Tuesday, July 03, 2012

The OA Interviews: Keith Jeffery, UK Science & Technology Facilities Council

Keith Jeffery has been developing ICT solutions to “support, enhance and assist in research and the management of research” since the 1960s. It is this mission that lies at the heart of his current post as Director of IT and International Strategy at the UK Science & Technology Facilities Council (STFC).
Keith Jeffery

Jeffery’s extensive knowledge and experience of how IT systems can be used to assist in the research process has also earned him a place at the table of the research outputs committee of Research Councils UK (RCUK), the umbrella body for the seven UK research councils. There he serves as a special “expert” member of the committee.

In light of his professional interests, it is no surprise that Jeffery is a keen advocate for Open Access (OA). OA, after all, assumes a world in which research is made freely and immediately available on the Web, not locked behind a paywall and thus available only to those who can afford access.

But Jeffery is not an undiscriminating advocate for OA. He has firm views not just on its benefits, but also on how it can best be realised. Specifically, he is convinced that Gold OA — where authors pay to publish their papers in OA journals — risks taking the research community down a very expensive, unproductive and risky cul-de-sac.

A far more fruitful and rational approach, he suggests, is to push for Green OA, where researchers continue to publish in subscription journals and then self-archive copies of their papers in their institutional repositories.

As he puts it, “I am firmly pro-Green Libre, anti-Gold and very anti-Hybrid … My arguments against Gold relate predominantly to its cost, coverage, access, and the way it encourages vanity publishing.” 

Next logical stage

Only Green OA, he adds, can usher in the next logical stage in the development of scholarly communication, which Jeffery characterises as “hyperlinked multimedia and Web 2.0”.

Importantly, he says, by exploiting institutional repositories, research organisations can ensure that their intellectual property is “in a place where it is surrounded by additional relevant information”.  Moreover, when it is located directly in the institution, the information can be “managed and curated more appropriately, and within the research business processes of the organisation.”

With exactly this vision in mind, Jeffery has devoted a lot of time and effort (as lead architect) to developing a new information format able to implement his vision — an information standard known as the Common European Research Information Format, or CERIF.

CERIF, an EU Recommendation to member states now used in 42 countries, allows the use of rich metadata. This provides greatly improved retrieval capabilities, and enables institutions to create linkages between all relevant data — not just research papers, but information about the authors, research datasets and software, funding information, project information, experiments, and even data on the facilities and equipment used by researchers.

Given his belief that Green OA based on a network of sophisticated institutional repositories offers the most cost-effective and rapid migration path from print to electronic for scholarly communication, we should not be surprised that Jeffery is disappointed with the recommendations made in the recently published Finch Report.

Headed up by Dame Janet Finch, the Finch Committee was set up last year by UK Minister for Universities and Science David Willetts, and tasked with establishing how access to research can be expanded. 

Retrograde step

When it published its Report in June, the Finch Committee recommended that the research community single-mindedly pursue the gold route to OA, with institutional repositories consigned to bit players as access providers for data and grey literature, and as preservation tools.

Jeffery believes the RCUK will now be obliged to follow the Finch recommendations. This would be a retrograde step: earlier this year RCUK — which has had an OA policy since 2006 — released the draft text of a new, stronger OA policy. Amongst other things, this made a commitment to Libre OA and proposed that the Research Councils stipulate an embargo period of no more than 6 months (except in the case of humanities and social sciences).

By contrast, the Finch Report argues that it would be “unreasonable” to require embargo periods of less than twelve months.

And while Finch encourages non-commercial re-use, there is no similarly explicit support for full CC-BY licencing. As such, it will not facilitate the unrestricted re-use of articles, including text mining, which Jeffery believes to be an essential component of OA.

“[A]s an employee of a Research Council, I am obliged to support Finch, although privately I have grave misgivings,” he says.

Jeffery expands on his work, and his disappointment with the Finch Report, in more detail below. 

The interview begins ...

RP: You are a director at the UK Science & Technology Facilities Council (STFC). Can you say, briefly, what the STFC is and does?

