A progress report on opening up scientific publications

Could you explain to us why the CNRS adopted a dynamic policy on opening up scientific publications?

Alain Schuhl: Open science means making research results accessible for everyone in compliance with the idea that science is a common good. As regards scientific publications specifically, the open access movement began in the early 2000s as a response to scientific publishers controlling the dissemination of work produced by scientists in articles that could only be read on a subscription basis.

Opening up scientific publications and giving access to their associated data means we can encourage collaboration and gain in collective efficiency because it makes public research results that are more transparent and reproducible. This reinforces the bond of trust between science and society which is particularly important because scientific results now make up the raw material to shape thinking about the world of tomorrow. The COVID crisis is a revealing example. A vaccine was produced in record time and the openness of scientific results enabled this feat.

The Roadmap for Open Science published by the CNRS set out the target of achieving 100% of publications in open access. How is that going?

A. S. : It's important to differentiate between publications by CNRS researchers and those from units under joint supervisory authority.

In the first case, since 2022 we have achieved a rate of 95% of publications in open access. This was made possible by our annual campaigns of researchers' activity reports in which scientists report publications to be considered from HAL.

For publications from research units under joint supervisory authority, we've reached a stable 80% threshold since 2021. We're aware we still need to work with our partners in higher education to increase this rate.

This is not without cost though as publishers are gradually replacing subscriptions with article processing charges (APCs) for publishing in open access. Is this transformation a good thing?

A. S. : Subscriptions are not being replaced by publication costs which in fact represent an additional cost. Not only are APCs on the rise, they're unrelated to the service provided by the publisher. In fact, these are prestige costs that are mainly linked to a journal's reputation. To sum up, research is now being paid for three times – the first to be produced, the second to be published and the third to be read. As I explained last year clearly this is an unsustainable business model.

The result is that reading and publishing scientific literature both depend on the financial resources of operators or countries which only increases existing inequalities even more. Worse still, this 'author-pays' model twinned with the pressure to publish has led to the emergence of predatory publishing. This development of 'junk science' is undermining people's trust in science. Here again the COVID crisis is a telling example because anti-vaxxers used this as an argument.

You mentioned that subscriptions are continuing. Could you tell us more about this? What is the CNRS policy on this point?

A. S. : Publishers will carry on offering subscriptions as long as all scientific production is not immediately available in open access. In the light of that, the CNRS will renew certain subscriptions where appropriate.

But if it's not too inconvenient then the CNRS does unsubscribe from certain resources like, for example, Springer journals in 2019 and from the Scopus database more recently. We don't take such decisions lightly - the subject is discussed collegially with our Institutes. Unsubscribing also sends a strong political message to publishers that we would like to withdraw gradually from their system. To have a credible negotiating position we need to be prepared to do without their services.

In fact, other subscription models do exist such as 'Subscribe to Open' or S2O. The idea of this is to support a publisher in opening up its journals by continuing to subscribe to these. Then, once a sufficient number of subscribers is reached, the journal that adopted this model goes into open access without APCs. This means that whether S2O is a success or not will need to be evaluated over the long term according to the number of subscriptions received from libraries. The CNRS is supporting this initiative by subscribing to the journals published by Annual Reviews and mathematics journals published by the Société de Mathématiques Appliquées et Industrielles .

The CNRS recently signed an agreement with Elsevier through the Couperin consortium which the organisation belongs to. This sort of contract has both a subscription component and an immediate open access publication component. Isn't that contrary to the position your institution has adopted?

A. S. : For a few years now 'transformative' agreements combining reading and publication have been developing. Certain publishers who are completely in open access also offer simple publication agreements given subscriptions with them actually don't exist.

Whether we're talking about agreements combining reading and publication or simple publication agreements the main idea is to centralise APC expenditure to control it. This means researchers don't have to pay APCs individually. Several types of publication agreement are available - either with a prepaid stock of APC or a fixed price for unlimited publication. Whatever the case, here the CNRS is opposed to this idea with for-profit publishers because it speeds up the transition to the author-pays system we reject. When these agreements expire, prices go even higher.

