The recent launch of the Compact for Open-Access Publishing Equity (COPE) has attracted both plaudits (e.g. here and here) and criticism (e. g. here and here).
What is COPE? It is a call to universities and research funding agencies to "recognise the crucial value of the services provided by scholarly publishers, the desirability of open access [OA] to the scholarly literature, and the need for a stable source of funding for publishers who choose to provide open access to their journals' contents."
Signatories to COPE are asked to commit to, "the timely establishment of durable mechanisms for underwriting reasonable publication charges for articles written by its faculty and published in fee-based open-access journals and for which other institutions would not be expected to provide funds."
Specifically, signatories are invited to create Gold OA Funds to assist researchers to pay to publish their papers in OA journals — which instead of charging readers to read (via a subscription), impose an author-side article processing fee (APC). The deal is that by paying a fee an author can ensure that the publisher will make his or her paper freely available on the Web for anyone to read, and thereby increase its impact.
COPE is the brain child of Harvard's Stuart Shieber, a professor of computer science, and director of the university's Office for Scholarly Communication. Shieber outlined the thinking behind COPE in an article published in August in PLoS Biology. COPE is necessary, he explained, because OA journal publishing is currently "at a systematic disadvantage relative to the traditional [subscription, or Toll Access (TA)] model".
The implication is that authors would be willing to publish their papers in an OA journal, if someone else was prepared to pay the associated publishing fee.
Universities need to support OA publishing, concluded Shieber, in order for it to become "a sustainable, efficient system". Only then, he added, can the two journal publishing systems (OA and TA) "compete on a more level playing field."
To date five universities have signed up to COPE, including Harvard, Massachusetts Institute of Technology, Cornell, University of California at Berkeley, and Dartmouth University.
What's the bigger picture here? For the past several decades scholarly journal publishing has been in the grips of the so-called serials crisis. This has made it increasingly difficult for research institutions to fund subscriptions to all the journals that their researchers need to do their work properly.
This is both an affordability problem (since it means that each year universities have to reduce the number of journals they subscribe to due to a shortage of funds) and an access problem (since researchers cannot read papers published in journals for which their institution has no subscription).
Open Access publishing (Gold OA) was devised as a way of overcoming these problems — by shifting costs from the reader side of the publication process to the author side by charging a one-off publication fee.
Although the access and affordability problems are part and parcel of the larger serials crisis if we want to understand the current situation more clearly, and assess the likelihood of COPE resolving matters, it may help to view them as two separate problems.
There are two reasons for doing this. First, as a result of pressure brought to bear on publishers by OA advocates the majority of TA journals now permit authors to self-archive copies of their papers in their institutional repository. Known as Green OA, this practice ensures that copies of papers published in TA journals can nevertheless be made freely available on the Web.
However, since universities still need to pay subscriptions to access all the papers that have not been self-archived, Green OA is unlikely to have any substantial impact on the affordability problem, at least in the short term. It does, however, hold out the promise of solving the access problem.
Indeed, so far as access is concerned Green OA has a significant advantage over Gold OA, because if all researchers began self-archiving their papers it would be possible to make 100% of the global research output freely available on the Web practically overnight. By contrast, it would take a considerable amount of time and effort to convert all journals to Gold OA — even assuming that publishers agreed to the conversion.
However, in practice today only around 15% of research is being self-archived by authors. And while research institutions can require self-archiving, very few have yet introduced self-archiving mandates, and it is not at all clear yet whether the growth in mandates will accelerate in the near future.
What are the origins of the serials crisis? It is partly a consequence of a continuous growth in the number of journals, but mainly a result of price inflation, which appears to be endemic to the scholarly journal market.
Hyperinflation, argues Shieber is the result of "systemic dysfunctionalities" inherent to a subscription business model when applied to scholarly journals. As in insurance-based health markets, he says, it leads to moral hazard — "the phenomenon of overconsumption of a good by a consumer who is insulated from the good's cost".
In other words, he explains in PLoS Biology, "The 'consumers' of scholarly articles (the readers, typically faculty, students, and researchers at universities and other research institutions) are insulated from the cost of reading, that is, from the subscription fees paid by the institutions' research libraries."
This is a problem that anti-trust economist Mark McCabe outlined to me seven years ago. As he put in 2002, "One distinctive aspect of this market is that end users do not pay for the material they use since the actual purchases are mediated by the libraries. This means that the principals (the professors, the scientists, the researchers of a particular institution) ask their agent (the library) to buy whatever they need, and the agent has no way of enforcing price discipline on the users. So there is a disconnect."