KJ: The Science and Technology Research Council has two main roles. First, it is responsible for policy, strategy and grant awarding in the areas of particle physics and astronomy, as well as securing access to appropriate facilities such as CERN and various telescopes.

Second, it operates major facilities at its UK laboratories — and also at some overseas sites that it manages on behalf of the whole UK research community, and for which it has associated strategic and policy decision-making responsibilities.

RP: What is your role at STFC?

KJ: Until a few years ago, I was Director, IT running a service with 1,100 servers, 140 staff, 360,000 users and some 8Pb of data per year, along with an associated research operation to ensure provision of leading-edge ICT to support leading-edge science.

Having been responsible for setting up international consortia, I was asked to do the same for our non-ICT facilities (neutrons, lasers, electron synchrotron light). So today, I am Director IT & International Strategy, with a secondary title of Director, International Affairs (a title that is better understood in places like the European Commission).

Special expert

RP: Where does scholarly communication fit in this?

KJ: My motivation — since the sixties — has been the utilisation of ICT to support, enhance and assist in research and the management of research. I see scholarly publication as just one aspect of this, alongside controlling observations from satellites or the set-up of an experiment on a beam-line or even the simulation ‘in silico’ of a biomedical process.

RP: You are also a special “expert” member of the research outputs committee of Research Councils UK (RCUK). What does that involve?

KJ: I have been involved in OA for years. Furthermore, I have in the past developed systems for several of the Research Councils, and acted as design authority for the new Research Councils’ shared services centre.

When I was Director IT at STFC I was also responsible for the library, and we set up an early institutional repository. In addition, I was lead architect of CERIF, (Common European Research Information Format) — an EU Recommendation to member states now increasingly used in 42 countries.

Given my experience, it was felt that I was well qualified to advise the RCUK.

RP: You say you have been involved in OA for years. Just to be clear: you are an OA advocate.

KJ: Yes, I have been an OA advocate for almost two decades, although in the early days it was not known as OA.

RP: What in your view are the main benefits of OA?

KJ: The major benefits of OA as I see it are these:

(1) It provides access to research output information (not just scholarly publications) for anyone, anywhere, and in a timely manner and appropriate form;

(2) To the end-user it appears to be without cost (Gratis OA), and so maximises (re-)utilisation;

(3) It can provide access in a way that allows others to process it with their computers (not just read it with their eyeballs) i.e. Libre OA as an addition to Gratis OA. This is of some importance for the text mining of scholarly publications, but becomes more important when creating associated multimedia publications or datasets.

I would add that the privileges that OA provides (Gratis and Libre) come with responsibilities: the reader/user must acknowledge, cite or otherwise accredit the originator. I also believe that the evaluation of research output should take account of such accreditation.

Green or Gold

RP: OA comes in a number of different flavours. Do you favour a particular form of OA?

KJ: I am firmly pro-Green Libre, anti-Gold and very anti-Hybrid.

RP: As you indicated, Libre OA is essential for text mining. Some OA journals do provide Libre access, so what are your objections to Gold OA?

JK: My arguments against Gold relate predominantly to its cost, coverage, access, and the way it encourages vanity publishing.

RP: Let’s go through each of these objections separately. What is your concern about the cost of Gold OA?

KJ: It is this: for any reasonably productive research institution the cost of Gold is ~3 times that of Green (based on the cost of article processing charges, or APCs — which range from $1,000 to $10,000, with a mean of around $3,000). Assuming budgets are constant, this reduces the capacity for a research institution to generate and distribute its scholarly output.

RP: Do you have any figures to support your claim that Gold is around three times more expensive than Green?

KJ: I do. We did the calculation based on the circumstances at Rutherford Appleton Laboratory some years ago. Of course, the argument is swayed heavily by what is taken as the “average” APC and indeed predictions of how these may change in the coming years.

But we have all seen what commercial publishers have done with subscriptions and I do not expect their behaviour to be any different with the gold business model.

RP: You are saying that you expect the price of Gold OA to increase over time, creating a situation not unlike the serials crisis associated with subscription publishing?

