The Open Access Interviews: Wellcome Trust’s Robert Kiley

Over the past year Open Access (OA) publishing has gained considerable mindshare, not just amongst researchers and librarians, but publishers too. This has been helped greatly by the perceived success of the Public Library of Science (PLoS) — which in 2010 managed to cover its operating costs with revenue for the first time. But as it becomes increasingly likely that OA publishing will prove no less expensive than traditional subscription publishing, a couple of key questions arise: How much will Gold OA eventually cost? And how will the research community pay for it? I explored these questions recently with Robert Kiley, Head of Digital Services at the UK-based Wellcome Trust, one of the world's largest medical research charities.

The consensus is that viability for PLoS was eventually achieved thanks to PLoS ONE and the “light” peer review model that it has pioneered. Indeed PLoS ONE is widely described as a “cash cow”, since it is believed to be subsidising not just the publisher’s flagship journals but practically the entire PLoS enterprise. For this reason, no doubt, traditional publishers are currently rushing to create clones of what PLoS itself describes as the first of a new breed of megajournals.

Amongst those to announce PLoS ONE look-alikes in recent months are the American Institute of Physics (AIP Advances), Nature Publishing Group (Scientific Reports), the Company of Biologists (Biology Open) and Sage (Sage Open).

In addition, practically all subscription publishers now offer a Hybrid OA option. This allows researchers to have their papers made freely available on the Web even when they publish in a subscription journal — if they agree to pay an article processing charge (APC). Designed by publishers as a way to offer OA without loss of revenue, Hybrid OA is invariable charged at premium rates — which range from between $3,000 to $5,000 per paper.

In short, despite their initial rejection (not to say repugnance) of OA, publishers now view it as a lucrative new revenue stream to have opened up in the scholarly publishing space. Indeed, they appear to fear that unless they move quickly they may lose out in what some have characterised as a “gold rush”.

This new attitude was articulated on the Liblicense mailing list recently by publishing consultant Joe Esposito, who assured list members “OA can grow and commercial publishers can become even more profitable, in part by co-opting OA publishing.”

Dilemma

In the meantime, however, many in the research community have resigned themselves to the fact that OA publishing may never provide the cost savings that it was expected to deliver. As former director of Penn State University Press Sandy Thatcher put it recently on Liblicense, “[W]ith the gold OA model, you are entirely at the mercy of publishers, who will charge what they need to make their preferred profit margin and will not be any more transparent than they are now about their actual costs. End users will benefit, but will the costs to the system be any less?”

If costs do not fall the research community faces a dilemma, since it was widely assumed that OA would provide a solution to the long-standing serials crisis — the phenomenon whereby library budgets have consistently failed to keep pace with the rise in the cost of journal subscriptions (effectively it is libraries that pay for scholarly publishing).

To add to the challenge, in the short term OA can only increase costs, because during the transition research institutions are having to continue to pay subscriptions while also paying for OA membership schemes. The latter are subscription-like deals offered by OA publishers like PLoS, BioMed Central (BMC) and Hindawi to enable research institutions to block purchase free-at-the-point-of-use publication rights so that faculty members can make their papers Open Access without being billed personally. These too are usually funded from library budgets.

In addition, many universities have created Gold OA funds so that one-off grants can be awarded to researchers wanting to publish in OA journals for which the institution has no membership agreement.

To cap it all, these additional costs come at a time when the global financial crisis is squeezing university budgets to death. Rather than being able to provide additional funds, most universities are beginning to find that they need to take an axe to their journal subscriptions.

In its 2010 Study of Subscription Prices for Scholarly Society Journals, for instance, Allen Press published a list of universities that made “significant institutional subscription cancellations” last year. This included Georgia Tech, the University of California, San Francisco, Oregon State University, and the University of Nevada, Las Vegas.

And this July The Chronicle of Higher Education reported that the University of Oregon and Southern Illinois University even felt compelled to cancel a number of “big deals”, which can mean depriving researchers of access to thousands of journals in one fell swoop.

Meanwhile on the other side of the Atlantic, Research Libraries UK (RLUK) has informed both Elsevier and Wiley-Blackwell that unless they agree to a 15% reduction in prices it will not renew its Big Deal contracts with them next year. This is serious: RLUK represents the so-called Russell Group of universities, whose membership consists of thirty major institutions, including Oxford, Cambridge and Manchester universities, Imperial College, the London School of Economics, and the British Library.

Safety net

In this kind of environment how can universities avoid eventually having to cancel their institutional membership schemes too — as in fact Columbia University did in 2009? And how long can they avoid having to close their Gold OA funds — as the University of Amsterdam did the same year.

In fact, Gold OA funds today tend to be more nominal than real. Having realised that demand could quickly outstrip available resources if OA took off, universities have in recent years started to impose rigorous eligibility rules on authors asking for assistance. When the Compact for Open-Access Publishing Equity (COPE) was founded in 2009, its stated policy made this explicit.

