and create positive new incentives for OA.

3. Favor inclusive publishing and distribution channels that never exclude authors on economic grounds. Take full advantage of OA repositories and no-APC journals (“green” and “diamond” OA). Move away from article processing charges (APCs).

4. When we spend money to publish OA research, remember the goals to which OA is the means. Favor models which benefit all regions of the world, which are controlled by academic-led and nonprofit organizations, which avoid concentrating new OA literature in commercially dominant journals, and which avoid entrenching models in conflict with these goals. Move away from read-and-publish agreements.

Major recommendations for the next 10 years

1. Open infrastructure and its governance 

We recommend hosting and publishing OA texts, data, metadata, code, and other digital research outputs on open, community-controlled infrastructure. By this we mean infrastructure built from free and open-source software, operating under open standards, with open APIs for interoperability, and whenever possible, on platforms owned or controlled by academic-led or other nonprofit organizations. Here we focus on platforms with these critical features, but we support the longer list of features in the Principles of Open Scholarly Infrastructure.

• 1.1. OA research is at risk of enclosure when hosted on closed, proprietary, or commercial infrastructure. The current owner might make the content open today, and might be bound by contract to do so for some time into the future. But one day the same owner could see more profit in restricting access. Or the owner might be acquired by a different entity with a different vision. OA research is safest in academic-led institutions and infrastructure not subject to future acquisition, not subject to profit-maximizing access decisions, and not subject to the risk of enclosure. Adopting academic-led open infrastructure is part of keeping OA research open and making it sustainable. It is also part of decoupling discoverability, peer review, impact measurement, and research assessment from publishing.

• 1.2. We recommend distributed, independent, interoperable modules of open infrastructure over centralized platforms. Centralization creates monocultures, and monocultures create risks of lock-in and stagnation. They create these risks even in the absence of monopoly and even when open.

• 1.3. When universities, research institutions, and funding agencies select or build new research platforms, they should insist on open infrastructure. When nations build research platforms, or call for bids on research platforms, they should insist on open infrastructure.

• 1.4. When it is important to use services from for-profit companies, research organizations should favor those using open-source infrastructure. If the vendor later changes priorities, or changes hands, the research community would still possess the software and undiminished rights to use it.

• 1.5. For a given purpose at a given time, the best closed infrastructure can be better than the open alternatives. While acknowledging that, we recommend that academic institutions always consider enhancing existing open infrastructure before adopting closed infrastructure. If the open alternatives are inadequate, that could be a good short-term reason to choose closed infrastructure. At the same time, it’s a good long-term reason to improve the open alternatives. If we can’t have first-rate open infrastructure today, for a given purpose, we should take steps to have it tomorrow.

• 1.6. Similarly, for a given purpose at a given time, the best centralized infrastructure can be better than the decentralized alternatives. But just as the best closed infrastructure diverts money from building better open alternatives, the best monoculture diverts money from building better decentralized alternatives. In the long run no monoculture platform can be better than a system of interoperable, free and open-source components operating under open standards. The distributed ecosystem lets every project and organization pick the modules that best fit local needs. It lets users develop new modules without building an alternative to the whole system. It lets the global research community scale up to meet global needs without accepting constraints on a central platform or persuading centralized decision-makers to make complex changes. These virtues matter because platforms and infrastructure, like policies and strategies, are most effective when they take local constraints and cultures into account.

• 1.7. Institutions should consider sharing pieces of open infrastructure. Every university should make use of an OA repository, but not every university needs its own. Every OA publisher should make use of management and workflow tools, but not every publisher needs its own. Many institutions and nations need OA publishing portals, but they don’t each need their own and can benefit from national or regional portals such as those in Croatia, Ethiopia, Finland, Greece, Norway, Serbia, Spain, and Latin America. Shared infrastructure has advantages that unshared infrastructure cannot match, even when the unshared pieces are interoperable. To use OA repositories as an example: It’s easier for universities to join a shared repository than launch separate repositories. It’s easier to support text and data mining at a large, shared repository than at multiple separate repositories, and the results will be more useful. It’s easier to preserve a large, shared repository than multiple separate repositories. As we know from large, shared repositories like arXiv and PubMed Central, researchers who use repositories as readers are more likely to see the point of depositing as authors. Shared infrastructure can have state-of-the-art features that no single institution could afford to develop on its own.