KJ: Yes. Moreover, with the transition to Gold from conventional subscription-based scholarly publishing the commercial publishers will extract from the public research budget twice; first from the indirect budget that already pays for subscriptions, second from the direct budget in order to pay the costs of APCs.

This is likely to reduce the direct budget — which funds research activity in the universities — by something like 4%, with this money going straight to the publishers. 4% of the UK Research Council funding of ~£3 billion is a considerable amount of money.

RP: In other words, universities will have to continue paying subscriptions costs from their central budgets, while increasingly having to pay to publish papers using money that would otherwise be used to do research. Of course, many argue that this is only a transitional problem, but we can come back to that. First, the figure of 4% you cite: Is this a generally accepted figure? When I spoke to the Wellcome Trust’s Robert Kiley he estimated that paying for OA would devour 1.25% of research costs.

KJ: As I indicated above, estimates vary depending on the assumptions. But in general, Gold is costlier than the more widely existing (subscription + Green) model in the following cases:

(a) For high production institutions (because of the number of APCs that have to be paid);

(b) For publishing in research disciplines that have more expensive journals (there is variation between disciplines, with some APCs being much less expensive than others);

(c) For publishing in the most prestigious journals (which tend to set high APCs)

RP: What about coverage? What coverage problems do you see with Gold OA?

KJ: At present, some 4-5% of scholarly publications are published using the gold model. This will increase slowly at best and is unlikely ever to be the universal model of OA. In comparison, green coverage varies from 15% to around 85%, with a higher percentage where an OA mandate is in place.

RP: This too is likely to be a transition issue is it not?

KJ: Not really; I suspect the transition period to be long and never completed. Before the gold model ever becomes universal it will be overtaken by other less costly, more efficient, and more effective, processes driven partly by technology and partly by sociological changes.

RP: Again, let’s come back to this. You said there is also an access problem with Gold OA. That sounds counter-intuitive. What do you mean?

KJ: Publisher repositories are numerous and it is tedious to have to access each one separately and then search (using different user interfaces, facilities etc.) for what you want.

True, some now offer OAI-PMH harvesting and OAISTER access, but that requires using extremely primitive keyword-based searching with — in my opinion — poor relevance and precision. Institutional repositories offer the same facility more or less universally.

Moreover, institutional repositories are increasingly linked with CRIS (Current Research Information Systems) using CERIF. This provides rich metadata and consequently much improved retrieval capability. It also provides linkages to research datasets and software, and to funding information, project information, and even to data on facilities and equipment used. 


RP: Can you say something about CERIF?

KJ: CERIF was developed at the behest of the EC by a group of nationally-mandated experts. It is an EU recommendation to member states (i.e. a standard), is now used in 42 countries, and has become the national standard for research information in nine of them.

I have published quite extensively on the advantages of CERIF over flat metadata such as Dublin Core and similar standards. CERIF covers all the objects in the research domain; not just publications. By using base objects (person, organisation, publication, equipment...) and link objects (person employed by an organisation between date/time 1 and date/time 2) it represents the real world accurately (as a fully connected graph in first order logic).

There is more detail on CERIF at www.eurocris.org. It is commonly used in the term CERIF-CRIS meaning a Current Research Information System utilising CERIF.

RP: You said that Gold OA encourages vanity publishing. It is true that Jeffrey Beall’s index of predatory OA publishers now lists around 100 publishers. But might this not also be a transition problem?

KJ: I would say that to date only a small amount of this has been detected. The central issue is that Gold OA has a structural problem — basically, if the author pays then the author expects a service. This can lead readers/users to assume the scholarly contribution concerned has been properly peer reviewed, and so is valid, when in fact it may not have been.

Given the obvious temptations for publishers to maximise income and profit inherent in the gold business model this should be a concern for the community. However, let me say that if it is a transition problem no one would be happier than I would be to discover that it was.

RP: Can you say why you are so sceptical about Hybrid OA, the model by which researchers are able to pay to have their papers made OA even when publishing in a subscription journal?

KJ: Hybrid OA is a compromise that — to my mind — satisfies no stakeholder fully.

RP: Can you expand on that?

