Accelerating open science in physics
Daniel Keirs considers open science from the perspective of a physics publisher, including some of the steps needed to accelerate progress
Open science has numerous definitions, but it includes making science more accessible, reusable and transparent, increasing reach and reproducibility, and fostering trust in its results.
It’s also about cultural change and increasing the diversity and inclusivity of people and ideas. It is the right thing to do, and strengthens scientific integrity and trust, but it can also help accelerate scientific discovery by enabling more people to collaborate and share ideas efficiently and cooperatively.
Physics was one of the earliest scientific communities to embrace internet-enabled open science. The arXiv repository, where researchers in physics and related disciplines can share drafts of their scientific papers publicly before submission to a journal, was established in the early 1990s. In the last decade SCOAP3 (Sponsoring Consortium for Open Access Publishing in Particle Physics), an initiative coordinated by CERN, enabled most journals in particle and high-energy physics to transition fully to an open access publication model.
By the end of 2014, arXiv hosted more than one million preprints and has seen annual growth in the upload of new papers of more than 10 per cent in recent years – suggesting researchers increasingly see benefit in sharing their work earlier and openly before formal publication. Simultaneously, the number of peer-reviewed physics articles published in open access journals almost doubled between 2016 and 2019. The expansion in potential access to articles can support faster and wider impact of research.
In 2020, the global coronavirus pandemic has underlined the role open science can play: highlighting the value in the open dissemination of scientific results in maximising access, combined with rigorous but prompt peer review to further scrutinise and verify claims to maintain trust and integrity in the scientific record. These needs are not new this year, but the urgency of the scientific response to the pandemic has emphasised them.
Learned society publishers like us, rooted within a scientific community, must help address the need to make science more open and reliable – both in the context of the current pandemic and for the longer-term advancement of the physical sciences.
When thinking about how we do this, we must ask: What can we do to improve scientists’ ability to discover, read, interpret, share and build upon the outputs of scientific research? And how can we enable researchers to better observe how scientific claims are developed and asserted, and better understand the checks and processes a scientific work undergoes before public dissemination? All have their own challenges, which we are working to address in our own scientific community context through our ‘open physics’ programme. This is our commitment to supporting increased access, transparency and inclusivity in the physical sciences.
All about access
Access is fundamental to making science more open. Accessible research results provide a foundation for broader scientific engagement and collaboration. A collaboration may be literal, through researchers from different backgrounds recognising a previously siloed common challenge. Others may be more abstract, as open research that can be read and built upon is better able to inform and even inspire future research projects.
There are several ways scholarly publishers can improve access to research. Providing research communities high-quality open access journals in relevant fields is one positive step.
We have seen a steady increase in the size and range of open access publication venues in the physical sciences. Over the past three years we estimate the number of larger open access journals (those publishing more than 100 articles per year) rose by around 30 per cent, from 23 in 2016 to 30 titles in 2019. However, many communities are not yet evenly served by open access publication venues – particle and nuclear physicists and those in optics and photonics are better served than most other physics subdisciplines for instance. And although we have those 30 open access publications, there remain more than 500 larger publication venues for physics that do not operate on an exclusively open access model.
We are among the growing number of publishers launching fully open access journals. We have launched eight since 2016, covering a range of fields in physics, materials science, energy, photonics, environmental science and machine learning. However, to ensure researchers can publish openly and continue to choose the right venue for their research, we still enable authors to choose what is known as hybrid open access (where an article is published under an open licence in a subscription-based journal). This model will remain crucial to growth in open research for some time.
The practicalities of getting from where we are to a more open publishing, open science landscape are often overlooked, given the considerable cultural and financial obstacles.
Scientific research is a global endeavour. But moving to a fully open publishing model may erect new financial barriers to those in relatively less well-funded subject areas and geographical regions. For instance, we know researchers based in lower-income economies have fewer resources to cover the costs of open access article publication charges. Even better-funded groups in many countries may not be incentivised to prioritise funds for open access publication (especially when existing subscription-based journals may serve their needs in all other respects). To provide some shorter-term support many publishers, including us, reduce or even remove open access charges for researchers based in lower-income economies. But this model is unlikely to be sufficiently sustainable to support transition to open publishing globally.
