1. What’s the Problem?

Open access (OA) research publication is gaining increasing traction (for more on open access and its definitions, see Eve, 2014a; Suber, 2012; Willinsky, 2006. For more on uptake, see Archambault et al., 2014; Piwowar et al., 2018). The recent establishment of the global ‘cOAlition S’ group of funders and its ambitious accompanying Plan are only intensifying the pressure on traditional publishers to change their models to those conducive to OA, in the face of opposition from both likely and unexpected sources (cOAlition S, 2020, 2019; Debat & Babini, 2019).

One of the primary concerns among stakeholders, though, is that Plan S and most forms of gold OA in the Global North continue to normalise the Article Processing Charge (APC), over and above other non-APC business models.1 Yet we have known for quite some time that this poses a substantial challenge within disciplines that far more rarely receive project funding for research: the humanities and social sciences in particular. It also creates challenges of economic distribution that become even worse when the model is scaled to long-form artefacts with higher production cost overheads – monographs, for instance (Eve, Inglis, Prosser, Speicher, & Stone, 2017). Institutions with low research library budgets, but with substantial research output portfolios – a not uncommon situation away from the prominent research forerunners – find themselves facing a manyfold localised increase in costs under APC and their book-equivalent models. Often, if paying to make outputs openly accessible at current market rates, the serials budget at such institutions will cover only several articles, and the research book budget only a single book.

This continues to come as a surprise to the many people who have argued that we simply need to ‘flip’ library expenditure from a purchasing economy to a supply-side service system. What this neglects is the distributive function of the sales model. When a large number of participants each pay a small amount into a central pot, the pot remains large enough to remunerate publishers, while the burden remains small enough at each institutional site to make this work. When a single entity is asked to bear the entire cost of an article’s production and publication, though, this is far trickier, since the money is not at the places where research production is taking place. It is a matter of distribution. Sales models act as collective pooling systems using a coercive logic of rivalrous exclusion in order to create a ‘club good’-type benefit with exclusive access perks for members (Hartley, Potts, Montgomery, Rennie, & Neylon, 2019; Potts, Hartley, Montgomery, Neylon, & Rennie, 2017). As well as creating a proliferation of difficult-to-administer micro-transactions, APC models, by contrast, act as cost-concentration mechanisms that nonetheless achieve OA by making open dissemination the membership good for which institutions vie.

Other models have long existed. Indeed, the Global North’s elite focus on maintaining its grip on prestige in the research dissemination world has led to accusations that initiatives such as Plan S are ignoring the many systems worldwide that are already built on communal, collective infrastructures – and that avoid APCs (Debat & Babini, 2019; Packer, 1998a, 1998b, 2020; Siler, Haustein, Smith, Larivière, & Alperin, 2018). Even in the Global North, systems such as arXiv, the SCOAP3 purchasing consortium (to some extent), Knowledge Unlatched’s book programme, and more recent membership schemes from punctum books and Open Book Publishers have all recognised the value of alternative models for OA that redress this problem of distribution (arXiv, 2013; Knowledge Unlatched, 2013; SCOAP3, 2014). These initiatives sit in stark contrast to the other threat to the traditional business models of academic publishers: the rise of pirate shadow libraries such as libgen and sci-hub (Bodó, 2018a, 2018b; Bohannon, 2016).

In 2013, Caroline Edwards and Martin Paul Eve began planning a model of open journal publication that would be suited to the funding situation in the humanities disciplines (Eve, 2014b). Aside from running a successful trans- or multi-disciplinary journal – a rare phenomenon in the humanities – the primary contribution of The Open Library of Humanities (OLH) to the OA debate is economic. Overcoming the free-rider model by remaining relatively modest in scale (27 journals at the time of writing) and financial ask, OLH has demonstrated the viability of a new consortial funding mechanism for journal publication. The platform has attracted considerable interest. Along with initial, now passed, generous funding from the Andrew W. Mellon Foundation, the OLH’s ongoing costs are met by a consortium of nearly 300 academic libraries worldwide – all of whom joined since 2015 (Supporters, 2020). These range from libraries at the big Ivy League institutions down to small theological colleges, on a banded scale. Several institutions have voluntarily opted to pay a higher rate in order to support the platform. Indeed, it is from each according to their ability, to each according to their needs. Institutions continue to join on a weekly basis and they enable the ever-growing output of the platform to be gold open access, without ever asking authors, member institutions, or their funders to pay any mandatory fee.