As the Director of the Office for Scholarly Communication at Harvard University Stuart Shieber explained at the time, “By design, the overall cost to a university of implementing the compact, in the short term, would be quite small. Hybrid open-access fees are explicitly eschewed, and true open-access fees tend to be found at present in just those areas of scholarship where grant support is most prevalent, reducing the underwriting load on the university substantially. Rough estimates based on the experience of the Berkeley Research Impact Initiative fall in the range of tens of dollars per faculty member per year.”

A year later Shieber was pleased to be able to report that, in practice, Gold OA funds are providing even less support than he had envisaged. Harvard’s own HOPE Fund, for instance, has to this day paid for only three OA papers.

In other words, the policy of many Gold OA funds is that if there is any alternative source of funding, and/or if the proposed journal does not meet a very strict set of criteria, no money will be made available. As Shieber points out, these funds should be viewed as “safety nets” alone. “Safety nets are useful even when they are not used.”

If institutions now start to cancel their OA membership schemes (which some believe provide pretty poor value anyway), the question inevitably arises: in light of the continuing financial squeeze, who on earth is going to pay for the “dramatic growth of Open Access” — as some characterise it?

An answer to this question is all the more pressing given that the fee waivers currently offered by many OA publishers could be under threat. With commercial publishers like Nature Publishing Group (NPG) entering the no-frills market created by PLoS ONE, it is anticipated that the waiver schemes could become rare, or even disappear altogether — a point made by Phil Davis on The Scholarly Kitchen blog earlier this year.

Dissemination costs are research costs

So who will pay for Gold OA? Shieber hinted at one alternative funding source in 2009: research funders. In his mind, no doubt, was the example of the Wellcome Trust, which in 2005 — when it announced its OA policy — stated that it believed “dissemination costs are research costs”. As a consequence, it said, in future it would foot the bill for any grantee who wanted to publish in an OA journal.

This decision cost Wellcome £622,000 ($1m) in 2005/6, and the bill has grown steadily over time. Last year the Trust forked out just over £3 million ($4.9m). And if and when OA publishing becomes the norm the cost could rise to £7.3 million ($12m) per annum, which represents around 1.25% of Wellcome’s current research spend.

Wellcome’s decision is good news for any of its funded researchers who want to embrace OA. Some, however, believe that it put the nail in the coffin of hopes that OA would reduce the cost of scholarly publishing. Unlike the funds created by COPE members, for instance, Wellcome is happy to pay for papers to be published in Hybrid OA journals — which, as noted earlier, charge premium rates.

With one of the world's largest medical research charities prepared to pay their asking price, critics complain, publishers have naturally come to assume that they can charge whatever they want for OA products — much as they have always done for journal subscriptions. As a result, critics add, the serials crisis will simply morph into a different kind of pricing impasse.

Others, however, argue that overpricing may not actually be possible in the current economic environment — as the stand taken by RLUK might seem to imply. Most research, they point out, is funded not by private charities like Wellcome, but by governments — and the current sovereign debt crisis afflicting the Western world has made money so tight that it just won’t be feasible to pay publishers the rates they want.

In fact, private funders are no less susceptible than governments to the financial turbulence we have witnessed in recent years. In 2008, for instance, the London Times reported that the Wellcome Trust had lost £1.5 billion ($2.4 billion) after some of its investments had turned sour the preceding year. This saw the Trust’s investments after debt shrink from £15.1billion to £13.1billion.

As a result, the amount of money Wellcome has available to fund research has fallen by £95m since 2007. And the continuing panics still sweeping through markets suggest that the situation is likely to get worse before it improves.

All in all, it is not at all clear today how Gold OA will be paid for going forward, or how OA publishing will develop.

Intriguingly, two months ago the Wellcome Trust announced that — together with the Howard Hughes Medical Institute (HHMI) and the Max Planck Society — it plans to launch a new OA journal. Even more intriguingly, the three partners stated that no APCs will be levied “for a number of years”.

This will inevitably further reduce the amount of money that the Trust has available for funding research projects. But what does it portend?

Confident

With these thoughts in mind I recently emailed some questions over to Wellcome’s Robert Kiley.

As will be evident from the Q&A below, Kiley remains confident that Wellcome’s decision to pay for Gold OA was the right thing to do, and he does not believe it will have a significant impact on the amount of money Wellcome is able to make available to fund research.

He rightly points out that — compared to the shifts in value that the ups and downs of the market can inflict on the Trust's resources — the $12m that it may have to spend on Gold OA is small beer. Moreover, Kiley adds, that figure “assumes we pick up 100% of the OA costs.” Since most research is funded by more than one organisation, he explains, it is safe to assume that the other organisations funding the research will share any publication costs.