New open publishing models that avoid author-based charges are becoming more common, albeit largely in Europe currently. Often called transformative agreements, they see institutions or institutional consortia wishing to publish their faculties’ peer-reviewed outputs openly make agreements with publishers combining the cost of open access article publication charges with the license fee for subscription-based publications. Such deals enable researchers from these institutions to publish their work openly without any direct cost to themselves. IOP Publishing currently has 12 such agreements, but challenges remain that exemplify wider barriers to a global conversion to open publishing.
Economically, there is the issue of winners and losers – meaning research-intensive institutions and countries will likely need to pay more as they take on more of the costs of open publishing models, while institutions and countries producing relatively little research would no longer have to pay for subscriptions to research journals and therefore likely pay less.
Some argue there is enough money in the system to afford a transition to open access. Whether this is the case or not, there is no current solution or global plan in place to adjust the allocation and flow of funding so it resides at the levels exactly commensurate to where research is produced. These are not intractable challenges. But they require global consensus on the goal of open science, coordinated action to build the infrastructure, and incentives to create lasting change. This will take time.
Preprints – a short-term solution?
While funding models evolve, preprints may offer a shorter-term way to make science more open. Community preprint repositories such as arXiv are recognised by researchers as an increasingly important part of the scholarly communications process. But they are not yet fully embraced by institutions and research funders, who could encourage or even require the results of their researchers’ work be shared more routinely as a preprint before journal peer-review and publication.
In recognition of preprints’ value to open science, our policy is now as liberal and supportive as possible for this early sharing of results, and we see more publishers also recognising their value and supporting them.
Access is perhaps where publishers can and should focus efforts in support of open science. But they should not ignore incorporating greater transparency in research communications to aid reproducibility and foster trust in science.
Some publishers are experimenting with more open forms of peer review. We have introduced transparent peer review, where the published article includes the full peer review history, with reviewer reports, editors’ decision letters and authors’ responses, and with reviewers electing whether to maintain anonymity. Following a successful year-long trial on several of our journals, we are considering how to extend this process to other communities we serve, so a scientific article’s journey through peer review, and the editorial decisions made about it, are clearer for all to see.
A significant route to increasing the reproducibility of research is open data sharing – making data underpinning a scientific article and any associated code publicly available. Access to research data can enable detailed scrutiny, reuse and replication of research, strengthen trust in the results, and provide the basis for future scientific innovation. In recent years, this has gained increasing traction in the research community, with publishers responding to support it in their policies and processes.
Many major research funders now require research data to be more, if not completely, accessible and re-usable. Such mandates are crucial for cultural change, but further consideration is needed into how to incentivise researchers to share data and credit others for data sharing. Scientific publishers are seeking to support researchers in complying with these new requirements for data sharing through evolving standards and systems, and with gaining more acknowledgement for doing so through encouraging and accommodating data citation. We have introduced new data sharing policies, where authors must include a statement in their article on whether data is available and where it can be found. In the months since introducing the new policies the response is encouraging, with more authors (where applicable) stating their research data is accessible than unavailable.
The importance of making data more openly accessible was highlighted recently, with the retraction of papers on a potential treatment for Covid-19 in prominent medical journals after concerns were raised post-publication about the validity of the data underpinning them. Nonetheless, the pandemic has underscored the need for the peer review process, when performed with due rigour, to be undertaken as efficiently as possible to ensure results gain further scientific validation as early as possible. It has been impressive to see publishers collaborating to increase the speed and scale at which peer-reviewed research on Covid-19 is disseminated. This has included making all relevant research freely available and discoverable as well as initiatives to coordinate pools of scientists to fast-track peer review of Covid-19 papers.
Although Covid-19 might have reinforced the value of open science, its benefits are well understood by many in the physics community, and we are a long-standing proponent. But there is still much work for all involved if we are to transition to a fully and sustainably open landscape in physics and beyond.
To that end, we will maintain an open dialogue with the physical scientists and scientific organisations we serve and continue to seek more insight into what’s specifically important to them. Over the coming year we will engage in a series of projects to speak to the global physical science community so we can contribute to more open science – more ‘open physics’ – and we look forward to reporting back.
Daniel Keirs is an associate director at IOP Publishing
