Open Access Archivangelism

Professor Ian Diamond
Chair, RCUK Executive Group
Research Councils UK Secretariat
Polaris House , North Star Ave
Swindon SN2 1ET

Date: 22 August

Dear Professor Diamond,

We are responding to the public letter, addressed to yourself, by Sally Morris (Executive Director of ALPSP, the Association of Learned and Professional Society Publishers), concerning the RCUK's proposed research self-archiving policy.

ALPSP says that the RCUK policy would have 'disastrous consequences' for journals, yet all objective evidence is precisely contrary to this dire prediction. In the point-by-point rebuttal attached (below) to this letter, we document this on the basis of the actual data and a careful logical analysis. Here is a summary:

ALPSP argues that a policy of mandated self-archiving of research articles in freely accessible repositories, when combined with the ready retrievability of those articles through search engines (such as Google Scholar) and interoperability (facilitated by standards such as OAI-PMH), "will accelerate the move to a disastrous scenario".

The disastrous scenario predicted by ALPSP is that an RCUK mandate would cause libraries to cancel subscriptions, which would in turn lead to the financial failure of scholarly journals, and so to the collapse of the quality control and peer review process that publishers manage.

Not only are these claims unsubstantiated, but all the evidence to date shows the reverse to be true: not only do journals thrive and co-exist alongside author self-archiving, but they can actually benefit from it -- both in terms of more citations and more subscriptions.

Moreover, there is a logical contradiction in the position adopted by ALPSP. On the one hand, ALPSP maintains that learned societies must be allowed to operate in a free market ("each publisher must have the right to establish the best way of expanding access to its journal content that is compatible with continuing viability"). Yet on the other hand, ALPSP is in effect asking RCUK to protect learned societies from the consequences of a free market -- specifically the right of those who have funded and produced research to make their product readily accessible for uptake by its intended users.

What no one denies is that today many researchers are unable to access all the research they need to do their work. As ALPSP itself acknowledges, researchers already have to make use of author self-archived articles in order to gain access to "otherwise inaccessible published articles," since no research institution can afford to subscribe to all the journals its researchers need.

In short, due to the current constraints on the accessibility of research results, the potential of British scholarship is not being maximised currently. Yet the constraints on accessibility can now, in the digital age, be eliminated completely, to the benefit of the UK economy and society, exactly in the way RCUK has proposed.

For this reason, we believe that RCUK should go ahead and implement its immediate-self-archiving mandate, without further delay. That done, RCUK can meet with ALPSP and other interested parties to discuss and plan how the UK Institutional Repositories can collaborate with journals and their publishers in sharing the newfound benefits of maximising UK research access and impact.

(A point-by-point rebuttal is attached below. A longer analysis, signed also by some non-UK supporters, is at http://www.ecs.soton.ac.uk/~harnad/Temp/alpsp.doc )

Yours faithfully,

Professor Tim Berners-Lee (University of Southampton)
Professor Dave De Roure (University of Southampton)
Professor Stevan Harnad (University of Southampton)
Professor Nigel Shadbolt (University of Southampton)
Professor Derek Law (University of Strathclyde)
Dr. Peter Murray-Rust (University of Cambridge)
Professor Charles Oppenheim (Loughborough University)
Professor Yorick Wilks (University of Sheffield)

Point-by-point rebuttal:

ALPSP: a policy of mandated self-archiving of research articles in freely accessible repositories, when combined with the ready retrievability of those articles through search engines (such as Google Scholar) and interoperability (facilitated by standards such as OAI-PMH), will accelerate the move to a disastrous scenario.

This hypothesis has already been tested and the actual evidence affords not the slightest hint of any 'move to a disastrous scenario.' Self-archiving is most advanced in physics, hence that is the strongest test of where it is moving: Since 1991, hundreds of thousands of articles have been made freely accessible and readily retrievable by physicists using the open archive called arXiv; those articles have been extensively accessed, retrieved, used and cited by other researchers -- exactly as their authors intended. Yet when asked, both of the large physics learned societies (the Institute of Physics Publishing in the UK and the American Physical Society) responded very explicitly that they cannot identify any loss of subscriptions to their journals as a result of this critical mass of self-archived and readily retrievable physics articles (footnote 1).

ALPSP: Librarians will increasingly find that 'good enough' versions of a significant proportion of articles in journals are freely available; in a situation where they lack the funds to purchase all the content their users want, it is inconceivable that they would not seek to save money by cancelling subscriptions to those journals. As a result, those journals will die.