2. Lessons Learned from the First Half Decade

As OLH approaches its fifth birthday since launch this year – and its seventh since inception – we here take some time to reflect on what we have learned from this first half-decade of a radical new economic model for open journal publishing.

2.1. Libraries are not Necessarily Classical Economic Actors

The history of libraries is contested and messy (Lawson, 2019, 2020). It is an over-simplification to say that libraries have always had a mission to provide universal access to knowledge with benign motivation but that they were simply technologically unable to do so. A history of class and racial/imperial exclusion also runs through historical library structures. Nonetheless, despite competitive market pressures in the academic world – introduced under the often-nebulously defined political paradigm known as ‘neoliberalism’ (Brown, 2015; Davies, 2014; Docherty, 2011; Holmwood, 2011, 2013) – the most usual view held by those working in academic libraries is that it is their role to maximise access for their patrons. Many see open, online dissemination as the logical outcome of that mission, to expand beyond the purview of a local student and researcher constituency and to reach out more broadly into societal access. Indeed, many of the assertions about the liberal humanist benefits of higher education in democracy seem lost without such broader access to research, particularly in countries where access to university education itself is stymied by fees.

The belief that it is the mission of academic libraries to broaden access to the widest constituencies, though, is powerful (not to mention that it seems a good, egalitarian belief). We think that, from our experience of asking libraries whether they would continue to pay – even when there is no direct benefit to their constituents beyond more work being generally open access – this belief is backed by an economic willingness, with some bounds on its scalability. Of course, there is an element of self interest: if one’s local constituents are using the resource in question, then one may be more likely to buy into such a model. This is the same gamble made by recent initiatives such as The Global Sustainability Coalition for Open Science Services (SCOSS): that the threat of losing an open service that is used by one’s constituents, through economic insolvency, could be as strong a motivator for library investment as was rivalrous access (Lewis, Goetsch, Graves, & Roy, 2018; The Global Sustainability Coalition for Open Science Services (SCOSS), 2020). This is akin to the logic that, if one does not pay one’s taxes, there will be no roads of a uniform standard.2 Academic libraries are able to behave collective-economically, so long as a demonstrable good arises from their investment.3

2.2. Implementing Consortial Funding Models Requires Much Marketing Labour

The economic willingness of libraries notwithstanding, it is far too easy to assume that financial support for open digital infrastructures is simply a matter of ‘build it and they will come’. It is also frequently assumed that ‘marketing’ in the journal world is not required for OA publications, because they are openly accessible and, therefore, everyone can see what is being published at the venue.

This is not the case. Persuading our supporting libraries to join and raising awareness among the library community of the platform required extensive work. We have always had a full-time staff member working on marketing and outreach. This includes digital marketing activities (social media, targeting free-rider institutions whose authors have published on the platform with requests that they join, emails, invoicing), real-world presentations at pertinent events, graphic design work (for posters, cards, and other awareness-raising materials), policy outreach work to funders and governments, and general networking.4 The person performing this role needs to be charismatic and must believe in the cause. They should also be rooted in the scholarly community and not a mere technocratic advocate. Above all else, they must be able to explain the offbeat economic model in terms that are understood by libraries. They must also articulate the benefits in a way that shows them to be congruent with the perceived mission of libraries, as set out above.

All of this is to say that scaling from zero to three-hundred supporting institutions took a lot of work. Certainly, we were aided by the use of intermediaries for billing and negotiation purposes. Jisc Collections in the UK and LYRASIS in the Northern continental Americas remain core partners for our efforts, centralizing payments and facilitating communications. The fact remains, though, that we have given in-person talks at the majority of our supporting institutions. Such extensive international travel and speaking engagements have significant resource and personal health implications for those working on these projects. There are also climate considerations with international travel, yet it remains our experience that speaking to people, in person, rather than (ironically) by electronic mediation is more persuasive.5

2.3. Governance and Administrative Overheads

It is important for the governance model of, and for trust in, OLH that corporate acquisition is impossible. Indeed, recent acquisitions by major publishers of entities that were thought to be ‘good actors’ have caused substantial alarm in the library community (Bond, 2017; Elsevier, 2017; Fitzpatrick, 2020; McKenzie, 2017; Page, 2020; Schonfeld, 2017). Hence, although OLH is established as an independent ‘company limited by guarantee’ under UK law, it is also an organisation registered as a charity. Its objects are educational and funds paid to OLH can only be used for the advancement of its mission to disseminate scholarship openly for educational purposes for the public benefit.