And while declining to say how much money Wellcome is investing in the new OA journal, Kiley insists that the percentage of its funds devoted to research dissemination (rather than research itself) will rise only “slightly” as a result.

Finally, Kiley expresses himself confident that OA publishing will prove cheaper than subscription publishing, particularly in light of the current wave of PLoS ONE-clones being launched. These, he points out, levy lower APCs than both Hybrid and pure OA journals (i.e. between $1,350 and $1,980).

This assumes, of course, that a sufficient number of Wellcome-funded researchers will be content to publish in less prestigious no-frills journals like PLoS ONE.

But given the straitened circumstances that the research community finds itself in today, perhaps the more important point is that if PLoS ONE’s prices are such that they can subsidise the larger PLoS enterprise (and spark a me-to “gold rush” as a result), then we must assume that publishers still believe they can ask the research community to pay premium prices, even when selling entry-level products that provide only a minimal service (PLoS ONE articles are not copyedited for instance).

The research community might justifiably wonder why publishers appear to be the only stakeholders that do not believe they have to tighten their belts.

Wellcome’s plans to introduce a new OA journal do, however, suggest an intriguing possibility: Although the three partners have certainly not said as much (in fact they have given a number of not entirely convincing reasons), it is tempting to speculate that the unspoken motivation for launching the journal could be a realisation that, in the age of the Internet, professional publishers have become an unnecessary and expensive luxury.

We could note, for instance, that the project’s FAQ states, “Online open access journals such as the PLoS titles and 'Nature Scientific Reports' are a step in the right direction. Our open access journal will also incorporate online technology to allow for an engaging, interactive experience, but will be staffed by experienced, active scientists and will use a faster, more transparent peer review system.”

Below I publish the email Q&A I had with Robert Kiley.


Robert Kiley 

RP: I believe the research funded by The Wellcome Trust generates around 5,000 papers a year. Is that right? 

RK: Yes it is. 

RP: In 2005 the Trust introduced an OA mandate that requires all its funded researchers to either self-archive their final peer-reviewed manuscript in the UKPMC repository within six months of publication, or to publish them in Gold or Hybrid journals. To help them do the latter the Trust agreed to pay any OA charges incurred. 

RK: Correct. 

RP: I am told that Wellcome’s contribution to Gold OA costs consists primarily of block grants that it makes to around 30 large institutions. Other Wellcome-funded researchers have to apply to the Trust to have their grants supplemented on a case-by-case basis. Are the block grants based on an estimate of the number of papers that Wellcome grantees at those individual institutions are likely to generate, or is there a one-for-one relationship? 

RK: The block grants are based on an estimate of the number of papers that will arise from Wellcome-funded researchers at a particular institution. 

RP: Do the beneficiaries of the block grants supply Wellcome with statistics showing exactly how that block grant has been used? 

RK: At the end of each year, each institution supplies the Trust with a spread sheet showing what they actually spent on OA publication costs.

As with any grant, the institution can only claim against the grant commitment for money it has actually spent. I.e. this explains why the commitment funding levels are often higher than the actual funding spend.

Regrettable? 

RP: As I understand it, the costs arising from the Trust’s OA policy have grown over the past five years from £622,000 ($1 million) to just over £3 million ($4.9m). Is that correct? 

RK: This increase reflects the fact that more and more authors are complying with the Wellcome mandate, and for this many more are selecting the "author pays" option.

I guess it also reflects the growing availability of "pure" OA journals, from the stables of PLoS and BMC, and more recently from Wiley, BMJ, NPG etc. 

RP: If OA becomes the norm, then I understand that Wellcome expects these costs to grow to around $12 million (£7.3 million) per annum, which would represent around 1.25% of its total research spend? 

RK: Yes, but that 1.25% assumes that Wellcome is picking up 100% of the research publication costs. In reality, much of the research Wellcome funds is also funded by other funders. Thus, if a paper is attributed to more than one research funder then we would expect the publication costs to be apportioned as well.

We recognise that this splitting of OA costs is probably not fully in place yet, but with the development of "middleware" solutions — such as Open Access Key — the splitting of publication costs across funders should become easier. 

RP: What I take away from this is that in a fully OA world Wellcome could have around $12 million a year less available for funding research (as opposed to communicating it). I assume that previously these costs would have been met by research institutions (through journal subscriptions). Is that regrettable in any way? 

RK: To repeat the point, the $12m assumes we pick up 100% of the OA costs. The scale of support available for research is driven by our investment portfolio. 

RP: What does that mean in practice? 

RK: For example the total amount spent on grants in 2009/10 was £450m, compared to £490m in 2008/09 and £545m in 2007/08. The introduction of our OA policy has not impacted the funding available for research grants. See here for the details.