• 1.8. When institutions see promising new open infrastructure that they are not ready to adopt, they should try to participate in the development process. When they can, they should help write code. They could ask their developer teams or incentivize their faculty and students to take part. When they can’t help write code, they should at least help the developers understand what institutions like themselves will want and use. Open infrastructure developed in close consultation with potential users will increase the likelihood of adoption, use, meeting local needs, and minimizing wasted time, effort, and funds.

• 1.9. For help in identifying open research infrastructure, we recommend the Global Coalition for Sustainability in Open Science Services (SCOSS); Invest in Open Infrastructure (IOI); the Posse of organizations committed to follow the Principles of Open Scholarly Infrastructure (POSI); and the Scholarly Communication Infrastructure Providers (SCIP). However, none of these lists is complete and some are deliberately selective. Every academic and research organization should watch for open infrastructure that will meet its needs.

2. Research assessment practices

We recommend reforming research assessment for funding decisions and for hiring, promotion, and tenure decisions by universities and research institutions. Careful reforms can eliminate disincentives for OA and create new incentives for OA, without limiting the topics, conclusions, or rigor of research. They are entirely compatible with academic freedom and the highest standards of quality.

• 2.1. Universities should drop disincentives for OA, such as the Journal Impact Factor (JIF) and journal rankings which depend on JIFs. The problem is not that OA journals have lower JIFs than conventional journals. (There’s evidence that OA increases citations, even JIFs.) The problem is that the average OA journal is younger than the average non-OA journal, and younger journals lack JIFs more often than older journals. For the same reason, JIFs discriminate against new journals (both OA and non-OA) on emerging topics, such as climate change mitigation. They also discriminate against journals not indexed in commercial services because of their language, geographic location, or institutional affiliation. Reliance on JIFs for assessment also mistakes an impact metric for a quality metric, and mistakes a metric about journals for a metric about articles or authors. (Also see Recommendation 1.5 from BOAI-10.)

• 2.2. Research assessment committees seldom deliberately create disincentives for OA, for example, because they oppose OA. The disincentives are generally unnoticed side effects of other practices, such as the use of journal rankings, journal-level metrics, rewards for publishing in certain journals, or the false assumption that all or most OA journals are ‘predatory’. Similarly, the dearth of constructive reforms arises from many obstacles and almost never from opposition to OA itself.

• 2.3. Eliminating disincentives for OA from research assessment practices would already be a large step forward. But we also support new practices to create positive incentives for OA. A good example is the University of Liege’s pioneering practice to assess faculty articles for promotion and tenure only when the articles are on deposit in the institutional repository. Several universities have followed suit, but not nearly enough. Clearly this practice does not touch on the topics, conclusions, or quality of research. (Note that we endorsed Liege-like policies in Recommendation 1.6 of BOAI-10.)

• 2.4. When promotion and tenure committees expect early-career researchers to publish in non-OA journals with high prestige or metric scores, the committees should understand the disincentives they create for publishing in OA journals. They should also understand the dilemma they create for researchers subject to funder OA policies. Because these dilemmas can be solved by repository-based or green OA (more in 3.10 and 3.13), committees should positively encourage green OA and not leave young researchers to discover the option for themselves or how it solves the problem. In any case, committees should base their evaluations on the quality of the candidate’s work, not the journals in which they publish. If they use metrics at all, they should use article- or author-level metrics, not journal-level metrics.

