Tuesday, March 13, 2012

Open Access, brick by brick

Last month Elsevier withdrew its support for the controversial Research Works Act (RWA). Had it become law, the RWA would have rolled back the National Institutes of Health (NIH) Public Access Policy requiring that funded researchers make their papers freely available on the Web within 12 months of publication. It would also have outlawed other US federal agencies from introducing similar policies. As such, the bill was a direct assault on Green Open Access. But while Elsevier’s retreat was a big win for supporters of Open Access (OA), OA will continue to be a brick-by-brick process — as evidenced by recent events in Australia.

In stepping away from the RWA, Elsevier acknowledged that it had made a strategic mistake. It clearly also made a serious PR gaffe. Whether the company has done lasting damage to its relationship with the research community remains uncertain, but the fact that researchers are continuing to sign up to the boycott Elsevier web site — created in protest at the publisher’s support for the bill — must clearly be a cause for concern.

What the RWA fiasco underlines is that while publishers are increasingly willing to embrace Gold OA (OA publishing), their antipathy towards Green OA (self-archiving) is growing, particularly where it is mandatory.

For that reason, Elsevier’s decision should be viewed as a political act alone, not a change of heart. Indeed, in announcing its withdrawal the publisher stressed that it remains firmly opposed to OA mandates.

Moreover, while OA advocates maintain that most publishers are now comfortable with the NIH policy this is surely only wish fulfilment. A week after the RWA died, after all, 81 publishers signed a letter opposing the Federal Research Public Access Act (FRPAA).

Introduced into both the US Senate and the House of Representatives on 9th February, far from outlawing the NIH policy, the FRPAA would propagate it — to around a dozen other US federal agencies. It would also reduce the embargo period from 12 months to six. As such, the bill would be a huge fillip for Green OA — although with the US elections approaching it seems highly unlikely to succeed, in the near future at least.

In short, the battle for OA goes on, but looks set to be fought primarily over Green OA henceforward.

Down under

Recent events on the other side of the globe would appear to confirm this. They also demonstrate that while the OA movement was victorious in the battle over the RWA, the war itself is far from over.

In a development generally under-reported outside Australia (pushed aside by the hubbub over the RWA perhaps), on 21st February the Australian National Health and Medical Research Council (NHMRC) announced that it plans to introduce an NIH-style mandate — effective July 1st.

This is undoubtedly another win for OA advocates but, as we have discovered with the open wars, every success is often accompanied by a failure.

Let’s examine what happened in Australia.
Colin Steele

The story appears to begin on January 5th, when Colin Steele, emeritus fellow of the Australian National University and convenor of the National Scholarly Communications Forum, published an opinion piece in The Australian. This alerted the research community down under to the RWA, and called on Australian universities to make a public statement in support of the NIH Policy and of Open Access.

With the title “Scholarly Licence to Print Money”, Steele’s piece concluded, “Ultimately, the prime issue is surely to disseminate university knowledge, which has been funded by taxpayers, as effectively and openly as possible, rather than for that knowledge simply to continue to be a source for large publisher profits and for manipulable metrics for research assessment exercises.”

Then on February 15th Justin Norrie, news editor at the Australian information service The Conversation, published a story about the Elsevier boycott, pointing out that 97 Australian academics had joined the pledge not to publish in or edit Elsevier journals.

In writing his article Norrie spoke to Danny Kingsley, the Australian National University’s manager of scholarly communications and e-publishing. And he quoted Kingsley saying, “The problem in Australia is that the research councils — the Australian Research Council and the National Health and Medical Research Council — award funding to academics who publish their work in the journals that are judged under a metrics system to have the most impact.”

As a result, Kingsley added, “if academics boycott those journals, it could really hurt their careers. It’s easier to give in to their extortionate efforts to extract money than to join a boycott.”

Kingsley’s remark stung the CEO of the NHMRC Warwick Anderson into responding: In a comment piece published in The Conversation six days later, Anderson denied that NHMRC funding decisions are based on journal impact factors. “This is not true. NHMRC removed journal impact factors from its assessments some years ago.”

And to demonstrate that he understands the rationale and necessity for OA, Anderson added that the NHMRC would soon be upgrading its OA policy. “One of the important benefits that the public expects from publicly funded health and medical research is access to the published findings of that research,” he wrote.

For that reason, he added, “From July this year, we will be mandating the deposit of publication outputs arising from NHMRC funded research into an institutional repository within 12 months of publication.”

For the moment the details of the mandate are not know, but the news was greeted enthusiastically. Since 2007 NHMRC has merely “encouraged” researchers to embrace OA, a policy now widely viewed as having failed.

Observers, however, were surprised at the turn of events, pointing out that it was an unusual way for a research funder to announce a change in policy. And the assumption was that Anderson’s announcement was a direct response to the controversy surrounding the RWA.

As always, of course, reality is more complicated than it seems. In fact, the NHMRC has been contemplating making OA mandatory for some years. Norrie had even signalled the policy change in his February article.

Insiders report that the NHMRC mandate was first mooted after a visit to Australia by Harold Varmus in 2008. Varmus has been an advocate for OA for many years and is a former director of the NIH. He is also a co-founder of OA publisher Public Library of Science (PLoS).

Following Varmus’ visit, says Steele, “the NHMRC were much taken with following the NIH policy.”

As first conceived, the NHMRC mandate would actually have been more OA-friendly than the NIH policy, since it was planned to allow publishers to impose no longer than a six-month embargo before papers were made freely available — half the period the NIH allows. However, following lobbying by publishers, notably Wiley-Blackwell, the NHMRC delayed its decision, and eventually watered it down to a 12-month embargo.

“The new NHMRC policy is stronger than the old one,” says US-based OA advocate Peter Suber. “The earlier policy encouraged OA without requiring it, but asked non-complying grantees to justify their non-compliance. That extra request put it above ordinary encouragement policies, just as the explicit requirement in the new policy puts it above the previous policy.”

