Where we have been: history of the foundation’s OA policy

The first iteration of the Gates Foundation’s open access (OA) Policy was publicly announced in 2014 and formally launched in 2015. The foundation sought to play a key role in driving systemic change in scholarly publishing by addressing the inequitable and unsustainable practices that undermine access to potentially life-saving knowledge. The OA policy ensures that new ideas, data and findings are made available and accessible to the broadest range of researchers as quickly as possible. It prioritizes access, transparency and equity so ideas and solutions can be shared, tested and built upon by researchers, collaborators and decision-makers globally.

When the policy first launched, the journal landscape did not widely support OA and many of the most selective journals were not considered compliant for grantee publication. Grantees therefore found themselves caught between their funder and their ambitions. To this day, the age-old ‘publish or perish’ paradigm that rules academia career advancement created obstacles to transforming publishing into a more open ecosystem.

All active grants were included in the policy and the OA policy was housed in the ‘Publications Clause’, which is standard in all grant agreements. The policy required that all articles stemming from funded research must be immediately available upon publication, i.e. outside a paywall or subscription, and licensed with a Creative Commons (CC BY) license. To date, the foundation only supports the CC BY license as it is one of the most liberal reuse licenses allowing both commercial and non-commercial partners access to research to foster innovation. There are also numerous reasons why an article may need to be translated into various forms and languages to have broader reach and reuse, for diverse communities.

With these factors in mind, the primary goal of the original policy was to ensure that research the foundation funded was immediately and openly available for reading and reuse. The secondary goal was to make OA the norm – that all the articles published globally would be done under an OA model. While there has been a steady rise in OA articles as more and more publishers and journals have adopted the model, most published research remains behind a paywall.¹

Early impact of the policy

The first few years of the foundation’s OA policy entailed educating grantees on their OA options while advocating for broader ecosystem change by engaging with other funders and impacted groups via talks, publications and membership in groups like the Open Research Funders Group.² From 2015 to 2017, the number of OA articles published rose from 986 to 2,000. The larger OA movement was gaining momentum and there were increased calls from other funders and institutions for more openness and sharing of research outputs.

In 2017, following in the footsteps of the Wellcome Trust, the foundation partnered with F1000Research (F1000)³ to launch Gates Open Research (GOR), a fully open post-publication peer-reviewed platform. At the time, it was considered a radical way of rethinking academic publishing, putting the publication before the peer review with indexing contingent on positive reviews. It demonstrated an effective and efficient model for communicating research leveraging the available technology. The European Commission launched its version, called Open Research Europe in 2021. However, some elements, such as open peer review and lack of impact factor, have slowed acceptance in the research community. Despite this, GOR is consistently one of the top five publication choices for grantees. The platform will never have an impact factor, which echoes the foundation’s commitment to the San Francisco Declaration on Research Assessment (DORA) and its goals around shifting the perverse incentive system culture. The platform’s version control allows authors to address peer reviewer reports and efficiently update the article, while the mandatory open data requirements ensure that readers have the underlying information they need to truly trust the conclusions. As the foundation’s OA policy has always included an underlying data sharing requirement – a requirement that most journals struggle to meet – GOR assists grantees in ensuring their underlying data is made as open as possible. The result is a more transparently sourced and potentially useful article.

Partnering with F1000 to build GOR has yielded great experience with alternative publishing models like Publish-Review-Curate⁴ with fully transparent costs. GOR articles are published at a very cost-effective rate when compared to the article processing charges (APCs), otherwise known as open access fees, other journals charge. The all-time average APC that the foundation pays is US$2,868 versus US$1,100 for GOR. The partnership with F1000 has given us further insight into the cost and labor associated with running a journal platform.

cOAlition S membership

In 2018, the foundation was invited to join cOAlition S and to adapt its own policy to reflect the Plan S principles.⁵ It was a natural next step for a funder with an OA policy as collective action is critical for change. Funders are conscious of the burden that different policies can have on researchers and want to make compliance easier for researchers with multiple funding sources. Plan S stood for ‘science, shock, solution, speed’,⁶ which felt like the antithesis of the current archaic state of publishing. The moonshot goal was to make open access the norm within an accelerated time frame of five years.

