The humanities have come late to open access (OA) – for a variety of reasons. The movement began with digital artists who wanted to share and ‘mash up’ material. It was greatly augmented by the recruitment of scientists, some of whom were furious at being held to ransom by big corporate publishers, who had taken over the publication of many of their journals and were extracting what some considered to be inordinate profits from them. Scientific data seemed to be (or seemed to ought to be) no one's property – and it offered prospects of much enhanced value if made accessible to other scientists and entrepreneurs (especially in the biomedical and tech fields) in a timely fashion. Governments and philanthropic funding bodies have more recently woken up to the latter attractions and sought to attach OA mandates to their funding offers, seeing open access both as a public good and as a contribution to the ‘growth agenda’. Publication costs formed such a small part of scientists' large research budgets that OA costs could, it appeared, be almost silently and painlessly absorbed. The rules and norms of open access were accordingly forged early on in a series of one-sided proclamations, with insurgent scientists' interests paramount.

Little account was taken of the humanities in these developments – and humanists for their part took little account of them. Our most important work does not appear in journals, but rather in books. (It is possible to make OA books, but it is a lot harder and more expensive, and because books feature so minimally in scientists' publishing protocols one could say that scientists, too, came late to open access, so far as books go.) Our journals are on the whole not owned by rent-seeking multinationals, and what profits they earn very often go to support the under-funded activities of learned societies, rather than to shareholders. Profits tend to be slender because our journals cost less to buy and cost more to produce – they publish fewer, longer articles, rarely with multiple co-authors (thus with fewer authors in total to charge for publication fees); they publish a large amount of commissioned material (reviews, comments, roundtables) and arguably require a great deal more editorial work on all of their published output. Furthermore, our ‘data’ is already mostly available on open access – in printed texts or in archives. Where it is not open, we do not own the rights – we are given limited permissions to make use of private archives, images, music, printed works in copyright. And our articles are not reports of data – they can even be considered literary works in themselves. They can, of course, be mashed up, but both law and ethics have generally held that they must not be without the explicit consent of the author, among whose ‘moral rights’ to their text is the right to preserve the integrity of the work. This need not impair free dissemination of the work, but it does put a check on ‘derivative use’, which changes the work (often in ways that cannot easily be discovered) while still attaching to it the original author's name and the authority of their words.

“Little account was taken of the humanities in these developments – and humanists for their part took little account of them.”

This mismatch between the protocols of the open access movement and the norms and practices of the humanities suddenly became glaringly obvious when the UK government decided in 2012 to throw its weight (and its mandates) behind open access to publicly funded research. It did so on the basis of the ‘Finch Report’, produced by a group of 16, predominantly made up of scientists and social scientists but with zero representatives from the humanities. Though the Finch Report included many cautions about the uncertain applicability of established OA protocols to the humanities, none of these cautions was regarded in the resulting policy put forward by the Research Councils in early 2013. Although a transition period of indeterminate length was allowed for, the goal announced was to move all Research Council-funded research to ‘gold’ open access. Gold OA, in which the author (or their sponsoring institution) typically pays for publication, aimed both to placate the commercial publishers (who arguably may even prefer this kind of guaranteed revenue stream to subscription income) and to introduce a market in article processing charges (APCs) that might eventually reduce universities' overall spend. In any case, universities would find themselves with growing responsibility for deciding which articles were worth publishing and where. Not many scientists liked this part of the package – nor even, to their credit, many managers – but for amply funded research such rationing did not appear to be too directly threatening. One grimly technical provision of the Research Council policy had the potential to polarize scientists and non-scientists more dramatically. This was the requirement that gold OA research be published under the Creative Commons CC BY licence – the original ‘mash-up’ licence. This requirement derived from a combination of open access idealism – ‘true’ OA, enthusiasts argued, put no restrictions at all on what could be done with published research – and government's growth agenda, which prioritized reuse of publicly funded research by small- and medium-sized enterprises. For reasons that would only gradually become clear, an apparently innocent licensing requirement based on the practices of scientists and digital artists hid all sorts of booby-traps for humanities scholars.

“… universities would find themselves with growing responsibility for deciding which articles were worth publishing and where.”