KJ: Hybrid is based on the principle that paying for visibility (publisher-provided OA) for your article through a kind of APC increases its impact, and it is postulated on the claim that publishers will reduce their subscription costs as their revenues from Hybrid fees increase.

In fact, I believe only two publishers running this model have reduced subscription costs. The Wellcome trust alerted the community to the danger of paying twice with the hybrid model in 2009.

RP: Ok, so for the reasons you have outlined, you are not a fan of Gold OA. Rather, you believe that Green OA offers a better solution. We discussed the issue of cost. What other advantages do you see Green providing that Gold does not?

KJ: My main argument for Green is that it places the intellectual property of the research organisation in a place where it is surrounded by additional relevant information. As noted, this can include research datasets, information about equipment, the expertise of people etc.; and by locating it directly in the institution it can be managed and curated more appropriately, and within the research business processes of the organisation.

RP:  And you believe that Green OA makes research more accessible?

KJ: Yes, in addition, it allows (especially when a repository of full text publications is linked with a CERIF-CRIS, or the publications are stored within the CRIS itself) the organisation to publicise (via web pages etc.) its output, and to generate easily and locally researcher CVs, bibliographies, and reports on research output as and when required by funders and other stakeholders.

Moreover, the existing peer-reviewing and publishing mechanisms associated with Green OA are already in place, well understood, and (more-or-less) respected.

Vitally, Green allows an evolutionary change to newer models of scholarly communication using more modern media, and driven by the community. It is a step between print and electronic that will better facilitate the hyperlinked multimedia with Web 2.0 that scholarly discourse will become next.

RCUK and Finch

RP: Can we look at the situation so far as RCUK is concerned. RCUK has had an OA policy in place since 2006. In March, it released the draft text of a new, stronger OA policy. Amongst other things, this made a commitment to Libre OA and proposed that the Research Councils 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 sciences). Was that a policy document you supported?

KJ: Briefly, yes. Personally, I see no justification for any embargo period on scholarly publications (although I do see their use for prior publication by research project teams in the case of datasets and/or software and/or research prototype artefacts). arXiv e-preprints are cited commonly, as is the arXiv later version of record; it does not seem to have harmed the publishers in particle physics and related disciplines.

I would add that Gratis OA is anachronistic in the internet age; Libre has to be the way forward.

By the way, the reaction of the publishers to the draft RCUK policy was rapid and aggressive (and in many cases inaccurate). We have to ask: does the community wish to partner with organisations like this in the business of scholarship?

RP: Is it not the case that papers archived by researchers in their institutional repositories by means of Green OA are far more likely to be made available as Gratis OA, not Libre OA?

KJ: There are two parts to this: are the articles used in a libre way, and can they be used in a libre way. The answer to the first is not commonly, but there are people experimenting in this area, including Peter Murray-Rust and teams elsewhere in Europe.

The answer to the second is that — unlike most gold publisher repository articles, under green conditions libre access is possible (anyone can use any software tool on any article) although some publishers are trying to forbid this and thus restrict research activity.

RP: I guess you are saying that while publishers may try to prevent researchers from doing it, text mining is not prohibited by copyright. But to go back to RCUK, three months after it released its draft proposal, the Finch Report — which was commissioned by the UK government to look at ways of expanding access to research publications — was published. What would you say were the major differences between the draft policy RCUK published in March, and the recommendations made by the Finch Committee in July?

KJ: The Finch Report has one positive message: that OA is essential. Thereafter, I find myself personally at odds with most of the reasoning and the conclusions — which strongly favour Gold OA.

But I should point out that RCUK — an umbrella organisation for seven research councils — is more in line with the Finch Report than I am personally.

RP: Why is that do you think?

KJ: It is partly because the research councils that fund biomedical research are already far down the gold track (and the central repository way of thinking); partly because of innate conservatism, based on adherence to the view that commercial publishing offers some sort of guarantee of quality; and partly because RCUK (being part of government administration and therefore cognisant of government thinking) recognises the value of the tax-take from UK-based commercial publishers.

RP: You are referring to the argument that says that publishers must be supported because they are important to the UK economy, and provide tax revenues for the government?