Within cOAlition S, the foundation adopted various strategies to ensure that grantees had the ability to publish in the journal of their choice but on OA terms that fulfilled the Plan S principles.⁷ This included discontinuing support for hybrid journals, implementing the Rights Retention Strategy (RRS),⁸ participating in the Transformative Journal program,⁹ and the still evolving ‘Towards Responsible Publishing’¹⁰ work.

The RRS was a natural fit for the foundation as one of the core components of the OA policy is ensuring that grantees keep sufficient copyright to share and reuse their funded research. The ‘Publications Clause’, which is included in all grant agreements, was updated to ensure grantee authors retained enough copyright to make a version of their article OA with a CC BY license. This was an important shift in the foundation’s OA policy as the first iteration focused solely on the journal’s version of record (VOR) which could be paywalled. With the RRS, the focus shifted to repository deposit of the author accepted manuscript (AAM). This is the version of the article that has gone through peer review, final author edits and is accepted by the journal, but has not yet gone through the final typesetting and journal branding. This changed the signal of what was of paramount importance – the content of the article and the peer review process, not the journal branding. One of the greatest achievements of cOAlition S has been starting a snowball effect of institutions following suit and developing their own rights retention policies.¹¹ In due course, this could have a very powerful impact in safeguarding OA and author choice. It is hard to assess the success of the RRS and how many papers have been made OA following this route. The intent of the RRS was to ensure that grantee authors had the freedom to publish in their journal of choice (OA or not) in the model they wanted (OA or subscription) and still remain funder compliant. Grantees were compliant when they posted their AAM in a repository with a CC BY license. It did not take long, however, before publishers began to deploy measures that were presumably done to block the RRS, such as eliminating the grantee author’s choice of publication route and forcing an OA workflow and APCs or by changing the journals’ copyright transfer agreement to disallow sharing of the AAM. Despite the cOAlition S attempt¹² to shed light on these publisher actions (specifically from hybrid journals) such clarity has not been achieved and we hear instead of the negative experiences of our grantee authors.

To help track reporting and compliance the foundation provided a grant to OA.Works in 2021 to bolster this function.¹³ It was important to partner with a non-profit organization that was similarly mission-driven and upholds the ideals of ‘open’. The goal is to create tools that help put OA policies into practice and deepen their impact.

Plan S policy alignment began on 1 January 2021. It was not an ideal time to update a major policy as the world was in the grips of a global catastrophe. Yet, unexpectedly, the Covid-19 pandemic became the ideal test case for what OA advocates had been striving for, for decades – an environment where information flowed freely leading to global collaborations to solve pressing issues. For the first time we saw open science practices being used and not just theorized. Paywalls came down and data was shared and accessible. Sadly, it was only a matter of time before these open practices receded as paywalls returned and data regained its elusive nature.

Lessons learned

The most important lesson learned through this work is our responsibility, as a funder and policymaker, to respond to the criticisms and learnings of the community. Funders need to follow the lead of OA advocates – especially librarians and researchers – who work tirelessly to improve the publishing process and infrastructure.¹⁴ Other things that we have learned are:

• publishing does incur costs, and we see this reflected in the wide range of business models. The highest cost, however, seems to be paying for prestige and the processing of rejected papers
• there are good examples of not-for-profit publishers who place the importance of access, quality publishing and community-building above profits
• because academic publishing is intrinsically tied to career advancement, authors have little incentive to experiment with new journals, new publishing technologies or different business models¹⁵ that do not conform to institutional expectations
• policy does drive the behavior change of researchers – less so for publishers. Change is slow, but it does happen
• policy focused solely on the journal VOR has stymied opportunities for other models to emerge and gain traction. It is important to support multiple open access routes. For the career incentives to change, the importance of venue of publication must be deprioritized¹⁶
• authors are not price sensitive – especially when their careers rely on publishing. They also may not be aware of the nuance and impacts of commercial versus not-for-profit publishing, thus inadvertently perpetuating an expensive and inequitable system¹⁷
• the main discussion on open access and academic publishing remains focused in the Global North. We cannot make progress without addressing this. Other countries have achieved cost-effective open access solutions. Instead of ignoring these feats the Global North should learn from them.