However, he adds, “The 12 month embargo is a disappointment. NHMRC is following the NIH policy, of course.  But in this respect the NIH is the outlier, not the norm. To my knowledge, NHMRC is the first medical funder with an OA policy, after the NIH, to allow an embargo longer than 6 months.”

For all that, the announcement of the NHMRC mandate, if not the decision itself, does appear to have been triggered by the controversy over the RWA, although Steele puts it this way: “I don’t think their announcement was provoked by the Elsevier controversy, but rather by the interview with Danny Kingsley in The Conversation. But they were nearly ready to go with the mandate by then in any case.”

War of attrition

Whatever the reason for the timing, and the manner, of the announcement however, it serves to remind us that the path to OA rarely runs straight. Indeed, OA often seems more like a war of attrition than a movement. And sometimes it is a case of one step forward, two steps back.

Our Australian story demonstrates this well enough. Observers suggest that one reason for NHMRC’s delay was that the funder had hoped to be able to make a joint announcement with the Australian Research Council (ARC), as had been done in 2007 when they both announced their current policy of “encouraging” OA.

In the event, it appears that ARC is not prepared to move to a mandatory position right now. At least we have to assume that, since its public statements about OA tend to be more gnomic than illuminating, and somewhat contradictory.

A week after Anderson’s announcement, for instance, The Australian raised the issue of an OA mandate with ARC chief executive Margaret Sheil. Sheil responded that “open access publishing” by grant recipients was encouraged where appropriate, but not demanded, by the ARC.

She added that there was less need for the ARC to impose a mandate, since ARC-funded research was not generally of interest to the public. “In the NHMRC's case, by and large, what the public is seeking access to is information regarding health and that's quite appropriate. In many areas of our research there is no community interest in the outcomes until much further down the track.”

Sheil seemed to be implying that the primary purpose of OA is to make research accessible to the public, rather than to other researchers. She also seemed of the view that to demand that researchers make the output of their work OA would jeopardise their careers and/or was not possible.

“In the humanities and many other areas it can be difficult to get published,” Sheil told The Australian. “And what do you do about research for which the main form of publication is books? It's not that hard for a scientist to get papers into some sort of repository that's open access.”

In the hope of clarifying ARC’s position I contacted the funder. Responding by email, an ARC spokeswoman said, “The ARC funds a more diverse range of disciplines and end users than the NHMRC, some of which would not be in a position to comply with a mandate — health and medical is more advanced than some ARC-funded disciplines.”

And she added, “The ARC encourages researchers to disseminate their findings as broadly as possible to allow access by other researchers and the wider community. In Australia, institutions, in general, have their own repositories and researchers have other means of attaining access to the papers they need through coordinated library subscription. The Australian Government Excellence in Research for Australia (ERA) system has advanced the cause of open access by providing funding for institutional repositories.”

As the answer I had received invited further questions I emailed ARC again. Why, I asked, does ARC believe that some researchers would not be in a position to comply with a mandate? “The ARC does not want to put further barriers up for those research areas that, and early career researchers who, find it difficult to get published,” I was told, but with no explanation as to what exactly these further barriers might be.

I also asked ARC if it could confirm whether or not it intended to impose a mandate. “Our comments relate to current policy, which is all we will comment on,” I was told.

Deeply frustrating

Unsurprisingly, ARC’s reluctance to mandate OA is deeply frustrating for local OA advocates. Since ARC manages around A$845 million ($900 million) of taxpayer’s money each year, they point out, it is a public interest issue, and one deserving of greater clarity.
Arthur Sale

“Unfortunately the lack of action by the ARC means that taxpayer funded research is hidden behind a price barrier for many people who might benefit from reading it,” comments Australian OA advocate Arthur Sale. “It also fails the acid test of accountability: taxpayer-funded activities ought to be transparent and visible to all, especially where, as in the case of research grants, no national security issues or political advantage arises”

There is also frustration at ARC’s apparent lack of understanding of OA. “The reasons ARC has given for not mandating OA seem very confused,” says Steele. “It also seems to be slightly confused between the gold and green paths to open access.”

For instance, explains Steele, open access book publishing, which is more common in the humanities, has enjoyed considerable success in Australia. “Yet ARC commentators do not seem to be aware of developments here.”

Steele adds, “It is also claimed that open access could be a barrier for researchers to getting published, and could restrict early career researchers as a result. However no evidence for this has been provided.”

In fact, suggests Steele, the evidence would seem to point in the other direction, as ARC itself has acknowledged. “Professor Andrew Wells, the deputy chair of the ARC, told a National Scholarly Communications Forum on the future of the monograph held in Melbourne in September 2011 that ARC data shows top humanities researchers experience no problems in getting published in this way.”

But it is ARC’s claim that ERA has advanced the cause of OA in Australia that most frustrates OA advocates. In reality, they point out, it is quite the reverse. “It is true that the Australian government — through the Australian Scheme for Higher Education Repositories (ASHER) program — provided funding for repositories to enable collection of data about scholarly output for the ERA program,” says Steele. “However, this funding did not specify that it be used for open access dissemination of research, and many universities used the funding for establishing ‘an enabling environment’ for reporting.”

The result was that, rather than leading to more research papers being made OA, ERA encouraged the creation of dark archives in which the content is used exclusively for internal reporting purposes invisible to the outside world. “While all Australian universities have institutional repositories partially funded by the ARC, most of them have dark contents,” says Sale.