• 2.5. We recommend that research institutions sign the Declaration on Research Assessment (DORA) and take steps to implement it. Along the same lines we support the Leiden Manifesto for Research Metrics, the Hong Kong Principles, and the Paris Call on Research Assessment. We recommend that other funders join the Wellcome Trust, Templeton World Charity Foundation, and UKRI in giving grants only to institutions that have signed DORA or an equivalent set of principles and shown a commitment to implement them.

• 2.6. Research assessment committees at universities and funders are in a critical position to create and change incentives. When they do not use this power with care, they can be a major bottleneck slowing progress toward OA. When they use it with care, they can be a major accelerator of OA. When they assume that any modification to assessment practices must compromise quality or academic freedom, they are confusing separate and independent questions. Research institutions should work conscientiously to align the incentives of the promotion and tenure process with the needs of open research.

• 2.7. Dropping journal ranking and journal-level metrics will require changes in research culture. It will require assessing works and people, not journals and publishers. It will require shifting from quantitative metrics to qualitative metrics. It will require that researchers see the artificiality and irrelevance of journal-level metrics and the research purpose for dropping them. It will require that academics stop outsourcing judgments of quality to publishers, or data arising from publisher decisions, and bear the responsibility to judge quality for themselves.

3. Article processing charges

We recommend inclusive publishing and distribution channels that never exclude authors on economic grounds. We recommend moving away from article processing charges (APCs). Viable alternatives have long existed, but they are systematically under-noticed, under-discussed, under-appreciated, under-funded, and under-used. We recommend taking full advantage of these alternatives to enhance the equity, quality, usability, sustainability of OA research. We recommend investments and creative exploration to identify other alternatives to APCs.

• 3.1. APCs exclude authors unable to find funds to pay them. This category includes disproportionately many authors in the global south. It also includes independent scholars and authors from less privileged institutions in the north. APC-based journals exclude authors for economic reasons and without regard to the quality and importance of their work. This harms the authors themselves and the readers who would benefit from their work. It harms research by excluding the perspectives of those authors, their fields, and their regions. It distorts how journals represent their fields and topics by adding acceptance criteria extraneous to the merits of submitted work. Just as we seek to remove paywalls to enfranchise more readers, we seek to remove APCs to enfranchise more authors.

• 3.2. APCs are as opaque and inscrutable as subscription prices. Authors, universities, libraries, funders, and other stakeholders outside a given APC-based journal cannot tell which journal expenses an APC covers or how far the APC exceeds the journal’s expenses. (For the largest publishers we have a clue in their annually disclosed profit margins.) Outsiders cannot tell whether an APC is based on the publisher’s production costs or the university’s historic spend; whether it charges a premium for prestige or impact factor; or whether it charges what the publisher believes the market will bear and aims to preserve or exceed legacy profit margins. For the same reason, APC “discounts” are opaque and inscrutable; they are discounts from prices of unknown fairness. Among other harms, the opacity of APCs supports APC inflation and payments far in excess of services rendered. However, even if APCs were lower, and even if publishers become more transparent about the costs covered by APCs (a goal of cOAlition S and the Fair Open Access Alliance), we still recommend taking better advantage of alternatives to APCs. Lower APCs and APCs transparently tied to a publisher costs are still author-side barriers, unrelated to merit, and obstruct progress toward a more equitable and inclusive system of research communication.

• 3.3. For these purposes it doesn’t matter whether the APCs are paid by universities and funders on behalf of authors. Many authors are not affiliated with institutions that are able or willing to pay these fees. Those authors would be excluded not for their individual economic means but for their institutional affiliations, another variable irrelevant to the quality of their work. Studies show that authors in the global north are best able to find APC subsidies and authors in the global south least able. The burden of paying APCs falls inequitably on authors who have the fewest funding options and who can least afford to pay them. Relying on institutional sponsors to pay APCs helps authors who need it least and exacerbates current disparities. Of course this could change if more institutions — in more regions and economic strata — were willing to pay APCs. But because APCs bring other problems, and because they are unnecessary for publishing research, we recommend that institutions spend new money on alternatives to APCs rather than APCs themselves.