Where we are going: supporting equitable practices in research dissemination

In its ten years, the foundation’s OA policy has driven real change in scholarly publishing, spurred by a broader movement to make research more accessible. When the policy was put into effect in 2015, most publishers had introduced one or more open access journals to their portfolio, with many entering the field with OA journal offshoots of name brands and testing the waters with hybrid models. Today, virtually every journal offers an OA workflow.

However, despite this progress, there is more to be done. To meet the urgency demanded by today’s greatest global challenges, there must be a culture shift in research publishing that ensures the prioritization of equity and access over prestige and personal interest.

The foundation’s 2025 policy refresh aims to expand support for a publishing ecosystem that prioritizes equity and inclusion. It builds on a decade of lessons learned and emerging best practices to address inequitable publishing models that undermine foundation-supported and other critical research from having an immediate and direct application towards new practices, methodologies and policies.

• 1) In the 2025 policy refresh, the primary changes are: all Gates-Foundation-funded research must be posted as a preprint or uploaded to a repository like PubMed Central. These funded manuscripts must be immediately available with a CC BY license and accompanied by a data availability statement. Grantees must acknowledge the foundation and include their grant number.
• 2) All journal articles submitted after 1 January 2025 will no longer be eligible for publishing fee payment by the foundation.

Why we are requiring preprints

The first preprint server, arXiv,¹⁸ was launched in 1991 with the vision of capturing the energy around the creation of the internet. However, the concept of a preprint began in the 1960s as a way for scientists to share information more quickly. While preprints have yet to undergo a formal journal peer review process, they are quickly and widely available for public review. Authors are able to share their work openly and rapidly at no cost to them. Citation practices are evolving so that preprints can be cited in grant applications and grant reporting, which is especially important for early career researchers who often cannot wait on lengthy publishing timelines to build their career. The speed of sharing research can be critical as was seen during the Covid-19 pandemic when the traditional publishing process struggled to adapt to the pressing needs of rapidly sharing information to find interventions. There are a multitude of global issues to solve that are as critical as the pandemic and the publishing process has not evolved enough to meet the needs of the next crisis.

Out of the various publishing models and practices, preprints servers demonstrate a high factor of equity as authors do not need to use funding to publish, the preprint servers are separate from journals so the research can be assessed on its own merits, and authors retain full copyright and control of the research. Many are calling for a ‘Publish, Review, Curate’ model¹⁹ and preprints can lead the way.²⁰ Shifting the OA requirement from the VOR to the preprint allows for reinvestment of millions of dollars in APCs funds into other more equitable business models. Other funders, the Chan Zuckerberg Initiative, Aligning Science Across Parkinson’s, Alex’s Lemonade Stand and others²¹ have paved the way for preprint policies.

Supporting preprint infrastructure

Building upon the success of the Gates Open Research model, the foundation, in continued partnership with F1000, soft launched VeriXiv,²² a verified preprint server in 2024. Preprint submissions to VeriXiv undergo a rapid and rigorous set of checks undertaken by F1000’s in-house editorial team before being published as a preprint with the status ‘awaiting peer review’. Apart from these technical and ethical checks, there is no traditional role of an Editor (or Editor-in-Chief) to decide on whether to accept or reject the preprint or to oversee the peer-review process. The reasons the foundation launched another preprint server include:

• GOR had an established integration with the foundation’s grant management system to confirm grant number metadata
• an expansive scope to fit all the disciplines in which the foundation funds
• availability of all article types to allow for grantees to easily share their outputs openly
• provide grantees an option with increased validation checks to alleviate any concerns about early sharing
• built-in data workflows to assist grantees with open data management.