Kingsley highlighted this problem in a paper she gave in Berlin last September. Citing surveys undertaken by the CAUL Australasian Institutional Repository Support Service (CAIRSS), Kingsley uncovered a worrying trend: “The CAIRSS Repository Managers Survey over the past three years (2009-2011) shows that the ‘percentage of material in repositories that is open access’ across Australian universities has fluctuated. It started strongly with 44% in 2009, dropped to 33% in 2010 and rose back up to 37% in 2011 (CAIRSS, 2011).
Danny Kingsley

She concluded, “While the preparations for ERA mean that all universities have a repository, ERA appears to be detrimental to the promotion of open access in Australia. It is no coincidence that the first round of ERA happened in 2010, which correlated with a drop in the number of open access items in repositories in that year.”

In retrospect, this is unsurprising. “University libraries in Australia are under-resourced, and many universities have a cross-over of repository and reporting staff, so the focus of these people was, by necessity, on complying with ERA reporting rather than open access during 2010,” explained Kingsley. “One Australian university (in personal correspondence) commented that their university repository held approximately 50% of all the preprints of work at the institution, but since the ERA reporting process had begun very few had been deposited.”


In short, ARC’s claim that its funding of institutional repositories has advanced the cause of OA appears to be erroneous.

Moreover, by failing to join with the NHMRC and impose an OA mandate, suggests Sale, ARC has done the Australian research community a disservice. “As Australia has just two research councils, the universities have been waiting to hear a parallel announcement from the Australian Research Council (ARC),” he says. “However, there has been total silence. This is astonishing, as the universities are entitled to expect the two research councils to have the same policy, so as to simplify the university administration of grants.”

What we learn from this perhaps is that successful OA strategies require a deeper understanding of the issues than ARC has demonstrated. We also learn that promoting OA half-heartedly or trying to combine it with other tasks (e.g. internal reporting) may have the opposite of the desired effect.  

This too is unsurprising. The history of OA is littered with stories of well-intentioned stakeholders consistently misunderstanding OA, and how best to achieve it. It does not help that many of the actors in the scholarly communication process have a vested interest in maintaining the status quo, while others are fearful of possible unintended consequences, or simply prefer to avoid doing anything that could prove difficult or controversial, or that might incur additional expense — regardless of the many benefits provided by OA.

“The ARC is not in a position to monitor researcher compliance should open access be mandated,” I was told by the ARC spokesperson, “nor could the ARC fund publishing costs above the two per cent of total non-salary funding provided under ARC funding rules to support the publication and dissemination of project outputs.”

Here some OA advocates would point out that providing money for researchers to publish in gold journals is in any case a misdirection of scarce resources. Besides, they might add, most researchers prefer to spend their grant money doing research, not paying publishers to disseminate it. “The 2% funding is welcome but one wonders how many researchers will remember or want to use the current underfunding of the majority of the grants,” says Steele.

It also turns out that the 2% allowance is not what it seems, and ARC has been accused of “moved backwards” on its commitment here.

In short, argue Green OA advocates, throwing more money at publishers, rather than mandating researchers to self-archive, is a counter-productive activity, and in the long run can only feed the anger that the RWA controversy uncovered — anger, that is, at the way in which publishers like Elsevier appear to be growing rich at the expense of the research community. This, after all, appears to have been the primary motivation for the Elsevier boycott. As the boycott site puts it, “They [Elsevier] charge exorbitantly high prices for subscriptions to individual journals.”

What those pledging to boycott the publisher incorrectly assume, however, is that Elsevier is the only culprit. They also fail to see that OA publishing looks set to prove just as expensive as subscription publishing. As such, it will not solve the underlying affordability problem that has the research community so tightly in its grip.

Acute shortage of local champions

Achieving OA in Australia, suggests Steele, is proving difficult because there is an acute shortage of local champions. “No vice chancellor since the retirement of ANU’s Professor Ian Chubb — now Australia’s Chief Scientist — has come out to support public funding, public access, or public knowledge.”

In addition, he adds, “There is no equivalent in Australia of the UK JISC and the US ARL SPARC to promote open access over and above the individual universities. While the Australian government has espoused open access to government information, the relevant department — Department of Industry, Innovation, Science, Research and Tertiary Education — has not been significantly engaged in this policy area.”

Sadly, the frustrations experienced by OA advocates in Australia are not unique. Inertia, foot-dragging, sabotage, misunderstanding, confusion, and misdirected anger remain widespread everywhere. It is for this reason that, some 20 years after OA advocate Steven Harnad posted his Subversive Proposal calling on researchers to make all their papers freely available online by self-archiving them, still so few papers are being made OA today. Had Harnad’s call been heeded, universal OA could have been achieved many years ago.

In short, history suggests that the struggle for OA has to be viewed as a brick-by-brick process. Moreover, even after a brick has been laid it may subsequently be ripped out, or prove too friable to hold. Consequently, successful OA advocates require both huge patience and persistence.

We should not doubt that OA is inevitable. The danger is that if the research community throws money at OA journals, while allowing Green OA to be subverted and held back, it will discover that publishers increasingly adopt Gold OA, but price it at a level that simply protects their current income. This will see the affordability problem at the heart of the crisis in scholarly publishing simply relocated to the new OA environment.

Green OA, by contrast, holds out the hope of forcing publishers to downsize their operations to the provision of basic peer review services alone, and reduce their prices accordingly.

This is a topic I have explored elsewhere, where I suggested that one way of viewing the current situation is as a race between Gold OA and Green OA. As I put it, “If Green OA wins the race, the research community can hope to finally free itself of both the access and affordability burdens that have for so long dogged it, and publishers will be forced to give up some of their profits. The research community will have won the war.  However, if Gold OA wins the research community will have freed itself of the access burden, but failed to free itself of the affordability burden. Publishers will have won the war.”

It is for this reason, of course, that publishers are rushing to embrace Gold OA, while becoming more and more antagonistic towards Green OA. And it is for this reason that the announcement that NHMRC is introducing an NIH-style mandate is good news, while ARC’s dilatoriness is bad news.