• 3.4. When institutions support OA primarily by paying APCs, they give no-APC or diamond journals a perverse incentive to start charging APCs. For the same reason, they give low-APC journals a perverse incentive to raise their APCs. These consequences aggravate the harm caused by APCs to excluded authors and readers who would benefit from their work. The harm to readers is global and the harm to authors does not fall evenly across disciplines and regions.

• 3.5. APCs feed fraudulent and predatory journals, which harm all the researchers they deceive. They also give OA itself a bad name, even though not all predatory journals are OA. The global OA community currently addresses this problem with guides to vetted journals and efforts to educate stakeholders about the honesty and quality of the majority of OA journals. These helpful strategies should continue. But at the same time, we should starve predatory journals by moving away from APCs.

• 3.6. There have always been two large alternatives to APCs: Repository-based (or “green”) OA and no-APC (or “diamond”) OA journals. Because green and diamond OA are open to readers and authors, not just readers, we recommend giving them more attention, more use, more funding, and more priority. Shifting resources from APC-based OA to green and diamond OA will enfranchise more voices in global research without reducing the quality or openness of research. It will increase the quality of research by sharing perspectives previously excluded. It will reduce the money flowing from nonprofit research institutions to for-profit corporations and enhance community control over scholarly communication.

• 3.7. We are not saying that articles in APC-based OA journals are not OA in the full BOAI sense. They are. When APC-based OA grows, OA grows. The task here is not to make non-OA works OA, but to stop excluding authors on irrelevant grounds, stop distorting research through those exclusions, stop obscuring the economics of OA, stop paying more than necessary, stop the flow of limited funds from the academic sector to the commercial sector, stop subordinating the sustainability of research to the sustainability publisher revenues, and stop entrenching a business model with these consequences. By analogy: Many institutions decided long ago not to pay APCs at hybrid journals even while acknowledging that OA articles at hybrid OA journals are bona fide OA. The task there was not to make non-OA works OA but to stop feeding, incentivizing, and entrenching a deleterious business model.

• 3.8. Universities and funders paying significant amounts of money in APCs should invest in green and diamond OA instead. Many APC-based OA journals, and many non-OA journals, could make the shift to diamond OA with the help of that investment, even if they could not do so without it. For journals that could not make the shift, authors and institutions should still favor journals that do not exclude authors on economic grounds.

• 3.9. Publishers formerly objected that green OA for peer-reviewed manuscripts was parasitical on peer-reviewed journals and threatened their survival. We still don’t know of any journal cancellations caused by the growth of green OA. (Indeed, publishers themselves have produced evidence that their own price increases are the leading cause of journal cancellations.) The objection is less common today, in part because publishers who formerly advanced it have started publishing their own APC-based OA journals. (For the same reasons, publishers have greatly reined in their former objection that APCs lower standards by paying journals to accept new work.) More importantly, the supposed threat of green OA to peer-reviewed journals does not apply to peer-reviewed OA journals, with or without APCs. Nor does it apply to green OA for preprints. Nor does it apply to peer-reviewed journals using open peer review, which requires green OA or the equivalent for submitted manuscripts. Nor does it apply to overlay journals, which are peer-reviewed and blur the distinction between green and gold by using the global network of OA repositories as their distribution infrastructure. We support growth on all these fronts — peer-reviewed OA journals, preprints, overlay journals. We also support the COAR next-generation repository project to enhance the global network of OA repositories with new layers of utility, including peer review, and help the global OA community take better advantage of green OA.