The second reason for treating access separately from affordability is that while Gold OA could in theory eventually produce 100% OA (and so solve the access problem) it is not at all clear that it can solve the affordability problem.
Since COPE proposes Gold OA as a solution to the predicament that the research community finds itself in, a key question, therefore, is whether Gold OA can avoid the systemic dysfunctionalities characteristic of TA publishing.
Currently the signs are not good.
When pioneer OA publisher BioMed Central (BMC) launched, for instance, it charged $525 to publish an article. Today its standard APC is nearly three times higher, at $1,535; and authors can pay anything up to $2,365 to publish an article in a BMC journal like the Journal of Biology.
Similar increases have been evident with other OA publishers. When Public Library of Science (PLoS) launched its first journal, for instance, it charged an APC of $1,500; today it charges nearly twice as much, with costs ranging from $2,250 to $2,900. (PLoS ONE charges $1,350, but is not a traditional journal).
Moreover with most researchers unable to fund the cost of paying APCs themselves a number of alternative payment options have been devised. This has led to the creation of Gold OA funds like those envisaged by COPE (there are currently 21 Gold OA funds, plus an ambitious project called SCOAP3 that hopes to convert the entire particle physics literature from a subscription model to Gold OA), and to the so-called "institutional membership scheme". Inevitably such schemes involve paying APCs from centralised funds.
Institutional membership, for instance, involves a research institution bulk-buying the right for all its researchers to publish their papers in any of a specific Gold OA publisher's portfolio of journals — with the costs generally charged to the library budget. Essentially, it is a quasi-subscription.
As a quasi-subscription institutional membership creates a similar "disconnect" between user and purchaser as occurs with TA publishing. Gold OA funds will inevitably have the same effect. As Shieber points out, the danger is that "since authors would not now have to pay the processing fee, they would over-consume in a price-blind fashion, and processing fees would hyperinflate just as the subscription fees."
It seems that this is not just a theoretical danger. In August 2007 unrest erupted over the institutional membership scheme operated by BMC, leading to the science and medical libraries at Yale University publicly announcing that they were discontinuing membership.
Explaining their decision Yale librarians pointed out that the university had been asked to pay BioMed Central less than $4,700 to publish articles in BMC journals in 2005, but that the figure had grown to $31,625 in 2006. "This experiment in open-access publishing has proved unsustainable," the librarians concluded.
A further fifteen universities also cancelled their BMC membership scheme.
BMC responded by saying that the price hike was a consequence of a rapid rise in the number of papers published in its journals. This is no doubt fair comment, but appears to confirm that OA publishing suffers from the same disconnect between principal and agent as is evident with TA publishing, and that this results in the same inflationary spiral.
In discussing this issue earlier this year I concluded that OA publishing funded with Gold OA funds, or institutional memberships, is likely to produce the same hyperinflationary effect that Shieber worries about. In other words, while Gold OA funds could eventually solve the access problem, it is far from clear that they can resolve the affordability problem.
OA advocate Peter Suber challenged my conclusion. "As soon as we shift costs from the reader side to the author side, then, we create market pressure to keep them low enough to attract rather than deter authors ... [and] ... precisely because high prices in an OA world would exclude authors, and not merely readers, there is a natural, market-based check on excessive prices."
Besides, he added, the majority of OA journals do not currently charge an APC. As he put it: "It's relevant to point out here that most OA journals charge no publication fees or institutional memberships at all. I've argued that even fee-based Gold OA is not the threat that Richard seems to think. But even if I'm entirely wrong about that: fee-based gold OA is a minority of Gold OA, and no-fee Gold OA doesn't pose any of the threats that Richard describes."
Shieber likewise cites evidence showing that the number of OA journals currently charging APCs is low. Writing in PLoS Biology he reported that fewer than 25% of the open-access journals in the Directory of Open Access Journals (DOAJ) are listed as charging a publication fee, "the remainder relying on other sources of direct or in-kind support."
Can we assume that this will continue to be the case?
Shieber thinks not. For this reason he reaches a slightly different conclusion to Suber: "[P]rocessing fees are likely to be an important revenue model for open-access journals, as they scale beyond the tiny fraction of overall journals that they currently constitute; processing fees are the only revenue source that inherently scales directly with the publishing services provided by a journal. The importance of the processing-fee model can be seen in the fact that of the open-access journals of sufficient standing to have an Institute for Scientific Information (ISI) impact factor, the proportion charging processing fees rises above 50%."