Meanwhile, back in the Northern Hemisphere, Research Council’s UK is planning to insert a strong new brick into the Green OA infrastructure — in the shape of an upgraded OA policy. Amongst other things, this will reduce the time publishers are permitted to hold papers captive behind paywalls. As the Enabling Open Scholarship web site puts it, “the Research Councils will no longer accept embargo periods imposed by publishers, but instead stipulate an embargo period of no more than 6 months except in the case of humanities and social science.”

Brick by brick.


Bill Hooker said...

"The danger is that if the research community throws money at OA journals, while allowing Green OA to be subverted and held back, it will discover that publishers increasingly embrace Gold OA, but price it at a level that simply protects their current income."

I retain some hope that, because OA makes journals economic competitors rather than complements, market forces will keep Gold OA from ending up as expensive as the subscription model. Besides, if we must pay the same but can get 100% OA for our money then I can live with that outcome too.

Stevan Harnad said...


"if we must pay the same but can get 100% [Gold] OA for our money then I can live with that outcome too."

So could I -- if Gold OA were actually happening as fast as Green OA could have happened. But it isn't. We're still just waiting, and waiting…

So let universities and research funders worldwide follow the UK's lead, not Australia's lag: Forget about Gold OA for now and mandate doing the researcher keystrokes that would have given us 100% [Green] OA 20 years ago, had they only been done, unmandated, 20 years ago.

The reward will not only be 100% [Green] OA now, but Gold OA at a fair price soon thereafter.

To keep on yearning and waiting for Gold is just throwing more good time and money after bad.

Mike Taylor said...

Thanks for this very comprehensive and informative article. Just one point of (I think) correction, and a couple of thoughts.

"What those pledging to boycott the publisher incorrectly assume, however, is that Elsevier is the only culprit."

I don't think this is at all true. As a boycott signatory myself, I can tell you that I see all the big barrier-based for-profit publishers are part of the problem. The Elsevier-only boycott is a matter of eating an elephant one bite at a time.

"They also fail to see that OA publishing looks set to prove just as expensive as subscription publishing."

I dispute this. Exact figures are of course hard to come by, but in a letter to Times Higher Education I estimated the total cost of publishing a paywalled article with Elsevier at £10,130. In response to a comment from Kent Anderson I revised that down to £6,078; in an article in preparation I arrived at what I think is a more accurate figure of £6,689. Whatever the precise figure, it's evidently a great deal more than the $1350 =~ £850 that PLoS ONE charges, and in the region of 10-18 times as much as the average Gold-OA fee of $906 =~ £570 calculated by Solomon and Bjork (in press).

"The danger is that if the research community throws money at OA journals, while allowing Green OA to be subverted and held back, it will discover that publishers increasingly adopt Gold OA, but price it at a level that simply protects their current income."

I don't believe that would be possible. If my figures above are close to correct, and Elsevier were to protect their income by winding their Gold-OA fee up to £6,689 =~ $10,500, the market would decide.

Peter Murray-Rust said...

Many thanks to Richard for a comprehensive and valuable post.

@Mike_taylor. What are the costs you quote FOR. Publisher-incurred costs or author-incurred costs or both?

Diane said...

I like to think my recent essay 'Unshackling basic knowledge' in the Australian magazine Policy 2011/12 27(4)48-5
This is a link also helped prompt the NHMRC's announcement.

Diane said...

Another comment (from ‘biomedical-land’):
Green OA is poorly appreciated amongst biomedical researchers and institutes (most outside USA have never heard of it or believe it is only practised in other fields), whereas Gold OA is gaining reasonable ground. Biomedical researchers see author charges as a major disincentive to publishing through Gold OA journals because few funding agencies provide for them. This is a shame because biomedical research is relatively expensive, making Gold OA publishing good value.
I find it surprising you say
‘The danger is that if the research community throws money at OA journals, while allowing Green OA to be subverted and held back, it will discover that publishers increasingly adopt Gold OA, but price it at a level that simply protects their current income.’
Surely, high publisher profits are not the problem. The problem is that their profits are coming at the expense of access to information. You need to factor in the cost of this arrangement.

Mike Taylor said...

Peter: I calculated total cost of publishing a paper with Elsevier simply by dividing their total income from journal subscriptions by the total number of articles that they publish. So the resulting number is what it costs the academic community as a whole to publish a paper and get their current level of access to it.

Richard Poynder said...

Thank you to everyone for your comments. I hope to address some of these points separately, but on the specific issue of Elsevier’s per-article costs I asked the investment analyst Claudio Aspesi for his views. Claudio tracks Elsevier for investors and so spends a lot of time perusing the publisher’s accounts. Below is his reply.

In 2011 Elsevier had total revenues of £2058 million. S&T journals account for about 45% of that (the company does not disclose the exact number), so that would make it £926 million. Assuming that the average operating margin of Elsevier (37%) applies to journals as well, costs would be £583 million. Since they publish about 230,000 articles/year, that comes to £2,536/article, which seems much more in line with what PLoS charges.

Also bear in mind that a) the margin is probably higher for journals, so costs are probably lower and b) that rejected articles also drive some costs; the higher the rejection rate, the higher the costs. If we had access to internal figures, we would want to look at cost/article submitted, rather cost/article published.

Wouldn't it be great if Elsevier were to publish the necessary data to allow us to estimate more accurately what its average cost-per article was?

Bjoern Brembs said...

With regard to gold OA article pricing, let's put ourselves in a gradstudent's or postdoc's shoes. In the hypothetical case of universal gold OA, imagine you have a great manuscript and have the choice to publish it in Cell (to stay with Elsevier) for 20k or with the new Imaginary Journal of Cellular Biology for 7 bucks (Arxive price). The market says that Cell will have to reduce prices to compete. The student/postdoc will say: I'd rather use my personal funds, if I have to, and publish in Cell.