• 3.10. Even in a world in which all new research articles are OA through journals (with or without APCs), we will still want green OA for many scholarly purposes. We will want green OA for preprints, for early time-stamps on new work, for updated versions after publication, for overlay journals, for preservation, for text and data mining, and for research outputs not published in journals such as datasets, source code, books, theses, dissertations, works digitized from print, and innovative new genres of scholarship. We will want green OA for policies from institutions and nations that choose repositories over journals as the preferred venue for OA. This includes policies from institutions that don’t want to limit author freedom to submit new work to the journals of their choice, and policies that want to enable early career researchers to satisfy a funder OA policy and a traditional promotion and tenure committee at the same time (more in 2.4 and 3.12). Finally, we will want green OA as a mature and widely-accepted channel for OA research in case other channels, including APC-based OA journals, fail or become less attractive for any reason, including author boycotts or economic unsustainability.

• 3.11. In many regions and disciplines, OA through APC-based journals has long been the most widely known kind of OA. In fact, many stakeholders still assert or assume that all OA is APC-based gold OA. This misunderstanding persists even though OA repositories predated OA journals and have coexisted with them for the history of OA journals. It persists even though APC-based OA journals are the minority of peer-reviewed OA journals. (As of January 18, 2022, only 30.4% of journals listed in the Directory of Open Access Journals charge APCs, though in 2020 65% of articles published in OA journals were published in the APC variety.) This misunderstanding causes harm. Authors covered by green OA policies (requiring deposit in a repository) mistakenly think they’re covered by gold OA policies (requiring submission to a certain kind of journal). Authors who want to make their work OA mistakenly consider just APC-based journals. If they can’t find funds to cover an APC, they mistakenly conclude that they can’t make their work OA at all. Institutions wanting to support OA journals mistakenly limit themselves to APC-based journals. Subscription journals considering a flip to OA mistakenly limit their deliberations to APC business models and don’t consider other models. Those thinking about what the global research community would pay for peer-reviewed journal articles in a world in which all journals had flipped to OA mistakenly limit the analysis to what the community would pay in APCs. Surveys of author attitudes toward OA mistakenly limit themselves to questions about APC-based OA journals. All stakeholders, including those who support APCs, should correct these misunderstandings wherever they occur. In their own communications, they should speak clearly. OA is delivered by many channels, including repositories, not just by journals. OA journals use many different business models, not just APCs.

• 3.12. How should we take better advantage of green OA? More institutional OA policies should be green-only or green-gold neutral, allowing compliance through green or gold OA at the author’s choice. Most university OA policies are still green-only, which we applaud. Most funder OA policies are still green-only, with a growing number favoring gold. Some like Plan S and the UKRI policy are green-gold neutral with compliance options of both kinds. Another way to take better advantage of green OA is for more universities to negotiate with publishers to autodeposit works directly into their institutional repositories. Nations should negotiate with publishers to autodeposit into national repositories or regional university repositories. More research institutions should support overlay journals that use OA repositories as their distribution channel. More universities and funders should adopt rights-retention policies, to allow green OA when publishers don’t and to support unembargoed and open-licensed green when authors and institutions would not otherwise hold the needed rights. More countries should adopt “Taverne-style” copyright amendments to the same effect, or strengthen their existing amendments by dropping restrictions and embargoes. More institutions and nations that want immediate and openly-licensed OA should understand that green OA can meet those conditions. More universities should launch systematic and adequately funded efforts to gather the institution’s research output for the institutional repository. Universities paying APCs (through publisher agreements or special funds) should redirect growing percentages of that money, and growing percentages of their subscription budgets, to the same green OA efforts. Universities and funders that encourage or require gold OA, and find it hard to pay APCs for all covered authors, should encourage or require green OA and invest instead in the project to collect the covered works for the OA repository. More authors and institutions should favor journals that use open licenses and do not demand exclusive rights, freeing more articles and more versions for repository deposit. More authors and institutions should understand that accepted author manuscripts usually contain the same substantive text as the version of record. Finally (following 2.4 and 3.10), funders and universities should both help early-career researchers use green OA to comply with funder OA policies, especially when their promotion and tenure committees encourage or reward publication in non-OA journals.