Rationale for ending support for publisher APC payments

OA publishing has gained its own momentum with a greater number of mandates from funders and institutions. Almost every publisher now offers an OA option with the vast majority leveraging an APC business model that shifts costs from readers (pay-to-read) to authors and their institutions (pay-to-publish). For many researchers, the barrier to participation in the research enterprise has shifted from an inability to read publications, which initially triggered the OA movement, to an inability to publish. APC prices have risen faster than the rate of inflation, with name brand journals commanding over US$12,000 per article. As Nigerian researcher Dr Seye Abimbola wrote recently²³ about APCs, ‘we have swapped one form of inequity for another’. Indeed, several notable journals have had their editorial board resign en masse²⁴ due to disputes of unfair and inequitable APC pricing. APC prices are also not adjusted based on the authors locality,²⁵ and waivers are sometimes given for those who cannot pay. But waivers are not a scalable solution; the global community needs inclusivity not charity.

Overall, the transition of subscription journals to fully OA has been unsuccessful. Publishers have introduced the hybrid journal model with both subscription and APC revenue, which has persisted for several decades. Only a small percentage of subscription journals have been able to flip to fully open access after enough authors opted for and paid for OA. A core component of Plan S was ceasing APC support for hybrid, and cOAlition S funders stopped paying these fees in January 2021 as described in this brief.²⁶ The decision centers on the practice of ‘double-dipping’, the increased cost to publish in a hybrid journal versus a fully open access one, and how the model has stalled adoption for other OA models. The foundation was one of several cOAlition S funders who supported the payment of hybrid APCs through the transformative journal program. Unfortunately, this program had little success in achieving its aim of flipping journals to fully OA. The latest report²⁷ further solidified the decision to end this program at the end of 2024. It is notable that a large portion of journals receiving foundation APC support will no longer be eligible for APC payment beginning in 2025, regardless of the overall policy refresh.

Per article pricing has historically been benchmarked on what came before, namely subscription prices to large institutions, and it is hard to ascertain if this reflects the cost of producing the article. The APC model includes the services provided by the entire journal profile (including rejections) and does not adjust for the specific work needed for an individual article. Efforts to gain more transparency in pricing, as tied to publisher services, have been unsuccessful. A journal comparison service²⁸ was created by Plan S to provide some transparency to APC pricing and help authors make informed decisions. Most publishers refused to participate, citing anti-competitive law.

It was with all these factors in mind that the foundation was motivated to end its support for APC payments, aiming to facilitate systems change in research publishing toward more equitable and inclusive practices, in turn supporting the development of more impactful scientific breakthroughs that empower the world’s poorest to transform their own lives.

Supporting alternatives to the APC model

Alternative models have been introduced slowly but have not gained much ground in replacing the APC. Other options, such as waivers and price caps, have a high administrative burden on publishers, funders and grantees. On the other hand, it should not be overlooked that the burgeoning diamond OA model has been successful for many journals. The Subscribe to Open (S2O) model is quickly gaining success as an ideal mode of achieving OA without relying on APCs.

Since the launch of the original OA policy in 2015, the foundation has thoroughly tracked compliance and APC spend to monitor the trends and real experiences for grantees. This data has been made available to the community through OpenAPC.²⁹ Grantees have many no-cost options for sharing their research, including preprints and repository deposit of accepted manuscripts. PubMed Central compliance for grantees is currently 89%. Requests for foundation payment for APCs have fallen to 45% of grantee published articles, yet the total annual cost is increasing. APCs fees on average cost US$3,000/article and can cost as much as US$12,000/article.

This leads us to conclude that most grantees are finding alternative options to pay for OA, including co-authors, institutions and other funding sources. Working with the University of Washington’s START program, the foundation produced a comprehensive range of options for managing APCs. The research revealed that increasing APC median year over year is outpacing inflation.

Conclusion

In looking back at a decade of OA policy and practice, the foundation will roll out its refreshed OA policy in January 2025 to address the inequitable practices perpetrated by the scholarly publishing community and to drive broader change in the field. OA is no longer the primary goal, but a milepost in the journey to open science practices that foster early sharing, broad dissemination and effective reuse of funded research.

As a community of fellow funders, institutions and OA advocates, we have ignited a tangible change in research dissemination and scholarly publishing, but our journey toward equity in knowledge remains unfinished. As part of our mission to promote equity for all people around the world, we must work toward a more inclusive future in research dissemination.