That's why, in the absence of a complete destruction of journal rank, there is no market driving down prices - and I don't see why gold OA should singlehandedly abolish journal rank. Thus, as I see it, there's not really much sense in pushing for gold OA before journal rank is destroyed. Or am I missing something?

Arthur Sale said...

It is plainly untrue for the ARC to imagine that members of the general public are not interested in research produced by ARC grants. The mistake the ARC makes is assuming that all researchers work in large institutions, especially universities. They should begin to think of school teachers, farmers including fish farmers, environmentalists, conservationists, engineers, computer professionals, and the list could go on - all members of the 'general public' and most vitally interested in Australian research funded by the ARC, but precluded from reading it due to a price barrier.

The net result is of course to depress the contribution that private enterprise can make to the Australian economy arising from Australian taxpayer-funded research.

Richard Poynder said...


I realise I would have been better writing a new blog post rather than adding such a long comment, but I wanted to keep the discussion all in one place. I have, however, had to break my comment into four parts in order to cope with the length limit of Blogger comments.

In what I say, I am going to assume we all agree that OA publishing is intrinsically better than subscription, or Toll Access (TA), publishing. Indeed, I will even assume that TA publishers like Elsevier would be perfectly happy to adopt Gold OA today if they could be sure that doing so would not cause any decline in their revenues/profits, either during the transition to OA or in the new OA environment. I also assume we all agree that OA is inevitable.

What is at issue then is when we can expect OA to become the norm, and what the consequences of that will be in terms of the costs of disseminating research.

And the key question, in my view, is whether OA publishing will solve the affordability problem that has plagued the research community for several decades now — a phenomenon commonly referred to as the “serials crisis”.

I am thinking that both @Bill Hooker and @Diane take the view that since the primary objective of OA is to solve the accessibility problem, then solving the affordability problem would be no more than a “nice to have” addition. I can entirely sympathise with this view. However, I suspect that taxpayers would not. And it is they, after all, who ultimately fund scholarly communication.

Let us suppose that all the issues facing scholarly communication, in all their complexity, were explained to taxpayers, and in a way that was both accurate and comprehensible. What would they conclude? I’m betting they would determine that both problems — accessibility and affordability — are of equal importance, and that it is therefore essential to solve both of them.

However, whatever taxpayers concluded the signs are that those responsible for distributing taxpayers’ money to the research community are likely soon to begin refusing to foot the bill for scholarly publishing unless prices do start to fall.

If I am correct, then costs are going to have to come down, one way or another. So whether OA publishing is cheaper than TA publishing is probably more important than most researchers appreciate.

None of us, of course, can predict the future, but let me offer a few thoughts on this.

In the early days of the OA movement (12 or so years ago) it was taken for granted that OA publishing would significantly reduce the costs of distributing research. Over time this claim has tended to evaporate, and OA advocates and publishers are now generally silent on the topic. @Mike and @Bill however say they still believe (or hope) that OA will solve the affordability problem. @Björn Brembs, by contrast, seems less convinced.

@Mike cites as evidence that OA publishing is cheaper than TA publishing the fact that PloS’ prices are orders of magnitude lower than what he estimates to be the per-article costs of publishing a paper with Elsevier. Where PLoS ONE charges $1,350 to publish a paper, he says, the cost per-article of publishing with Elsevier is somewhere between £6,000 and £10,000 ($9,400 and $15,700).

Richard Poynder said...


In passing, it might be instructive to compare @Mike’s estimates of Elsevier’s per-article costs with those given to the UK Select Committee Inquiry into scientific publishing in 2004 by Richard Charkin of Nature Publishing.

Charkin told MPs that the per-article costs of publishing in Nature were simply too high for OA to be a viable business model for the journal. “[I]n order to replace our revenues you would have to charge the author somewhere between £10,000 and £30,000 [$15,800 and $47,500) because the costs of editorial design and support are so high,” he said.

How did Charkin arrive at that figure? “Very crudely, £30 million of sales: we get income of £30 million and we publish 1,000 papers a year. That is your £30 million,” he told MPs.

@Mike appears to have used a similar method to arrive at his figures for Elsevier. But how accurate a method is it for calculating per-article costs?

Not very perhaps. Aspesi suggests above that a more likely per-article cost for Elsevier would be £2,536 ($3,705). We could also note that Elsevier itself estimates that it costs on average $2,000 [£1,260] to peer review a paper. (Although it is not clear whether these are Elsevier specific estimates or not).

There are other aspects of @Mike’s comparison that we might want to question. As we know, PLoS ONE pioneered a “lighter” review system. This has proved somewhat controversial, not least because one consequence of this is that the journal accepts 65% to 70% of the articles it publishes. (Elsevier estimates average acceptance levels at between 25% and 50%).

Since publishing costs are directly related to acceptance rates this must inevitably significantly lower PLoS ONE’s costs. A fairer approach, therefore, would be to compare Elsevier’s costs with what PLoS Biology or PLoS Medicine charge, which in both cases is $2,900. Aspesi concludes therefore that Elsevier’s costs seem “much more in line with what PLoS charges.”

On the Solomon and Björk study that @Mike cites, we should note that this included many journals based in the developing world catering to local authors. As such, it might not be very helpful for present purposes. Moreover, for some unaccountable reason, the authors also included a number of predatory OA publishers. It is far from clear that these publishers send all their papers out for review, so they can presumably afford to offer much lower rates.

We also need to do more than just compare current costs. We need to look at costs over time as well. This is important because many believe that the serials crisis is a product of the fact that scholarly publishing is a dysfunctional market — that is, a market without adequate mechanisms to restrain prices.

@Mike believes that OA publishing will be different, because it will send out appropriate market signals that will prevent the run-away price increases characteristic of subscription publishing. @Björn suggests that this would only happen if we first saw the “complete destruction of journal rank”.

I think @Björn is right, but I am aware that Stuart Shieber has published some suggestions on how he thinks a market signal could be introduced into OA publishing.