First, neither research topics nor research journals have national boundaries. RCUK-funded researchers publish articles in thousands of journals, and those articles represent the output of only a small fraction of the world's research population. It is therefore extremely unlikely that a 'significant proportion' of the articles in any particular journal will become freely available as a consequence of the RCUK policy.

Second, as we know, some physics journals already do contain a 'significant proportion' of articles that have been self-archived in the physics repository, arXiv -- yet librarians have not cancelled subscriptions: the journals continue to survive and thrive.

ALPSP: The consequences of the destruction of journals' viability are very serious. Not only will it become impossible to support the whole process of quality control, including (but not limited to) peer review, but in addition, the research community will lose all the other value and prestige which is added, for both author and reader, through inclusion in a highly rated journal with a clearly understood audience and rich online functionality

Wherever authors and readers value the rich online functionality added by publishers they will still wish to have access to the journal, either through personal subscriptions or through their libraries. This is obviously the case for the physics journals. Publishers who add significant value create a product that users and their institutions will pay for.

Researchers who cannot access the journal version, however -- because their institutions 'lack the funds to purchase all the content their users want' -- should not be denied access to the basic research results, which have always been given away for free by their authors (to their publishers, as well as to all requesters of reprints). Nor should those authors be denied the usage and impact of those users. Such limitations on access have always hampered the impact and progress of British scholarship.

ALPSP: We absolutely reject unsupported assertions that self-archiving in publicly accessible repositories does not and will not damage journals. Indeed, we are accumulating a growing body of evidence that the opposite is the case, even at this early stage.
For example:
[1] Increasingly, librarians are making use of COUNTER-compliant (and therefore comparable) usage statistics to guide their decisions to renew or cancel journals. The Institute of Physics Publishing is therefore concerned to see that article downloads from its site are significantly lower for those journals whose content is substantially replicated in the ArXiV repository than for those which are not.

And what is the evidence supporting the assertion that 'the opposite is the case' and journals are damaged? None. As we know, the Institute of Physics Publishing (like the American Physical Society) has already stated publicly that it cannot identify any loss of subscriptions as a result of 14 years of self-archiving by physicists (footnote 1). Moreover, institutional repository software developers are now working with publishers on ways to ensure that the usage of articles in repositories is credited to the publisher.

ALPSP: [2] Citation statistics and the resultant impact factors are of enormous importance to authors and their institutions; they also influence librarians' renewal/cancellation decisions. Both the Institute of Physics and the London Mathematical Society are therefore troubled to note an increasing tendency for authors to cite only the repository version of an article, without mentioning the journal in which it was later published.

Librarians' decisions to cancel or subscribe to journals are made on the basis of a variety of measures, citation statistics being just one of them (footnote 2). But self-archiving increases citations, so journals carrying self-archived articles will perform better under this measure.

Citing the canonical version of an article wherever possible is a matter of author best-practice; it is misleading to cite momentary lags in scholarliness as if they were an argument against self-archiving. All of this can and will be quite easily and naturally adjusted, partly through updated scholarly practice and partly through institutional and publisher repositories collaborating in a system of pooled and shared citation statistics -- all credited to the official published version, as proper scholarliness dictates. These are all just natural adaptations to the new medium.

ALPSP: [3] Evidence is also growing that free availability of content has a very rapid negative effect on subscriptions. Oxford University Press made the contents of Nucleic Acids Research freely available online six months after publication; subscription loss was much greater than in related journals where the content was free after a year...
[4] The BMJ Publishing Group has noted a similar effect...
[5] In the USA, the Institute for Operations Research and the Management Sciences ... made freely available on the Web... noted a subscriptions decline

In all three examples whole journals were made freely available, in their entirety, with all the added value and rich online functionality that a journal provides. This is not at all the same as the self-archiving of authors' drafts, which are simply the basic research results, provided by the author on a single-article basis. The latter, not the former, is the target of the proposed RCUK policy. It is hence highly misleading to cite the effects of the former as evidence of negative effects of the latter.

(And although the RCUK is not proposing to mandate whole-journal open access, it is worth noting that there is also plenty of evidence that journals have benefited from being made freely available: Molecular Biology of the Cell's (MBC's) subscriptions have grown steadily after free access was provided by its publisher, The American Society for Cell Biology (footnote 3). MBC also enjoys a high impact factor and healthy submissions by authors encouraged by the increased exposure their articles receive. The same has happened for journals published by other societies [footnote 4].)