KJ: Indeed and I recognise that that is so. There is considerable value to UK plc from the commercial publishers.

However, I suspect the cost of slower research turnaround (embargo periods), reduced availability of research publications (that are locked behind paywalls, or possibly not published in the first place due to the cost of APCs), and the confusion that researchers experience when faced by a plethora of restrictions, rights and commercial models, all represent a much greater impediment to the wealth creation and improvement in the quality of life that research can provide.

And if the current models for research publication are extended to datasets and software the stranglehold on research production will be complete.

At odds

RP: With what aspects of the Finch Report are you most at odds?

KJ: Let me be clear: as an employee of a Research Council, I am obliged to support Finch, although privately I have grave misgivings.

I believe the Finch Report is driving scholarly communication into an expensive cul-de-sac without any vision of future models of scholarly communication. From the business model point of view, it is anachronistic: users are not charged for Facebook or Google.

From the utility point of view, it is backward looking: it leads to publisher silos instead of an interconnected high quality retrieval environment — one that when linked to CERIF-CRIS means that the whole context of research output is understood.

From the value for the public purse point of view, it is also a backward step, as I indicated when talking about the cost of Gold.

RP: Could it be that your views on Green OA are as influenced by your interest in CERIF-CRIS as publishers’ views are influenced by their financial interests in the current scholarly publishing model — which inclines them to prefer Gold OA. Both you and they have a vested interest in a particular outcome perhaps?

KJ: That is a good question. As I indicated, I am passionate about ICT supporting and enhancing all aspects of research.

However, CERIF-CRIS are used predominantly for the management of research by funders and universities (evaluation, benchmarking, management information decision-making, business intelligence, stakeholder relationship management), for the support of the researcher (automated CVs, bibliographies, webpages, finding collaborators), for the support of the innovator (presentation of new ideas with contact information) and for the general public — usually through the media (research stories).

Scholarly publications are a relatively minor (although important) part of the research environment and the CERIF-CRIS may well only hold metadata about the article while the article itself resides in a repository.

RP: So what follows the publication of the Finch Report? Presumably, the government will respond. But what happens after that?

KJ: There is already a lively debate. There will be strong lobbying from the scholarly community, but unfortunately this is likely to be incoherent due to varying factional views based on historical experience.

Meanwhile, there will be strong defensive lobbying from the commercial publishers. My guess is that the Finch recommendations will be accepted as a compromise between scholarly interests (the community gets OA) and the commercial publishers (who get a mandate for an even more lucrative business).

RP: I assume you anticipate that the RCUK will therefore need to adapt its draft OA policy to fit with the recommendations made in Finch?

KJ: Yes, already there is work going on concerning this.

However, there is generally — not only within RCUK — a deep discussion on the costs projected in the Finch report, and deeper analysis is being undertaken. This could result in the recommendations not being adopted.

Web 2.0 scholarly discourse

RP: It is certainly the case that the main discussion point surrounding the Finch Report right now is that of costs — both the transitional costs associated with migrating to OA and the question as to whether OA will prove more or less expensive than subscription publishing. You are clear that Gold will be more expensive?

KJ: I am. Green offers the best route to achieve what the public, the innovators, the scholars and the research managers (funders and universities) require because it is most effective, least costly and has only two barriers: the commercial aspirations of publishers, who generate FUD (fear, uncertainty, doubt) among academics over rights, refusals to publish, litigation etc., and the consequent Zeno’s Paralysis — as Stevan Harnad aptly describes it — that afflicts researchers. This prevents authors undertaking the few keystrokes required to achieve universal OA.

On the latter point, let me admit that the ICT systems used today could do more to assist researchers by, for instance, reducing the effort threshold for metadata input. That is why I crusade for CERIF-CRIS, where contextual information can be used to ‘pre-fill’ much of the required metadata for scholarly publications (and other research outputs).

One further word on transition: it is in the interests of the commercial publishers to maintain the dual system of subscription and Gold for as long as possible. Because of the extended timescale this will entail, I believe estimates on the costs of transition have been underestimated by a large factor.