To its credit, the humanities community woke up pretty quickly to the pitfalls – and potential – of open access in the months after the publication of the Finch Report. It was apparent that humanities scholarship had little to contribute to the growth agenda – especially as our most commercially valuable work (published in books) had been deliberately excluded from the policy, again largely to appease the commercial publishers. Furthermore, in any internal university competition for publication funds, humanities scholars were likely to find themselves at severe disadvantage to researchers in more obviously revenue-bearing disciplines; the effects of a market for publication, adjudicated by university managers, were likely to be catastrophic for academic freedom, not to mention the quality of journals which, as we have noted, tend to cost more to produce in the humanities due to their longer, more intensively edited articles. Many humanities scholars are not even attached to universities, or are only sporadically attached, and any business model which relied on ‘pay to say’ threatened to disfranchise entirely a large body of valued early-career and independent scholars. And, of course, most humanities scholars work outside the UK, and it was far from clear that international scholars would have access to such funds, or indeed that international journals would accept them. Nor was the sharing of data and research findings at top speed such a priority for disciplines that were more diverse, contemplative and argumentative, and based on data already open or not owned by the researcher.

“… the humanities community woke up pretty quickly to the pitfalls – and potential – of open access …”

On the other hand, although the humanities are not principally valuable for their contribution to GDP, they are quintessentially educational enterprises, and I personally have encountered few if any humanities scholars who doubt that free public access to our scholarship would be a very great prize indeed, if it could be won without sacrificing academic freedom and quality. Emphasis in the immediate responses to the Finch Report and the Research Council policy was thus placed on finding forms of OA publication that would best suit the needs and interests of humanities scholarship. A coalition of 20 learned societies in the arts and humanities, the Arts and Humanities User Group (AHUG), had already by March 2013 formulated a strong statement in favour of open access and indicating the terms on which this could be made to work for their own disciplines.

Special weight was placed on the need to develop a viable ‘green’ route to open access which did not require APCs and thus did not discriminate against scholars with reduced or no access to institutional funds, and did not threaten academic freedom by giving managers the final say in what and where academics publish. The Research Council policy had offered as a transitional measure a green route, whereby articles would be withheld from open access for a short embargo period, which might allow journals to publish them for free and still reap some subscription income to support their costs. AHUG asked for this green alternative to be made a permanent feature of the policy and to be given parity with gold OA.

Then it remained to determine the terms under green OA in which humanities journals would be able to sustain moderate and responsible subscription income. The Research Council policy, again driven by science lobbyists, had denigrated the green route as an inadequate and temporary alternative, and deplored anything but minimal embargo periods as depriving the public of ‘its’ research (although of course the public had paid for the research but not – yet – for its dissemination). The embargo periods set in the policy were six months for science, 12 months for others, with extensions to 12 and 24 months where funds for gold OA were ‘unavailable’, whatever that means, although again the Research Councils made clear that their goal was a maximum six months' embargo for all. No rationale was given for these stipulations. Since nearly all green OA experiments hitherto had been undertaken with science publications, there was no evidence about what lengths of embargo were necessary to sustain moderate and responsible subscription levels for humanities journals. We are often asked for evidence of this kind, but it is difficult to provide evidence about something that has not yet happened. Even the Finch Report urged caution on this front repeatedly. Yet under pressure from OA enthusiasts, the Research Councils became wedded to their original stipulations, without providing any evidence of their own as to why they had chosen the embargo periods they did.

“Most humanities research is not downloaded from journals but from archive sites …”

Evidence has since emerged that longer embargo periods do not reduce the public utility of humanities research as they might for some science research. Measuring utility by means of article ‘half-lives’ (how long it takes for an article to achieve half its lifetime downloads), a recent British Academy study has found that only biomedical research consistently registers half-lives as short as the 24-36 months cited by Finch to justify a six-month embargo period. Computer science may well fall in the same range; other science articles seem to have a somewhat longer half-life. But most humanities articles fall in the 50-60 month range. That would seem to justify a 2:1 ratio such as the 6/12 and 12/24 month provisions of the Research Council policy. But these half-lives are measured only with data from journals' websites. Most humanities research is not downloaded from journals but from archive sites such as JSTOR. One leading journal with which I am associated gets 80% of its downloads from JSTOR. And the same BA study found that JSTOR half-lives for humanities and social science articles average at 20 years. In other words, much less public utility is lost from humanities publications even by quite long embargo periods.