Costs 

RP: In a presentation you gave earlier this year at the annual conference of the Association of Subscription Agents and Intermediaries you reported that during a three month period at the end of last year Wellcome funded the publication of 440 articles. This cost the organisation just over $1million (£611,000), which averages out at $2,367 per paper (£1,447)? 

RK: Correct. Those figures were based on the 440 papers that we paid for during the last quarter of 2010. These articles were published in journals with a wide range of APCs, up to and including $5,000 for Cell Press. The median cost was $2,250 and the mode $3,000.

The recent RIN/CEPA/Ware study on transition models, by the way, argued that if the average Article Processing Charges were set at about £1,457 (or $2,185 — the estimate by Outsell of the 2009 average charge) then "UK universities would benefit from substantial annual net savings that we estimate at £2.8m (or £3.0m including non-cash savings), .....whilst at the national level the UK makes annual savings in funding costs of £9.7m"

As you say, in the piece of work I undertook I calculated that the average APC levied on Wellcome Trust authors was $2,367 — thus validating the estimates provided by Outsell, and confirming that the "low APC" modelled in this report is highly realistic.

We believe that the benefits of maximising the dissemination of Trust-funded research findings, via our open access policy, outweigh the costs. 

RP: As you imply, the RIN figures you cite assume that APC rates settle at a lower level. However, the report also says, “if average APCs were set at a level equal to the estimated current global average cost per article (£2,634), UK universities’ annual cash costs would rise significantly, leading to a high net cost to the UK relative to the other scenarios (other than licensing).” Your assumption presumably is that the average APC cost will fall by at least £1,177 from today’s price? 

RK: If you look at the raft of new, "pure OA" journals that have been announced over the past six months you will see that the APC is akin to the PLoS ONE APC. So, the APC for NPG's Scientific reports is $1,350; BMJ Open is £1,200 [$1,980], Open Biology (Royal Society £1,200); Biology Open (Company of Biologists) $1,350 etc. — all of which suggests that an average APC of around £1,457/$2,185 is realistic.

The RIN report also looked at the benefits (not just cash savings) that would arise if greater access were provided to each of the core user groups. Again, with an APC of around $2,200 the benefits cost ratio (BCR) is "very substantially positive, in the range 10.8-15.7".

New journal 

RP: As well as committing to pay all the Gold OA fees incurred by its grantees Wellcome recently announced — with HHMI and Max Planck — that it plans to launch its own OA journal. In doing so it does not intend to levy any publishing charges on authors "for a number of years". Presumably the Trust’s financial commitment to dissemination can be expected to grow even more going forward, removing further money from actually doing research. Would you agree? Do you have any figures on what these additional costs might be? 

RK: Yes, our support for this new journal will mean that we are spending (slightly) more on dissemination costs.

We have not disclosed how much we are investing in this new journal, but remember that this initiative is being jointly funded by the HHMI and MPS (as well as the Trust). Details about the new journal can be found here. 

RP: Wellcome is a private funder, and it was one of the first funders to embrace OA, so it has been very much in the vanguard of OA. Most research is funded by governments. Do you believe that they will follow Wellcome's lead and assume, as Wellcome does, that "dissemination costs are research costs", and so agree to pay all OA publication costs? 

RK: I think you will need to talk to RCUK etc. to get their view on this. That said, the RCs do already provide mechanisms for their grant holders to meet open access publications cost — either as direct costs through the grant, or indirect costs.

Anecdotally we hear that RCUK-funded researchers find it more difficult than Wellcome-funded researchers to meet OA costs. And, though I would agree that our mechanisms for meeting OA costs greatly simplify the process, a quick search on PubMed shows (for example) that in the past year there have been some 196 Medical Research Council-funded papers published in either PLoS ONE, PLoS Medicine, PLoS Biology or PLoS genetics.

The point I am trying to make is that, in this simple example, MRC researchers were able to meet the APCs levied by PLoS. 

RP: Or perhaps they were beneficiaries of the APC waivers currently provided by PLoS. As its FAQ puts it, “PLoS is committed to ensuring that our fee is never a barrier to publication and so we offer a waiver to any authors who do not have access to funds to cover our publication fees.” The problem with waivers, of course, is that they can be discontinued at any point. I am wondering, however, whether some of these costs are unnecessary in any case. It is widely believed, for instance, that far too many papers are published today (partly because researchers are evaluated by their institutions on the number of papers they publish). 

RK: Umm.. I think I might take issue with the assertion that "it is widely believed that far too many papers are being published" — I'm not sure this is a problem as such and there are actually some drivers for more papers to be published, such as those reporting negative results etc. And anyway, we — and I'm sure most other funders — would never evaluate purely on quantitative terms of course. 

RP: Thank you for your time. 

I discuss the affordability problem confronting scholarly publishing in an article on the “Big Deal” published in the September issue of Information Today. (More here)