ALPSP: In addition, it is increasingly clear that this is exactly how researchers are already using search engines such as Scirus and Google Scholar... 'At this point, my main use of both [Scirus and Google Scholar] is for finding free Web versions of otherwise inaccessible published articles... Both Scirus and Scholar were also useful for finding author-hosted article copies, preprints, e-prints, and other permutations of the same article.'

Scirus, Google Scholar and the other search engines that retrieve open access articles serve the research community by enabling researchers to find and access articles they would otherwise be unable to read because they are hidden behind subscription barriers. These services help to maximise research access, usage and impact, all to the benefit of British science and scholarship, exactly as their authors and their institutions and funders wish them to do.

ALPSP: In the light of this growing evidence of serious and irreversible damage, each publisher must have the right to establish the best way of expanding access to its journal content that is compatible with continuing viability.

So far no evidence of serious and irreversible damage inflicted by self-archiving has been presented by ALPSP. This is unsurprising, because none exists. Publishers should do what they can to expand access and remain viable. But they certainly have no right to prevent researchers, their institutions and their funders from expanding access to their research findings either -- nor to expect them to wait and see whether their publishers will one day maximise access for them.

ALPSP: This is not best achieved by mandating the earliest possible self-archiving, and thus forcing the adoption of untried and uncosted publishing practices.

Self-archiving -- and what the RCUK is mandating -- is not a publishing practice at all: it is an author practice. And it has been tried and tested -- with great success -- for over 15 years without 'forcing the adoption' of any 'untried and uncosted publishing practices.' What UK research needs now is more self-archiving, not more delay and counterfactual projections.

ALPSP: This in turn will deprive learned societies of an important income stream, without which many will be unable to support their other activities -- such as meetings, bursaries, research funding, public education and patient information -- which are of huge benefit both to their research communities and to the general public.

Please contrast this double-doomsday scenario ('self-archiving will not only destroy journals but all the other good works of learned societies') with the following quote from Dr Elizabeth Marincola, Executive Director of the American Society for Cell Biology, a sizeable but not huge society (10,000 members; many US scientific and medical societies have over 100,000):

"I think the more dependent societies are on their publications, the farther away they are from the real needs of their members. If they were really doing good work and their members were aware of this, then they wouldn't be so fearful...... When my colleagues come to me and say they couldn't possibly think of putting their publishing revenues at risk, I think 'why haven't you been diversifying your revenue sources all along and why haven't you been diversifying your products all along?' The ASCB offers a diverse range of products so that if publications were at risk financially, we wouldn't lose our membership base because there are lots of other reasons why people are members."3

This perfectly encapsulates why we should not be taking too seriously the dire warnings from those learned societies who warn that self-archiving will damage research and its dissemination. The dissemination of research findings should be a high-priority service for learned societies, but not a commercial end-in-itself that generates profit to subsidise other activities, at the expense of British research itself.

RCUK should go ahead and implement its immediate-self-archiving mandate, without any further delay, and then meet with ALPSP and other interested parties to discuss and plan how the UK Institutional Repositories can collaborate with journals and their publishers in pooling download and citation statistics, and in other other ways of sharing the benefits of maximising UK research access and impact.

References

1. Swan, A (2004). American Scientist Open Access Forum 3 February, 2005

2. Personal communication from a UK University Library Director: 'I know of no HE library where librarians make cancellation or subscription decisions. Typically they say to the department/faculty 'We have to save £X,000" from your share of the serials budget, what do you want to cut?'. These are seen as academic -- not metrics-driven -- judgements, and no librarian makes those academic judgements, as they are indefensible in Senate... [S]uch decisions are almost always wholly subjective, not objective, and have nothing to do with the existence or otherwise of repositories.'

3. The society lady: an interview with Elizabeth Marincola (2003) Open Access Now, October 6, 2003

4. Walker, T (2002) Two societies show how to profit by providing free access. Learned Publishing 15, 279-284.

Copies also sent to:

The Lord Sainsbury of Turville Parliamentary Under Secretary of State for Science and Innovation Department of Trade and Industry
Professor Sir Keith O'Nions Director General of Research Councils, Office of Science and Technology
Drs. Astrid Wissenburg, RCUK Secretariat
Professor Colin Blakemore, Medical Research Council
Frances Marsden, Arts and Humanities Research Council
Professor Julia Goodfellow, Biotechnology and Biological Research Council
Professor Richard Wade, Particle Physics & Astronomy Research Council
Professor Alan Thorpe, Natural Environment Research Council
Professor John O'Reilly, Engineering and Physical Sciences Research Council
Professor John Wood, Council for the Central Laboratory of the Research Councils
Andrea Powell, Chair of ALPSP Council (Director of Publishing, CAB International)

In: Re-engineering the Open Access Movement I: Addressing What is Currently Missing,Sherif Masoud (SM) wrote:

SM: "I suggest that the efforts of the open access (OA) advocates be utilized in a different way from the current one so that the open access potential successes are expedited. I propose that roughly one third of such efforts be directed towards increasing the use of current OA publications."