But there's also the case of Elsevier which accounts for about a quarter of French scientific publications. The Couperin consortium's management board adopted a negotiating mandate despite the CNRS voting against this and now Elsevier and Couperin have concluded an agreement for the 2024-2027 period. Although this doesn't correspond to our policy, the CNRS has a duty of solidarity with its partners in higher education and research.

The CNRS encourages researchers to turn to virtuous publication models. Could you give us an overview of the possible options available for scientists?

A. S. : The first and most immediate option is to use preprint servers and open archives. The latter mean research results can be archived on a long-term basis. The CNRS was a pioneer in this area with the creation in 2001 of the HAL open archive. By 2023, HAL had racked up a tally of nearly 100 million consultations worldwide.

As well as open archives, the CNRS supports bibliodiversity which refers to the plurality of publishing stakeholders, forms of publishing and scholarly communication languages. So by 'virtuous' publications, we mean those that are free for authors and readers alike. But this does not mean these so-called 'diamond' publications are free because they all require financial support from public institutions.

That's why alongside the HAL open archive the CNRS provides financial or HR support to preprint platforms like arXiv or bioRxiv, open access publication platforms like SciPost, the Centre Mersenne, OpenEdition and peer review platforms like EpiSciences and Peer Community In. This support was paid for by the savings from unsubscribing to Springer journals in 2019.

And yet, not all disciplines have virtuous journals.

A. S. : We're aware a lot of work remains to be done to develop a sufficient offer of these. A paradigm shift in the way scientists publish is taking place and that takes time, like any cultural shift.

We are in the midst of a period of transition and the end goal will be for there to be viable options for researchers in all disciplines to publish in open access.

To achieve an objective like that presupposes a change in the way research is assessed individually and also collectively (a laboratory, institution or country). However, this reform can only be achieved if all the right stakeholders are on board internationally. The CNRS is working on this internally by changing its assessment principles and externally through its involvement in Coalition for Research Assessment (CoARA).

People sometimes criticise open science policies for restricting the academic freedom of researchers by preventing them from publishing in the journals of their choice. What do you say to that?

A. S. : But can we even talk about academic freedom when researchers are trapped in a system that encourages them to publish in prestigious journals or actually risk compromising their future careers? Also, can we talk about academic freedom when researchers are forced to give away their rights exclusively? Transferring your rights exclusively to a scientific publisher, as commercial publishers actually demand, reduces your freedom. Conversely, publishing in open access with a Creative Commons licence opens up usage rights and therefore offers researchers freedom of action.

France has a proactive policy aimed at developing 'diamond' publishing which will require boosting public scientific publishing. However, some believe this could jeopardise private publishing or even ultimately lead to scientific publishing being under public State control.

A. S. : It's ironic that private publishers are pinpointing the risk of public control when private scientific publishing already relies on governments paying subscriptions, APCs and even public subsidies. For example, French scientific publishers actually benefit from State aid through the national programme to support French scientific publishing.

We don't aim to just reject private scientific publishing in a Manichean way or deny that publishing scientific literature comes at a cost but the idea is to reduce the budget we dedicate to this down to the right level. Until now, prices have corresponded to a measure of prestige rather than to a service provided or production costs. This is particularly the case with multinational publishers like Elsevier, Springer and Wiley who have been making obscene profits for years now.

The latest figures from the French Barometer of Open Science show diamond publishing in France will only account for 10% of the articles published in 2022 and disseminated in open access. This means there is only a very remote and highly hypothetical risk of state control. And what's more that wasn't even the objective in the first place!

The Subscribe to Open example shows that 'diamond' publishing - i.e. without APCs for authors or payments from readers - is not contradictory to the model of publishing with a private publisher as the case of the Astronomy & Astrophysics journal clearly illustrates. This is a key journal in astronomy published by the private publisher EDP Sciences which switched to S2O a few years ago.

Researchers can also apply the rights retention strategy and, if all else fails with publishers, there's still the French Law for a Digital Republic which authorises free access to publications from French public research. Scientists can distribute their accepted author manuscripts at the latest after a short six-month default embargo period (up to twelve months for the humanities and social sciences) whatever contract they may have with a publisher. Of course, publication won't immediately be in open access  but in any case this still remains an option for researchers who don't wish to pay APCs.