This charitable status comes with some advantages, most notably that OLH does not have to pay UK corporation tax. It also means that OLH cannot be taken over by a for-profit corporate entity (a merger with a not-for-profit entity, such as a UK university, would still be a possibility). We have bound ourselves to the Odyssean mast in case of temptation by for-profit sirens. However, being registered as a charity also carries substantial administrative overheads. We must act with total propriety and ensure that every financial decision is taken to benefit our charitable mission. We must file all requisite annual company returns – as would any business – but also complete charity commission declarations. This includes a breakdown of the locations in the world from which we have received income, by percentage of annual turnover, and specific details of separation of powers and legal oversight within the organisation. We have a set of Trustees who govern the charity, and two co-CEOs who act as the Management Executive on a day to day basis.

In addition to legal reporting duties and formal corporate oversight, we also have a responsibility towards our library supporter members. These members are responsible for voting on any new journal admissions to the platform that we believe would necessitate an above-inflation price increase in future years. However, coordinating a vote among 250+ institutions is an extremely challenging matter. The main problem we have faced, here, is that emails asking people to vote go missing (to spam folders), to the wrong person, or are left unactioned. This can result in strong percentage support for new titles among those institutions that voted, but often on lower turnout than we would like. In turn, this means that institutions then face a shock at the time of renewal. Librarians are busy, so this is not surprising, but it carries an important governance lesson for those seeking to coordinate between a large number of voting actors.

Further, though, many institutions ask us for metrics on an individual institutional basis. This is substantially problematic, both in terms of theoretical grounding and practical implementation. The problems are twofold. First, libraries that ask us for readership statistics do not understand the challenges of measuring this in a situation without authentication. We could do some form of geolocation, but this would then only capture on-campus usage, and very roughly so. In short: because we cannot with any accuracy determine the institution to which a reader may or may not belong (and have little desire to do so for reasons of privacy), OLH articles may appear to fare worse compared to paywalled resources that can accurately measure this and report it back to paying organisations. Second, libraries that ask us for authorship statistics are conceiving of the model as though it were one of APCs. This is a theoretical misunderstanding of the economic model. This misconception may be higher in the case of institutions who join after we point out that their authors are publishing in OLH titles. Such institutions appear to conceive of OLH as a retrospective waiver model, in which the default is to waive all APCs unless an institution is able to pay. This is not what we intended, but it does seem a powerful driver of new signups.

Nonetheless, across both readership and authorship metrics demands from libraries, we face a substantial challenge in delivering such statistics. This is because we are, in some senses, a meta-organisation. We have publications that span multiple platforms: at Ubiquity Press, at Liverpool University Press, and on our own Janeway software. Extracting such information, in a timely fashion, from third-party organisations can be challenging as, even with COUNTER-compliant metrics, the platforms are not interoperable. It is often hard, also, accurately to determine authorship information from university sources. Bad metadata provided by authors, including outdated institutional affiliations, mean that we must perform loose queries. A good example of this are the three entries: ‘Cambridge University’, ‘The University of Cambridge’, and ‘The Massachusetts Institute of Technology, Cambridge, MA’. Even with a loose query for ‘Cambridge’ – again, spread between multiple organisations who must perform the lookup – we can end up with entities that are radically different from one another and this must all be checked manually. Given that we distrust author-provided metadata, requests for authorship affiliation are usually fulfilled by using Google’s ‘site’ operator to search and then by manually verifying that the affiliation is correct. A future implementation of an organisational identifier, such as GRID, could help with this problem, but it will not eradicate the challenge of coordination between providers. Again, given that there may be misspellings or inaccurate information, it is hard to estimate how good these statistics are.

All of which is to say that the demands of libraries for usage statistics – on both author and reader sides – add to our costs substantially. We would estimate that at least twelve working hours per week are given over to providing such information to supporting libraries, while trying to convey the dangers of appraising our model in such terms. We understand that library budgets are under substantial pressure to reduce expenditure and to find places that they can cut. We also understand that our model may appear particularly appealing to slash from the budget – the projected ‘free-rider problem’ – but note that this has not panned out to be the case in reality. We are also not asking institutions not to make a judgement on our value proposition. We remain extremely competitive in our costs and prices.

However, we are concerned by the growing increase in demands for transparency and metrics from funding initiatives such as Plan S. It can feel as though we will need, at some point in the near future, a full-time member of staff just to comply with reports on our financial situation and on readership and author metrics. This will, of course, raise the prices that we have to charge to stay afloat.