Richard Poynder said...


But what does the experience of the last decade or so tell us about OA publishers’ prices? When they launched both PLoS Biology and PLoS Medicine levied a $1,500 article processing charge. Nine years later, this fee is 93% higher, at $2,900. PLoS ONE has also increased its prices since it launched.

(For purposes of comparison, consider that a typical BioMed Central journal like the Journal of Translational Medicine initially charged $525. Today it charges $1,970 — a 275% increase).

These price increases are striking, particularly given that when I spoke recently with PLoS co-founder Mike Eisen, he said to me, “the marginal cost of processing an article is going down, and will continue to do so, asymptotically approaching zero”.

Why, then, I asked Eisen have PLoS’ prices risen rather than fallen. He replied, “the holy-grail — when the cost of the publication part of the process drops close to zero — is when we get publication ready content from authors … Our costs for PLoS ONE haven’t dropped that much yet because every paper still requires manual attention.”

It might help here to consider how the views of Vitek Tracz, the founder of BioMed Central, have changed over time. Initially Tracz too assumed that OA publishing would cause prices to fall to the point where they approached zero. When I spoke to him in 2006, Tracz described OA as a service. A service, he added that “will become a smaller and smaller component of the publishing business. More and more of what we do for authors today they will be able to do for themselves in the future, and as we develop more tools to allow them to do it themselves, so what we charge them will be less and less.”

As an OA publisher, Tracz has first-hand knowledge of the issues and practicalities of OA publishing. Today, however, he says he does not expect OA to cause prices to fall much. When I spoke to him again last year he told me, “I always assumed that OA publishing would be cheaper, but OA publishing is more complicated than we had initially envisaged. The logistics of it all quickly becomes very complicated, so it turns out not to be much cheaper than traditional publishing. This in turn means that the amount of money publishers can make is similar.”

Richard Poynder said...


Time will tell who is right, but as I understand it, Eisen imagines a future somewhat like that envisaged by Tracz in 2006. That is, a world in which the writing, copy-editing, typesetting etc. etc. is all done by the author. Since the reviewing is done for free by other researchers one is left wondering what the role of publisher might be in such a scenario.

Presumably, the bulk of a publisher's job would simply be to post papers into an online database (After all, researchers are now generally asked to recommend their own reviewers and the process is almost entirely automated).

As publishing consultant Joe Esposito has pointed out, this amounts to little more than providing disk space. And this is essentially what the physics preprint server arXiv does. To do this, arXiv estimates, costs just $7 per paper.

But in a world increasingly populated by both institutional repositories, and central repositories like PubMed Central, one might wonder whether there is any meaningful role for publishers to play in the future. The research community can do it all itself.

Why, therefore, asks former BioMed Central publisher Jan Velterop, does not the research community move to the endorsement model used by arXiv. If it did, he adds, it could save the taxpayer $3 billion a year of unnecessary expense.

Perhaps Tracz’s initial views were right all along?

In the meantime, one is tempted to ask: Why does PLoS ONE charge $1,350 to publish a paper? Eisen told me that this generates a significant surplus for the publisher. Since PLoS is a non-profit committed to “making the world’s scientific and medical literature a freely available public resource” should it not be charging a price more in line with the costs it actually incurs?

Could it be that when it became a publisher PLoS lost sight of its mission, and succumbed to the temptations and luxuries associated with “being a publisher”.

Recently Esposito commented on the Liblicense mailing list, “For those who have visited the offices of PLOS, which are among the more sumptuous of any publishing offices I have ever set foot in, there is the disconcerting feeling that all this is fleeting.”

Taxpayers might be tempted to mutter, “Let’s hope so.”

In any case, if PLoS truly remains committed to its mission should it not aim to operate out of less luxurious accommodation?

Let me finish with one final question: Should PLoS ONE not consider lowering its article processing charge?

Alex said...

Let's say that (despite the comments of other experts quoted) the ARC is correct that the humanities are not "in a position to comply with a mandate". They could simply require that only the funded scientists make their articles freely available, e.g. 6 months after publication. The ARC has a classification system for disciplines (FoR codes), so if they cared to be completely explicit about who the mandate applies to, they could list he FoR codes for which the mandate applies.

Kamal Mahawar said...

We have been saying for some time now that "open access" the way it exists today will not solve the problem of affordability in scientific communications. Yes, there are some people making too much profit in all this but the more fundamental problem is not that but the processes involved in academic publishing. If we do not examine, how scientists should communicate at a more basic level, it would be difficult to resolve these issues. We at WebmedCentral have provided the community with an alternative. If all scientists engage with our model, databases like PubMed include us, and citation tracking bodies start tracking us, our model based on instant publication followed by post publication peer review has the potential to offer a lasting solution to this problem.

Mike Taylor said...

Yeesh! I have only now seen that there was followup to my comment -- sorry to be so late in replying. I have not yet read Richard's long four-part comment, but wanted to respond to Claudio Aspesi's numbers.

First, this: "In 2011 Elsevier had total revenues of £2058 million. S&T journals account for about 45% of that (the company does not disclose the exact number)"

I, too, had difficulty in finding this percentage. Here's best I've been able to do. Tom Reller kindly pointed me to an investor seminar report. The chart on page 8 says that Elsevier revenues are split 50-50 between S&T (Science and technical), and that of S&T's 50%, 39% is from research journals. That implies that 78% of S&T revenue is from journals, which is confirmed by the chart on page 18. Unfortunately, the report says nothing about how revenue in Health Sciences breaks down, and my emails to various Elsevier personnel asking for this number have not received informative responses. In the absence of that information, I assumed for my new article in The Scientist that the same 78% applied (so that the overall percentage was also 78%).