• 3.13. How should we take better advantage of diamond OA? More universities, funders, and governments should go beyond moral support for diamond OA to financial support. They could contribute directly to diamond journals managed or edited by their faculty or grantees. They could support diamond OA initiatives such as Open Library of Humanities, OPERAS, and Redalyc-AmeliCA. Or they could support diamond OA indirectly through organizations such as the Fair Open Access Alliance (FOAA), Free Journal Network (FJN), JISC, LingOA, or the Open Access Community Investment Program (OACIP). They could act on the recommendations of the Action Plan for Diamond Open Access. Universities and funders paying APCs should redirect growing percentages of that money to support diamond OA journals. Libraries should redirect funds from their subscription budgets for the same purpose. Those that provide financial support for APCs should provide at least as much financial support for no-APC OA journals. Scholars and diamond OA publishers should help document the variety of ways in which diamond journals find the funds to pay their bills. Just as subscription journals often explore the possibility of converting to OA, and hybrid journals often explore converting to non-hybrid OA, more APC-based OA journals should explore the possibility of converting to diamond OA. Existing diamond OA journals should support this exploration by working with journals considering the model on why they adopted the no-APC model and how it works in practice.

4. Read-and-publish agreements

When we spend money to publish OA research, we should remember the goals to which OA is the means. We should favor publishing models which benefit all regions of the world, which are controlled by academic-led and nonprofit organizations, which avoid concentrating new OA literature in commercially dominant journals, and which avoid entrenching models in conflict with these goals. We recommend moving away from “offset”, “read-and-publish”, or so-called “transformative” agreements.

• 4.1. We’ll refer to these agreements collectively as “read-and-publish” agreements, while acknowledging that there are many variations on the theme. We don’t use the term “transformative” because it’s not self-explanatory, because it preempts a useful term that could apply to other transformative or transformational initiatives, and because it’s more honorific than descriptive.

• 4.2. First we note that our respondents are divided on these agreements. Some support them and some don’t.

• 4.3. Read-and-publish agreements have the beneficial effect of increasing the number of OA articles. We welcome this growth but worry about making it a foreground goal pursued at the expense of background goals. These agreements depend on the APC model and entrench it even further. Hence, they grow the OA corpus by methods that exclude some authors. We want to grow the OA corpus by methods that include authors without regard to their institutional affiliations or economic circumstances. We want to refocus on the goals served by OA itself. Finally, we believe these agreements are unsustainable, by paying more than necessary and putting short-term growth ahead of long-term growth.

• 4.4. Proponents often say that these agreements dispense with APCs. But that is misleading. While authors don’t pay APCs at covered journals, and needn’t look for funding to pay APCs, the negotiations eventually agree on a certain cost of publication that the institution should bear. These costs charged to the institution are APCs or the equivalent for present purposes. Instead of eliminating APCs, these agreements arrange for institutions to prepay them on behalf of authors. (See 4.10 for some nuances.) This arrangement is an advance over author-facing APCs, but still carries the other disadvantages of APCs. APCs prepaid by these agreements are just as opaque and inscrutable as author-facing APCs. The same is true for the APC waivers and discounts they are supposed to deliver. We may know the amount the two parties used for the calculation, but we don’t know what publisher expenses it covers or how far it exceeds the publisher’s expenses. In some variations on the theme, agreements give a university an unlimited number of waivers. In those cases the APC or equivalent reduces in price as more covered authors publish in covered journals. That is better than variations without this feature, but does not address other problems with this model and even aggravates the problem of steering the bulk of covered research to the commercially dominant tier of publishers (see 4.9). It excludes authors from institutions that cannot afford these agreements. These agreements also reduce incentives for authors or institutions to favor journals with lower APCs, or no APCs (the majority of all OA journals, remember), and for funders to pay APCs alongside universities.