2.4. Complex Tax and VAT Considerations

Consortial models such as the one we run are subject to complex arrangements with respect to Value Added Tax (VAT). In the United Kingdom – and the European Union – services and most goods are subject to 20% VAT at the point of sale. Under UK law, VAT is applicable when a transaction meets the following four criteria:

  1. it is a supply of goods or services
  2. it takes place in the UK
  3. it is made by a taxable person (someone who is, or is required to be registered)
  4. It is made in the course or furtherance of any business carried on or to be carried on by that person.

According to Her Majesty’s Revenue and Customs (HMRC), ‘In order for a supply for a consideration to exist, there must be at least two parties and a written or oral agreement between them under which something is done or supplied for the consideration. Also, there must be a direct link between the supply and the consideration; the supplier expects something in return for his [sic] supply and would not fulfil his obligation unless he [sic] thought that payment would be forthcoming’ (HMRC, personal communication, April 24, 2015).

Despite the above remarks on how we believe that some institutions misconceive of OLH as an APC-driven service with a default of waivers, our membership agreement clearly states that ‘members should be aware that they are not paying for their own authors to publish, nor for a subscription, but rather to build and sustain and govern a collectively-funded platform that returns extremely good value and that could not exist without such investment’. We believe that OLH, is, thus, infrastructural and not providing a direct service to those who pay.

HMRC agreed. They wrote that:

VATSC50000 provides examples of when a payment is made but there is no supply for VAT purposes, such as donations, grants and project funding. This explains that where any benefit received by the payer in return for the payment is incidental, the payment is not consideration for the supply of such benefits. The primary purpose of the OLH membership fee is to fund the open-access to academic journals. The membership benefits of a seat on the Board, yearly reports and listing on the OLH website, are the only benefits provided and are incidental to the primary purpose; it is not the aim of the members to obtain such benefit. It is, therefore, HMRC’s view that the membership fee is not a consideration for the benefits, or any other supply, and is outside the scope of VAT (HMRC, personal communication, April 24, 2015).

This brings, again, mixed fortunes. For institutions who cannot reclaim VAT, this lowers the overall membership cost of participating in OLH quite substantially (by 20%). At the same time, though, it increases our internal costs by 20% as we cannot reclaim VAT (we do not have enough VAT-able activities to reach the threshold for registration).

Further, the fact that no supply exists creates some havoc when interacting with state-funded institutions that insist on competitive tender processes with direct benefit. Some German institutions, for instance, have reported that they are not allowed to pay for goods that do not confer a direct benefit to their institution. In short, governmental oversight policy appears to create obstacles to take-up of our model. We have circumvented this through diversifying our revenue stream into channels that are VATable under supply so that organisations can purchase this with direct benefit to their universities, while we use the revenue to cross-subsidise the general model. We hope that future government policy around expenditure in this area will recognise, though, the benefits of collectivity in funding publications infrastructures for the public good.

2.5. Janeway, Diverse Revenue Sources, but Staying Open

At our launch, we worked entirely with Ubiquity Press. This was a wholly positive experience; their low rates, combined with no requisite up-front investment, make this organisation an attractive choice for new presses – as seen in the recent raft of young university press startups. We continue to have a productive relationship with this organisation. However, our Trustees felt that OLH needed to spread its platform risk. In particular, there was concern over the potential acquisition of any supplier by an entity that did not share the mission and values of the Open Library of Humanities.

While much of what we needed for an in-house publication platform could be found in the truly pioneering and excellent Open Journal Systems (OJS), we were concerned that the hiring implications of maintaining and synchronising with an external PHP codebase would have economic implications. For consider that, in the 2019 list of the most popular programming languages, PHP was almost five times less popular than Python according to Github and Tiobe statistics (Putano, 2019). This is an indicator that it is easier to hire good Python developers at this point than it is to hire those well versed in PHP. We therefore wanted to build an easily modifiable system, with a clear codebase, in a modern Python web framework to which we could migrate a portion of our journals. From this was born Janeway (Eve & Byers, 2018).

An important principle for us was that Janeway would be developed wholly ‘in the open’. The code would be 100% openly licensed and available for anyone else to re-use and deploy without charge. Again, this followed the principles set by trailblazers such as OJS. We opted for the GNU GPL Affero license to ensure, though, that anyone re-using the code – even if just in a modified server-side deployment – would be forced to make their changes similarly open. We believe that anyone should be able to use Janeway. But this is not the same as believing that they should be able to do so without a commensurate obligation to the public good to make any modifications open.