RP: I assume you are not sympathetic to publishers’ claims that Green OA is parasitic, and could bankrupt them.

KJ: Parasitic on what? Academics do the research, author the publications, do the peer review, and usually do the editing — all free for the benefit of the community. The publishers organise peer reviewing and maybe set standards for editing and layout.

In fact, the costs paid by academia to publishers are predominantly for the benefit of having work appear in a reputable journal (a publication channel with high impact factor). Researchers want to publish in the same prestigious journals as members of their peer group do.

One final point I would make is that commercial publishing is not well aligned for the data deluge associated with future scholarly communication; nor is it well suited to the fast interaction between academics (and between academics and innovators) that will ensue.

RP: I guess this takes us back to the point you made earlier about how future scholarly discourse will need to operate in an environment of hyperlinked multimedia and Web 2.0 tools. You believe that Green OA offers a much better transitional path to this than does Gold OA. What strikes me about most discussion on the costs of OA is that it appears to be posited on the assumption that all that will change with OA is the way in which publishers are paid for the services they provide. When people talk about Gold OA being cheaper, therefore, they are saying no more than that in future APC costs will be lower than historic subscription costs. Not only do you disagree with the proposition that Gold OA will be cheaper, but you believe the claim is based on a misunderstanding of how scholarly communication is set to develop?

KJ: I prefer to see scholarly publishing as just one component of scholarly discourse. I would like to see all aspects integrated for the convenience of the researcher, research manager, innovator and media.

But to respond to your point about costs:

(a) I believe the transition will be long and costly, with double payments being made to publishers — first, through indirect payments for subscriptions; second, through payments for APCs, which will come directly from research funds;

(b) I believe the costs of Gold will be greater than the current subscription model: first, based on an extrapolation from the currently announced APCs; and second, because the experience of publisher subscription increases leads one to believe they will do the same with Gold once they have obtained a near-monopoly in the new environment.

RP:  Thank you for your time.


Marcin Wojnarski said...

Thanks for very interesting points. Can you elaborate more on the libre access to 'green' papers?

My understanding is that every publisher and journal can impose its own licensing terms on the authors. Even if they allow self-archiving, they usually do NOT allow redistribution of the paper (essential part of 'libre', right?). Moreover, every single paper in repository may come with a different license and different legal caveats buried inside, so the prospective reader would have to study a brand new license each time he aproaches new paper. Am I right?

How can 'libre' OA be possible in such case? And how is the problem of multiple licenses resolved?

Richard Poynder said...

Marcin, Peter Suber has commented on your question in this way:

"Marcin is generally right that repositories can't authorize libre green on their own, and therefore depend on permissions from others. That's why libre green is harder than libre gold.

"But libre green is not at all impossible, and in fact it's growing. See sections 2 and 5 of my article, ‘The rise of libre OA,’ in the June 2012 SOAN."

Keith G Jeffery said...

Marcin; the response from Peter via Richard is absolutely , correct. However there are some further twists. (1) it is worth checking carefully the contract of employers with an organisation for clauses on IP (Intellectual property). It may well be that author agreements with publishers are invalid and thus the organisation is free to do what it will with the green article; (2) an organisation could - on behalf of its employees as authors - negotiate with publishers (the ones used predominantly by its employees as authors) for libre rights on the green articles.

My hope is that by shaming the publishers on the basis of double-dipping, restricting the progres of research etc we may get movement. Recently even Elsevier moved in the face of researcher antipathy!

Stevan Harnad said...


The answer is so familiar, and so obvious, and so ignored:
Yes, universal Libre Green is better than universal Gratis Green. And perhaps universal Libre Gold is better still.
But we don't even have 22% Gratis Green yet. And Libre (Green or Gold) is much harder to get, and further from reach, than Gratis Green. And universal Gratis Green is the fastest and surest way to get not only 100% Gratis Green, but to eventually reach Libre Green and Gold too.
So why do we go on and on and on fussing about Libre and Gold instead of first mandating Gratis Green? It's like fussing over the lack of sustainable organic food and letting people keep starving, instead of feeding them non-organic food that's already available and enough to feed everyone for now.