Is Claudio is right that in fact research journal account for only 45% of Elsevier revenue in total, then that would mean that only 6% is contributed from Health Science, i.e. only 12% of all HS revenue is from journals. That seems unlikely to me, but then as Richard points out, "Claudio tracks Elsevier for investors and so spends a lot of time perusing the publisher’s accounts", so I wouldn't necessarily want to argue with him. Nevertheless, I would very much like to see his source for the surprisingly low figure of 45%.

Moving on ... "so that would make it £926 million. Assuming that the average operating margin of Elsevier (37%) applies to journals as well, costs would be £583 million."

Wait, wait! Why would we care what the cost to Elsevier is? The fact that they take out 37% in profit neither increases nor reduces the costs to their customers -- all the customer cares about is what they have to pay, not how much of that is costs to Elsevier. So the number we care about here is £926 million (assuming that low 45% figure is correct).

Next up ... "Since they publish about 230,000 articles/year" -- for the record, I should note that Elsevier's 2011 annual report gives a higher number of 240,000, which would make the per-article price 4% less.

" ... that comes to £2,536/article, which seems much more in line with what PLoS charges." But that number is uninteresting: it's what it costs Elsevier to publish each article, but we care about what they charge the customer. So the relevant figure would be £926m/230,000 = £4026. That is currently $6400, which is nearly five PLoS ONE articles. So still astoundingly less good value for the scientific community, and remember that is using the very low figure of 45% for the proportion of Elsevier revenues that are from journals.

Into the home straight now ... "Also bear in mind that a) the margin is probably higher for journals, so costs are probably lower " -- irrelevant, since we care what Elsevier charge, not what it costs them.

"and b) that rejected articles also drive some costs; the higher the rejection rate, the higher the costs." Also irrelevant, since this is true of all publishers. (PLoS, too, could cut the publication fee if the rejected fewer articles.)

So it seem the best case for Elsevier has them costing the academic community nearly five times as much as PLoS ONE (which, remember, is a relatively expensive OA option, costing half as much again as the average Gold OA fee of $906.

Richard Poynder said...

Claudio Aspesi has asked me to post this comment for him:

I read Mike Taylor’ posts with interest. A few specific comments:

1) I never said that journals account for 45% of Elsevier revenues. I said that S&T journals probably account for 45% of Elsevier revenues. I agree that the likely Elsevier S&T revenues from journals are in the region of 80% (in 2010, according to the company, they represented 78%, and journals grew faster (4%) than the rest of the division in 2011 (close to zero), still according to the company). Since S&T revenues were more than 50% of total Elsevier revenues in 2011, the 45% number for Elsevier S%T journals seem a reasonable assumption for 2011.

2) In discussions with Reed Elsevier in March 2011, it indicated that journal revenues in Health Care are 20% of Health Sciences revenues (not 80% or anywhere near that number).

3) The 230,000 articles number comes from a slide in the December 2011 investor seminar. It refers specifically to S&T journals. In fact, elsewhere in the presentation, the company indicates it publishes a total of about 300,000 articles/year, which would seem to also include the Health Sciences journals. In fact the ratio of articles between S&T and Health Sciences (230,000/70,000) appears reasonably close to the revenues ratio (40-45%/9-10%).

4) Costs incurred in rejected articles are a cost that have to be incurred by someone (whether it is the publisher, the subscriber or the published authors in an OA journal), unless a journal abandons altogether screening for articles. While it is true that these costs are borne by all publishers, not all publishers bear the same ratio of costs incurred on rejected articles per article published. This ratio is not necessarily determined by each publisher, but by the behaviour of authors (when they submit their articles to a specific journal). If we assume two journals to publish the same number of articles each year, have exactly the same costs to publish an article and the same operating margin (or return on invested capital, or on any other metrics relevant to their respective owners), the one which receives the most articles (and therefore rejects the most) will incur higher total costs. These costs will have to be borne by someone, and they cannot be ignored. What we do not know is how relevant these costs are for different publishers.

I agree, on the other hand, with Mike Taylor’s point that the costs incurred by the academic community (including the publishers’ operating profit) to disseminate an article on Elsevier S&T journals are still higher than on PLoS. After all, £4000/article (including operating profit) at Elsevier is a bit over twice the $2,900 charged by PLoS Biology (the publication mentioned as a closer comparable by some of the other commentators). Again, I want to underline it is somewhat dangerous to assume too much without knowing the cost of “rejecting” articles and the rejection rates of different journals, and – therefore – the real costs. Ultimately, however, if the academic community wants to lower its dissemination costs, it has to demonstrate it is willing to stand behind its stated convictions, abandon commercial subscription journals and live – at least for a while - with lower journal impact factors, rankings, etc. I suspect it is a classical prisoner’s dilemma for every scientist.

Mike Taylor said...

Thanks, Claudio, for reading and responding to my message.

First, thanks for the clarification on journals vs. S&T journals. That is a horrible lack of clarity in Elsevier's reporting, and it's good to get things straighter from someone who knows more about what's going on. That said, it's weak sauce for me to cite "Claudio Aspesi, pers. comm. 2012" for this stuff. Is it published anywhere? Maybe in financial reports that I've missed?

Great that you were able to get the 20% figure for proportion of Health Sciences revenue that is journals (though infuriating that it is still unsourced).

Yes, I see the 230,000 figure for published articles on page 20 of the investment seminar report. Obviously that figure's going to vary from year to year. It's a shame that no one document has all the information we need for these calculations.

Again, the costs incurred by a publisher's rejection of articles are not really of interest to a customer buying a subscription, who never sees these articles. From the customer's perspective, it's irrelevant whether Elsevier's revenue goes on editorial services for published articles, rejecting others articles, or indeed private jets or shareholder dividends. It's all money out of pocket, and that's all that matters.