Putting a cap on it
Shieber nonetheless suggests that Gold OA can avoid the hyperinflation that has plagued TA publishing — if research institutions impose a cap on the money made available to authors when reimbursing their publishing fees.
With a cap, he explains, authors "would have to trade off whether using a certain amount of their limited allocation of funds for a given journal was appropriate in relation to the services and imprimatur that the journal provides, thereby reintroducing exactly the economic trade off that is missing from the current system."
He concludes, "In essence, the caps would act as inverse deductibles still allowing the economic signal to pass through to authors. In this approach, decisions about what is a reasonable fee are delegated to authors who choose on the basis of a market mechanism; the institution needn't stipulate reasonableness a priori."
This invites us to wonder what constitutes a reasonable fee, and whether a researcher would know a reasonable fee when presented with the bill. But would a cap work? "On the plus side," says Suber, "a cap will give authors a reason to husband their resources, or their allocation, introducing a note of price competition into their decision about where to submit a new work."
On the other side, he adds, "[A]fter an author spends her allocation, she might just submit subsequent work from the same fiscal year to a no-fee journal (OA or TA). That's compatible with price competition for the first article; but it's also compatible with disregarding price competition even for the first article. It's compatible with OA for the subsequent articles; but it's also compatible with TA for the subsequent articles."
However, the more important point, perhaps, is that universities will inevitably struggle to provide sufficient money to allow a Gold OA fund to ameliorate the predicament the research community currently finds itself in. This point was graphically demonstrated by Cornell doctoral student Phil Davis in May, when reporting on the plan of COPE signatory Cornell to make $50,000 available for a Gold fund.
"Considering that the Cornell University Library spends nearly $18 million dollars on collections, $50K seems like pocket change," he said. "From an management standpoint, it may take much more than $50K in staff and faculty time to administrate and process author charges one article at a time."
With APCs costing up to $3,000 per article (and some, like Cell Press, charging as much as $5,000), pocket change like this can hardly be expected to make much impression on the serials crisis. And since most of a library's budget will inevitably remain locked up in traditional journal subscriptions — which can be expected to continue to increase in price each year — it is hard to see how COPE-like initiatives will have much impact on either the affordability or the access problem.
"The notion of a COPE cap on the amount that funders and/or universities commit to subsidising authors for Gold OA fees is predicated on the enthymeme (i.e. the unstated or unrealised premise) that publishers abandon subscriptions and convert to Gold OA publishing," says Harnad.
In a post on Liblicense Harnad adds: "There are 25,000 journals, most of them not Gold OA, let alone equitably priced Gold OA, publishing 2.5 million articles a year from 10,000 universities worldwide. The tacit hope of COPE is to persuade all journals to abandon subscriptions and convert to equitably priced Gold OA by offering to pay for equitably priced publication today."
In short, COPE could only prove efficacious if universities cancelled all or most of their journal subscriptions and reallocated the money to pay for Gold OA (which they cannot risk doing), or of publishers voluntarily converting their journals to Gold OA. But as Harnad points out, "publishers have no reason to stop charging successful subscriptions just because some universities and/or funders commit to offering authors a capped Gold-OA subsidy."
In response to a claim on the Liblicense mailing list that COPE is a key OA initiative, and will allow "a smooth and successful transition to Open Access, Harnad responded: "it is just a very expensive way of generating some OA for a small fraction of a university's research output."
Like Harnad, Executive Editor for Social Sciences at Penn University Press Sandy Thatcher is doubtful that Gold OA funds will achieve very much. Commenting on a press release that the Scholarly Publishing and Academic Resources Coalition (SPARC) posted on Liblicense extolling the creation of Gold OA funds at the University of Calgary and the University of California at Berkeley, Thatcher said: "I wonder how such initiatives really will change the economics of the system overall, especially in the short term. Universities will still be paying for subscriptions for many journals and [are] now adding fees in addition."
In short, for so long as universities have to continue paying constantly increasing subscription fees to traditional publishers they cannot hope to buy OA for more than a handful of their researcher's' papers.
This, of course, is the crux of the systemic disadvantage that Shieber bemoans: research institutions cannot force a transition to Gold OA if their library budgets are already committed to traditional journal subscriptions. But it is far from clear that COPE can do anything to resolve it. Once again, we have to conclude that not only will COPE fail to resolve the affordability problem, but it will have little impact on the access problem.