And of course we still have no data on how long an embargo period is necessary to sustain the moderate and responsible subscription income necessary to fund high quality editorial product. Based on the well-established norm of a three-year embargo on humanities journal content before that content is available on JSTOR, a group of history journals associated with the AHUG statement proposed a three-year embargo as a starting point, and regular monitoring thereafter to see whether that limit needed to be extended or could safely be contracted. This proposal has often been ridiculed by OA enthusiasts who, it could be claimed, know little about humanities publishing, but since the half-life evidence points to very great differences between humanities and science publishing, and there is little evidence as yet about the impact upon journals of embargo lengths in the humanities, it seems prudent and flexible. However, it has not yet been recognized by the Research Councils. All that has been conceded – on this and on other points – is a regular ‘review’. We must hope that ‘review’ does not mean, as it often does in government pronouncements in my experience, a swift kick into the long grass.

“… the half-life evidence points to very great differences between humanities and science publishing …”

A final point of concern in the AHUG response focused on the CC BY licence. In some respects, humanities scholars are more favourable to generous licence terms than scientists. We would prefer that the OA version be the version of record, not the unformatted and edited ‘accepted manuscript’ or ‘preprint’ (because formatting and editing is an important part of the final product). We have no objection to text mining (mechanical analysis of our work in bulk), nor to data mining (although as noted above, our data is not normally under our control). But ‘derivative use’ is a sticking point. CC BY permits, indeed encourages, the creation of ‘derivative works’ out of the original – that is, anyone can take the original words and manipulate them in any way they wish, without making clear how they have done this. Much confusion surrounds the fact that CC BY is an ‘attribution’ licence – it requires acknowledgement that the original work has been altered, but not how: thus it is enough to say, ‘this is a work by Simon Schama, adapted by Peter Mandler’. Often it is very difficult to work out how the work has been changed, and meanwhile the new work acquires authority not only from the name but from the words of the original author. There are lots of reasons why humanities scholars – and indeed many scientists, who when given a choice most often prefer a ‘non-derivative’ licence over CC BY – have promoted other CC licences that facilitate open access but not this kind of reuse. For one thing, we do not have full ownership of our texts ourselves – we use others' words and images, often by permission. For another, we have our own norms of how best to incorporate one work within another – e.g. by quotation – which derivative use denies. Most important is our moral right (long acknowledged in law and ethics) to protect the integrity of our work. By all means read and disseminate our work free of charge, but do not change it as you are doing so – write your own work.

“Most important is our moral right (long acknowledged in law and ethics) to protect the integrity of our work.”

While the Research Council policy remains unchanged (pending ‘review’), the critiques of AHUG and others have been listened to more attentively by the university funding councils (HEFCE, etc.), which have recently released their own OA policy for the next Research Excellence Framework (REF). Green is indeed given parity with gold. Scholars without institutional affiliations at the time of publication are given some exemption. A major and important (if ambiguous) exemption aims to keep international publication open to UK scholars and to preserve their academic freedom to publish what and where they wish – journals which do not offer OA or have longer embargo periods are permitted where they are ‘the most appropriate publication for the output’. AHUG's favoured licence, CC BY-NC-ND (which permits free use but not derivative or commercial use) – also the most favoured by scientists when they are given a choice, and by other bodies such as the US Social Science Research Council – is accepted as valid, though grudgingly. All in all, this is in my opinion a pretty satisfactory outcome for the humanities, and a significant one, as many more of us are funded through REF than by the Research Councils. It proves the value and efficacy of making a loud noise, especially when funders and policymakers are inclined to take science norms as research norms in default of loud noises to the contrary. Concerns remain. We are leaping into the dark so far as embargo periods go. Here, RCUK policy and the REF policy are aligned; elsewhere, despite their authors' protestations to the contrary, they are not, storing up great potential for confusion. Heaping yet more behavioural nudges on academics through the REF (in return for less money) is not making the UK higher education environment more attractive in a global market. Most problematic, probably, is the way in which new policies such as these are likely to be mis- or over-interpreted by risk-averse managers, leading to perverse consequences that in such circumstances rarely benefit the humanities. Given how many managers appear to have refused to countenance the carefully drafted provisions for double-weighting of monographs in the humanities for REF 2014, the carefully drafted OA exemptions (and some confusingly countervailing incentives) for REF 2020 are likely to produce more confusion and caution and, for all the good intentions, do the damage to academic freedom that the funders themselves seem earnestly desirous of avoiding. Who decides what is the ‘most appropriate’ journal? Who decides what kind of licence the author should sign? Managers are unlikely to leave these decisions to authors.

The answer to these objections to government policy is not for humanities scholars to turn against open access, but for them to redouble their efforts to ensure that the public gets access to our research in forms that sustain rather than undermine the quality of that research. It will be a long and difficult haul, but…the prize is great.

“The answer to these objections … is … for humanities scholars … to redouble their efforts …”