(As an aside, I question how much value rejection adds -- it's always seemed strange to me that journals boast about their high rejection rates as though that is a service that they provide rather than an unfortunate side-effect of space constraints. We all know the truth of rejected articles: they get reformatted (a tremendous waste of time) and sent off to another journal, and the process is repeated until the article is accepted anywhere. So rejection doesn't in any meaningful sense serve as a filter that keeps "bad" articles out of the pool -- at best, all it does is move them around. I wonder what proportion of all the articles Elsevier rejects end up published in another Elsevier journal? But anyway, that's beside the point. I should really blog about it separately.)

Finally, it seems a little disingenous to compare the cost of publishing in an average Elsevier journal with that of publishing in PLoS's flagship journal, PLoS Biology (impact factor 14ish). Since PLoS ONE's healthy impact factor of 4.411 is far above the average IF of Elsevier journals, that is the correct point of comparison (and even that is being generous when one looks beyond Cell to the rank and file of Elsevier journals).

In the end, I think your invocation of the Prisoner's Dilemma is spot on. That is an interpretation that deserves more exposition. I may blog about it on SV-POW!.

Stevan Harnad said...


It is rather arbitrary and unrealistic to reckon the costs of Gold OA publishing independently of other factors that could raise or lower them.

The two most important causal factors are (1) Green OA and (2) institutions' subscription budgets.

Institutions cannot cancel essential journals if their contents are not otherwise accessible to their users.

If Green OA is universally mandated, then authors' final, peer-reviewed drafts of all journal articles are deposited in institutional repositories and freely accessible to all users whose institutions cannot afford subscriptions to the journals in which they appeared.

This makes it possible for institutions to cancel subscriptions, eventually making the subscription model unsustainable as the means of covering the costs of publication.

Subscription cancelations force journals to cut inessential costs.

With the refereed final drafts of all articles accessible to all through Green OA, journals no longer need to (1) provide the print edition, (2) provide the online edition or (3) provide access or archiving: The distributed network of Green OA repositories provides all that is needed. The rest are all obsolete products and services in the universally mandated Green OA era.

When the costs of (1), (2), and (3) are unbundled from publication products and services made obsolete by universal Green OA, the only essential cost remaining is that of implementing peer review.

Peers review for free, so the cost of peer review is just the cost of managing the peer review process, including the editorial expertise and judgment in choosing referees, adjudicating referee reports, and adjudicating revised drafts.

If peer review is provided as a "no fault" service to the author's institution, per submitted draft, regardless of whether the outcome is rejection, revision, or acceptance, the cost of rejected articles can be unbundled from the cost of accepted articles; this not only lowers and distributes the cost of peer review, but it removes the risk of lowered peer review standards and over-acceptance for the sake of making more money through Gold OA.

This much lower cost of post-Green OA no-fault Gold OA -- my guess is that it would be between $200 and $500 per submitted draft -- would not only be incomparably more affordable than today's pre-Green OA fees for Gold OA, but the money to pay for it would be available, many times over, from a fraction of institutions' permanent annual windfall subscription savings released by the cancelations made possible by universally mandated Green OA.

The only essential element for having Gold OA at this much more realistic and affordable price is one cost-free act on the part of the universal providers of all research output: Institutional Green OA mandates (reinforced by research funder Green OA mandates).

Without taking these costs and causal factors into account, estimates of the costs of OA are arbitrary and the wait for universal OA will continue to be long.

Harnad, S. (2007) The Green Road to Open Access: A Leveraged Transition. In: Anna Gacs. The Culture of Periodicals from the Perspective of the Electronic Age. L'Harmattan. 99-106.

Harnad, S. (2010) No-Fault Peer Review Charges: The Price of Selectivity Need Not Be Access Denied or Delayed. D-Lib Magazine 16 (7/8).

Harnad, S. (2010) The Immediate Practical Implication of the Houghton Report: Provide Green Open Access Now. Prometheus, 28 (1). pp. 55-59.

Harnad, S. (2011) Open Access to Research: Changing Researcher Behavior Through University and Funder Mandates. JEDEM Journal of Democracy and Open Government 3 (1): 33-41.

Mike Taylor said...

Having now (finally!) reach Richard's giant four-part comment, I find that I have nothing to add in response to it beyond what I wrote in my most recent response to Claudio. The key points remain (A) we care about what Elsevier charges us, not what their expenses are; and (B) we must compare like with like, not compare PLoS's flagship 90%-rejection journal with Elsevier's rank-and-file.

David Roberts said...

Mike Taylor wrote:

"I wonder what proportion of all the articles Elsevier rejects end up published in another Elsevier journal?"

something like this was mentioned in the Evolution of Science panel discussion, though it may have been NPG who said it. I can't remember when in the video it happened, unfortunately, though it was later on.

Anonymous said...

I enjoy reading these comments. I was happy to see the cost issue related to editorial work being addressed by people at PLoS.
As a programmer, aside from a PhD student, I wish to help to obtain such a goal; publication quality PDF, EPUB3, or HTML5, for scholarly work, done by the researchers themselves. I'm looking towards expanding Pandoc for such an authoring tool. I'm surprised they didn't prioritize this goal rather than heavily investing in commenting/web2.0/etc. style Journal homepages, which is more or less deserted by researchers (though I do like the idea myself, and would have used it, had my research been in the field of biology).
Knowing that this is indeed part of the long term goal for PLoS is very encouraging.

mike mcgrath said...

As a recently retired editor of a journal for Emerald can I just support Stevan Harnad's comments on the cost of peer review and its relationship to rejection rates? As editor I selected reviewers (which included myself), they reviewed for free. I uploaded the final and agreed articles to the publisher's web site using publisher software that converts to html and pdf. I then assembled and confirmed an issue of the journal. Only at this stage is the publisher's intervention required in order to attach metadata etc and ensure accessiblity to customers from their web site.Rejecting manuscripts is often a lengthy process because of the need for careful criticism but is still at no cost to the publisher. So from where does Elsevier derive its reviewing costs?