• 4.5. A journal APC can be high because it’s tied to high expenses, for example, at highly selective journals. In those cases the problem is not that the APC is more than necessary to cover the journal’s expenses, although it might still be more than necessary for that purpose. The problem is that it’s more than necessary to publish the same article in another venue, for example, in a journal with a lower APC, a journal with no APC, or an OA repository. This problem is intrinsic to APC-based OA journals, even those with lower production costs and APCs. There must be a special and weighty justification to pay a fee to share research when the same research can be shared without paying a fee. When the author can publish the same article elsewhere, the justification cannot be that the payment improves the article’s quality. Nor does the payment  improve the article’s quality when its purpose is to pay for a journal’s brand, prestige, or metric scores. Those fees don’t pay for improved quality, but for the perception of improved quality. We understand that career advancement can depend on that perception. But that is a problem to solve, not an immutable reality to accommodate. The most equitable, sustainable, and academically honest solution is for research assessment committees to pay less attention to where research is published and more attention to the quality of research itself. (See Recommendation 2.) Those committees, above all other stakeholders, should focus on quality over prestige, when the two differ. Whether or not APCs are closely tied to production costs, paying them gives up the fight to promote models that scale sustainably for all disciplines, regions, and economic strata.

• 4.6. Journals covered by these agreements continue to charge subscriptions and make only some of their articles OA. They are hybrid journals. We advise against paying APCs at hybrid journals, as many universities and funders have long since decided. One reason is to make limited APC funds go further. Another is that hybrid journals charge higher average APCs than full OA journals, even though hybrid journals also benefit from subscription revenue. Another is that hybrid journals often double-dip (charge twice for their OA articles, once through subscriptions and once through APCs). Full or non-hybrid OA journals cannot double-dip. However, the main reason is to avoid creating perverse incentives for authors and journals receiving the money. Paying APCs at hybrid journals pays the journals to stay hybrid. It pays them to resist the conversion to full OA that many institutions intend and predict when they enter the agreements (more in 4.7, next).

• 4.7. Proponents of these agreements often predict that the covered journals will convert from hybrid to full or non-hybrid OA. But we aren’t seeing many conversions. Nor are we seeing publisher plans for conversion. Nor are we seeing agreements that make conversion an enforceable provision of the contract. On the contrary (4.6), we see incentives for journals covered by these agreements to remain hybrid and resist conversion. Note, however, that even if we start to see agreement-based conversions from hybrid to full OA, that will not be a significant kind of progress if the new OA journals rely on APCs.

• 4.8. Many institutions make these agreements in part to see whether they will bring about the expected journal conversions. We support experiments and recommend that these institutions make their support appropriately experimental or provisional. For example, cOAlition S is willing to support these agreements until the end of 2024 but not after that. It was recently joined in this decision by Sweden’s Bibsam Consortium, based on Bibsam’s first-hand experience with read-and-publish agreements. Similarly cOAlition S funds is willing to pay APCs at hybrid transformative journals that are making demonstrable progress toward conversion to full OA, but only until the end of 2024.

• 4.9. These agreements tend to be made by the largest and wealthiest publishers, steering much of the new OA literature toward themselves (4.4). This aggravates the monopoly effect of the current publishing landscape and excludes diamond OA journals (the majority, remember), born-OA journals, full or non-hybrid OA journals, and those that may be smaller, nonprofit, less expensive to research institutions, or some combination of these. In that sense, these agreements not only entrench the APC model, but also entrench the current system of journal prestige and the current winners under that system. Likewise, these agreements tend to be made by the largest and wealthiest universities, widening rather than narrowing the publishing-access gap between them and less wealthy institutions.

• 4.10. The financial support we recommend for diamond OA journals (3.13) differs in two ways from the financial support for OA journals provided by these agreements. First, diamond OA journals are not hybrid journals. Second, diamond OA journals offer no-fee publishing to all authors, not just those affiliated with institutions taking part in the agreement.

• 4.11. Because we support experiments, we are ready to change our position if these agreements start to cause journal conversions that meet our objections. In this sense, it’s encouraging that so many institutions and publishers are developing new variations on the theme.

March 15, 2022