I am not sure whose efforts are being targeted here (OA advocates, libraries, publishers, researchers, universities, research-funders) nor how much effort any of these is actually expending for OA today. But the fact is that about 20% of research output is being made OA today (5% by publishing it in OA journals, 15% by self-archiving it in OA archives).

The objective would seem to be to raise that 20% to 100% as soon as possible, rather than to try to squeeze more usage and functionality out of the impoverished 20% that exists to date. For this, I would say that 100% of effort should be expended on inducing researchers to make 100% of their research articles OA.

SM: "Using the OA resources more extensively will result in higher awareness and impact for the OA publications. This will lead to readers (users) and authors demanding more OA publications and repositories. This increased demand will result in higher OA supply in the form of OA journals and repositories, which is the end goal of the OA movement."

This is what I too thought 6 years ago, when OAI and OpCit and Citebase and OAister were launched: Surely as the research community experiences the power and utility of OA as users (consumers) they will translate their satisfaction into their behaviour as authors (providers). But that didn't happen -- partly because of the limited functional potential of such sparse coverage (20%), no matter how deluxe, and partly because of the (wrong) perception that much more effort is entailed in being an OA provider rather than just an OA consumer.

This perception is wrong, and the number of keystrokes and minutes involved in doing one moderately intensive online search are comparable to the number of keystrokes and minutes involved in making one of your own articles OA by self-archiving it (Carr & Harnad 2005).

Yet not even the repeated demonstrations of the dramatic increase in citation impact for articles made OA by self-archiving (Brody & Harnad 2004) has yet proved sufficient to persuade nearly enough authors to self-archive .

The solution is clear, and has been clearly stated by authors themselves, in their responses to two international, interdisciplinary JISC surveys showing that most authors say they will not self-archive until and unless their employers and/or their research-funders require it -- but if/when they do require it, over 94% of authors say they will do it (81% of them willingly, 14% reluctantly, with only 5% saying they still will not do it).

So what is urgently needed worldwide today is definitely not a diversion of effort toward enhancing user functionality for the sparse 20% OA content that already exists, but a concerted effort to get researchers' employers and funders to require OA self-archiving. Research Councils UK looks as if it will adopt a policy of requiring OA self-archiving at the end of this month, and if they do, it is very likely that other institutions worldwide will follow suit. The blueprint for the policy requirement is already there:

RCUK Proposed Policy Requirement
UK Select Committee Proposed Policy Requirement
Berlin-3 Proposed Policy Requirement
Registry of Institutional OA Self-Archiving Policies

SM: "Reviewing a few Internet publications on open access [1], I found that almost every open access (OA) advocate's efforts are in one of two directions: analyzing, understanding, and explaining that OA is much more useful than closed access for everybody (except maybe for giant publishers) or promoting the foundation of OA serials / repositories and publishing in them. While the domination of OA serials and repositories form the end goal of the OA movement, I believe the current direction of efforts is not the best one to achieve that end goal in the fastest way possible. Re-engineering the open access movement to address what is currently missing can be very useful in expediting the domination of open access."

The end goal is 100% OA, and the two roads to 100% OA are the "golden" road of publishing articles in OA journals or the "green" road of self-archiving in OA archives all articles published in non-OA journals. Neither informing researchers about the benefits of OA nor showing them how to derive those benefits is enough: Their employers and funders also need to require OA provision (just as they already require publishing itself: "publish or perish") for the sake of maximizing individual and institutional research impact and maximizing the return on research funders' investment in research, in terms of research uptake, usage, impact and progress.

Diverting efforts instead toward increasing the functionality of the sparse OA content that already exists would simply delay still longer the optimal, inevitable, and already long overdue outcome: 100% OA. That (not something else) is the 100% doable thing that still remains 80% undone. That (not something else) is what needs doing. Endless diversions have kept it undone far too long already. Please, let us not keep finding more diversions. OA provision needs to be mandated by researchers' institutions and funders: "that is all ye know on earth, and all ye need to know."

