Open excess: remove open access burden from REF

• This HEPI blog was kindly written by Patrick Grant, the Pro-Vice-Chancellor for Research at the University of Oxford; Tanita Casci, Director of the Research Strategy & Policy Unit at the University of Oxford; and Stephen Conway, Executive Director of Research Services at the University of Oxford.
• This piece is the latest in a series of pieces on the Research Excellence Framework. In March, we discussed People, Culture and Environment (PCE); in April, we discussed small and specialist institutions; and in May, we talked about research integrity and returned again to PCE.

Whatever your opinion of the UK Research Excellence Framework (REF), it has driven important changes in how we define and assess research excellence. The inclusion in the REF of societal impact, its emphasis on quality over quantity, and plans to evaluate the quality and sustainability of the research environment all show how the UK derives value from public investment in research.

However, the latest REF proposals for Open Access (OA) — currently under consultation — depart from the generally positive direction of the forthcoming REF 2029. The proposed OA requirements for research outputs risk being unaffordable and excessively bureaucratic, and will bias the selection of outputs that we submit for assessment.

The consultation, which closes on 17 June, invites responses to technical questions concerning the OA compliance rules for publications submitted for assessment: the preferred length of embargo periods, acceptable publishing platforms, copyright licensing options, publication types that are in-scope, which publication types should be treated as exceptions to the policy, and what level of tolerance (i.e. violations of the policy) should be permitted. For the first time, an OA policy will also apply to long-form outputs such as books and book chapters.

It is easy to provide pragmatic answers to the technical questions to try to contain or selectively delay some of the excessive burden of implementation.

A more fundamental question is why we should have a REF OA policy at all.

First, the costs and the bureaucracy of implementing the expanded policy will be unaffordable. To make all research outputs OA eligible for quality assessment — the output pool from which the best are selected — involves processing and checks on upward of 15,000 separate outputs produced by Oxford’s researchers each year. The OA compliance cost for the new requirements for long-form outputs has been estimated at £20M over a REF cycle for our university alone; add up these costs across the sector and it easily runs into hundreds of millions. This estimate does not include the cost of researchers and professional services teams engaging with a convoluted, technical policy.

Second, OA costs are not commensurate with value. What does OA have to do with research excellence? Not very much: an output that is OA is no more excellent than one that is not. Arguably the REF OA agenda has driven researchers to a compliance-centric view of OA, rather than providing them with positive choices to increase the reach of their work.  

Third, there is the opportunity cost. A focus on OA detracts from opportunities to promote open research practices, which help to ensure that research is verifiable, replicated, and built upon. The REF OA agenda has had little positive effect on diversifying the range of submitted output types, different publication formats, and contributions from different members of the university community. We could take the opportunity of REF 2029 to develop more robust policies, incentives, and evaluation mechanisms for open research.

Fourth, REF OA rules unhelpfully distort the REF itself because the associated expense necessarily constrains the pool of eligible outputs that institutions can put forward. Institutions will be judged only on what outputs they can afford to make eligible. Some financial help is available from some funders, but for many outputs, especially books, funds are insufficient or non-existent. For smaller institutions, compliance requirements will depend disproportionately on institutional subsidies and will increase the risk of not having enough eligible outputs to submit.

Our national obligations to OA are more logically driven through UK funder mandates attached to individual research projects. REF should be encouraging institutions to submit their best research outputs selected from the widest pool. It would then stimulate institutions to produce globally competitive outputs, not ones that satisfy a technically complex carve-out. Are our global competitors constrained in this way? The strength of an institution’s support for OA would be much better assessed in the new REF People, Culture and Environment component, in the context of OA and open research practices.

As a sector we decried the administrative burden of REF 2021 and vowed to do better. We are fortunate to have a funder — Research England — that has embraced new and progressive REF designs for the good of the sector. The removal of an OA policy from REF 2029 will, in a simple single act, excise complex and costly bureaucracy for all concerned. It may also promote research quality and fair assessment.