Indeed, not only are budgets already seriously over-stretched, they face further cuts as a result of the current financial crisis. At least two of the COPE signatories — Cornell and MIT — are in the process of reducing their library budgets. There must therefore be doubts as to how long Gold OA funds can be supported. One of the first Gold OA funds to be created was introduced at the University of Amsterdam in 2007. Earlier this year, however, it was closed — "Due to a precarious financial situation."
Asks Harnad: "Would it not be more timely and useful (for OA) to encourage every university to provide OA for its own research output, by mandating Green OA self-archiving, rather than making formal or financial commitments before or instead of doing so?
Green OA, after all, is easy to implement, incurs no additional costs, and offers a much quicker route to OA.
As it is, he says, only two of the five signatories to COPE (Harvard and MIT) have introduced mandates, suggesting that COPE signatories may have lost sight of the primary objective — to make as much research OA, as quickly as possible.
COPE, says Harnad, is, "in effect simply encouraging universities to put up the cash today and then just sit and wait to see whether offering the capped subsidy will take the bait. Meanwhile, access continues to be lost, year after year."
Let a thousand flowers bloom?
One can of course argue that COPE does have the virtue of drawing attention to OA and the way in which the research community is being held hostage by scholarly publishers. Is that not a valuable thing to do? In any case, what's wrong with experimentation? Should we not let a thousand flowers bloom?
Sure, says Harnad, people are free to experiment, and they are free to speculate; but the first step of any institution serious about OA should be to do the most obvious and rational thing: introduce a self-archiving mandate. "With Green OA mandates safely seeing to access, we have nothing to lose in getting into speculative economics. But without Green OA mandates we lose our very raison d'être, OA, in favour of speculative economics."
Besides, he adds, taking a scattergun approach simply disperses the energies of the research community, and distracts them from the goal — to achieve OA as quickly as possible. And that, he says, is precisely what Green OA can do. "This is definitely not a case for 'let a thousand flowers bloom,' in parallel. It is much too late in the day for that, especially with the Green OA mandate option fully within every university's reach."
For Harnad, therefore, achieving OA should be treated as a serial process, with Green OA taking priority over Gold OA. Suber disagrees, arguing that there is no reason why Gold and Green OA initiatives should not be pursued simultaneously. "I do think the two can and should run in parallel."
Suber adds however: "[A]ny university willing to launch a fund to support Gold OA should also be willing to adopt a Green OA mandate. More: one of the primary arguments for a Gold OA fund (namely, to provide OA to a larger fraction of the university's research output) applies even more strongly to a Green OA mandate."
Consequently, he concludes, "I'd like to see universities support both Green and Gold OA with strong, effective policies. If they do, I don't care whether they adopt the Green policy first and then the Gold, or vice versa. But I do worry [that] when a university adopts a Gold OA policy without seeing that its reasons for doing so are even stronger reasons to adopt a Green policy as well."
All in all there are good reasons to be highly sceptical about the likely efficacy of COPE. For as long as library budgets are tied up with subscriptions it is hard to see how Gold OA funds can break the logjam, unless they are accompanied by Green self-archiving mandates.
The COPE signatories have perhaps mistaken intent for action.
The good news is that it would take very little to turn COPE from a well-meaning but probably ineffectual initiative into one with real teeth. All that is needed is to require that, in addition to creating a Gold OA fund, COPE signatories commit to introduce a Green OA mandate. Not only would this have the merit of pleasing both Harnad and Suber, but it would be far more likely to help the research community achieve OA (as in access).
As it is, COPE gives the impression of being a step back from the historic moment last year when Harvard's Faculty of Arts and Sciences became the first university department in the US to adopt a Green mandate — an initiative that was also masterminded by Shieber.
That said, although adding teeth to COPE could help solve the access problem it is not certain that it would do much to address the affordability problem — since there is no evidence that either Green or Gold OA are capable of reducing the costs of scholarly communication (as currently conceived).
Consequently affordability will doubtless remain a continuing concern for the research community, and for taxpayers — the people who ultimately fund scholarly communication.
It may be that if Green mandates multiplied they would eventually generate enough OA to induce subscription cancellations, releasing the subscription funds to pay for — and hence encouraging publishers to convert to — Gold OA. But that is speculation. It is also a discussion for another day!