At the same time, in-housing part of our platform hosting brought with it, again, both positives and negatives. On a per-article basis, we save substantially by working directly with typesetting suppliers, rather than incurring third-party overheads. However, this has also necessitated bringing development and editorial expertise into the organisation itself, thereby incurring ongoing staffing costs. In order to cover these costs to some degree, we decided to establish a Janeway hosting service. While all proceeds are used to support our open publishing/educational mission, this also allowed us to diversify our revenue model to defray the staff costs of running our own technology. Also, as above, it allowed us to present a VATable service to institutions who would otherwise be unable to join OLH for reasons of public accountability.

2.6. Flipping Journals

Part of the OLH model that makes it appealing lies in our journal flipping programme, where we have sought to convert existing subscription titles to an open-access model without fees. This has been extremely popular and we have been unable to grow economically at a rate that would allow us to accept all applications here. Indeed, we have had to reject over two thirds of journals that applied, even when they were previously subscription. (This relativism is one of the absurdities of library budgets being measured in terms of percentage increases; an organisation that charges £100 in year one and £200 in year two registers as a 100% increase, while an organisation that charges £1m in year one and £1.1m in year two appears as only 10%, whereas organisation one receives £100 extra and organisation two receives £100,000 more.)

Flipping journals has been valuable for us in many ways. Firstly, the high-profile transfer of the editorial board of Elsevier’s journal Lingua to a new title, Glossa, published by Ubiquity Press but funded by OLH and LingOA was not only a PR success but has demonstrated to an entire discipline that open access can be compatible with quality control and prestige. It also showed up, in our personal view, a rapacious for-profit publisher in the international media. Secondly, libraries can see some degree of offsetting in the transfer of titles. That is, they no longer need to subscribe to migrated titles. This, though, can be harder in reality than in principle since titles are often bundled in ‘big deal’ packages that make individual cancellation difficult.

Our sense from this experience is that many editorial boards desire a fee-free OA model for their journals. They are sometimes put off, though, by the complexity of the transition, not understanding the contractual relationships they have with respect to a title and their current publisher, and not trusting a relatively young organisation such as OLH, even when we are backed by major funders such as The Andrew W. Mellon Foundation, the Wellcome Trust, and Riksbankens Jubileumsfond – as well as our many library supporters. Indeed, our model is good for readers, authors, editors, and appears financially sustainable. We will close with some remarks on how we might extend this good out further into the world.

3. Where Next? Flipping Societies and University Presses

We have now made thousands of articles open access under our no-author-fee system. We have demonstrated that academic libraries are willing to pay for open infrastructure as part of their mission belief. Essentially, when we asked libraries ‘would you continue paying a subscription if we made our publications openly accessible’, the answer was a resounding ‘yes’ – and we did not even have an existing journal portfolio. We have shown that academics in the humanities are not opposed to high-quality, peer-reviewed open-access publication, when there are no financial hurdles in the way (in fact, we have shown that there is a massively untapped enthusiasm for this mode). We have existed for half a decade.

We cannot change the world alone though. Certainly, there is some benefit in the limits of our scalability; scaling small, as Janneke Adema (2018) has it . But we need large publisher and intermediary, mission-driven organisations such as Cambridge University Press, Oxford University Press, Project Muse, and JSTOR to experiment with this model. Why not try contacting subscribers to a list of titles and asking them: would you continue to subscribe if we made these titles open? There is nothing to lose and everything to gain. This is the principle behind the flipping model that Martin Eve devised in parallel to the similar ‘Subscribe to Open’ model at Annual Reviews (Crow, Gallagher, & Naim, 2019; Eve, 2018). If these organisations are really serious about open access and want to make it work – for the good of the world – then take OLH as an example and take a small risk to try it out. It is all very well to think that experiments should always take place elsewhere, but sometimes we need to see leadership from those who profess to be field leaders. We would like to see the OLH model spread and will willingly advise any organisation that wishes to adopt it for itself.

Finally, if OLH is to be seen as a pilot light for a new model for fee-free OA, we cannot let the pilot go out. It is easy at this stage in an organisation’s lifecycle to sit back and celebrate accomplishments. Yet we have not yet achieved our mission: to propagate a model of fee-free OA worldwide at scale. We have shown it can work. However, if libraries cancel and we cannot persuade new libraries to sign up, the faith in our model may dwindle over time and APCs could win out. It is imperative that we continue to operate OLH at full capacity, showing an alternative to the rapidly emerging (and damaging) APC status quo, paving the way for a more just model of journal publishing that will benefit readers and authors worldwide.