SM: "I propose that only a certain part of the OA advocates' efforts be directed towards supporting the extensive use of OA publications. I don't mean a total change; I am saying that roughly two thirds of the efforts should stay in the current directions, but the rest should be in the direction of promoting the extensive use of OA publications. "

There is no point promoting the extensive use of sparse content! What is missing is the content, not the usage or the functionality. The greater usage of OA content is already fully demonstrated by its already-existing citation advantage of 50%-250+% as well its usage (download) advantage. Those few cows (20%) cannot be milked more fully! We need more cows (100%).

SM: "Amplifying the use and citations of the OA publications will result in higher awareness and impact for the OA literature. As the impact and awareness keep amplifying, the readers (users) and authors' demand for OA literature (serials and repositories) will intensify. This increased demand will result in higher OA supply (in the form of OA journals and repositories), which is the end goal of the OA movement."

But the usage and citations of OA articles are already greatly amplified; this has been happening for years, and has lately been demonstrated repeatedly, in field after field. These findings are useful, and will certainly persuade some more authors to provide OA to their work, but not nearly enough. What is needed is an OA-provision requirement from researchers' employers and funders. The employers and funders, in turn, are the ones who are being persuaded to adopt the policy of requiring OA-provision by the already amply abundant evidence that OA enhances research usage and impact. Is a 50%-250% advantage not already advantage enough?

SM: "So the question that rises is "how to promote the use of OA publications"? Well, the degree of using and citing OA literature is proportional to how easy it is to access that literature. Therefore OA advocates, both librarians and academic experts, have vital roles to play here. Librarians, the guardians of libraries, should be natural OA advocates because OA helps them mitigate the serials crisis. Hence, they should mine for OA resources all over the Internet and present it to their academic communities in an easy to use way."

But Sherif, the ease-of-access effect and the evidence for it are already there! (And although librarians are great allies to the OA movement, it is a great mistake to imagine that the rationale for researchers providing OA is to mitigate the serials crisis! The rationale is to maximize research usage, impact and progress.)

SM: "Innovative techniques are expected to achieve good results. I mean that the starting point is: the Internet is there containing all the open access publications and the search engines are there giving millions of page results to searchers. Of course a starting point like that --if it is also the final point-- is not very promising. Innovatively, the final point, could be a website that integrates all the sources in an institution's library and the Internet (e.g. open access journals) with a scope encompassing a researcher's specific point of research (e.g. gear design in mechanical engineering)."

This was the reason the OAI interoperability protocol was designed in 1999. Please have a look at OAister and Citebase and Google Scholar. The functionality is already there (and being heavily used). What is missing is 80% of the target content!

SM: "That's only one innovative example; I am sure there are many creative ways in which librarians can catalog Internet OA content to make them easy to use. Let's start thinking that not properly guiding the library patrons to available OA resources is analogous to having very generous donators that donate lots of publications to a library and also pay for storing them to be only left in the warehouse without being added to the library catalog (although the keys of the warehouse are available for patrons upon request for digging on their own)."

Sherif, I'm afraid these are rather old-fashioned views on functionality in the online age. What users need today is not more cataloguing or help from librarians: They need more OA content (preferably OAI-compliant): 80% more. That's all. The rest will take care of itself.

SM: "On the other hand, as experts in precise fields of research, academic OA advocates are also urged --individually or collaboratively-- to form Internet research guides on their fields of studies. "

Why? Why produce guides to a sparse 20% of content? And why have guides at all, when OAI full-text content can be inverted and searched boole/google-style, with the help of citation-links and -metrics as well as similarity-metrics? What's missing today, and urgently needed, is more OA content, not more OA visibility/classification/navigation resources.

SM: "Another point, the design of such Internet research guides should always be user-centered. That means for the user (researcher or reader) a closed access database (e.g. only table of contents, abstracts, or even pay-per view) could be very useful along with OA publications."

Anyone who wishes to build value-added services on top of OA content is welcome to do so -- but first we need the OA content, otherwise there is precious little to add the value to.

SM: "I mean finally what will get the impact and awareness is open access, but for the research guides to be successful in attracting researchers they have to be hybrid in order to fulfil all of the researchers' requirements. Along with that, the benefits of open access should always be advertised, so that researchers not only know about them practically (when using the OA movement research guides) but also understand the theory behind them (how this is happening). Needless to say, the OA advertising should never annoy or disturb the users; it should come naturally. Moreover, it will be easy to recruit at least some of the users of those research guides to the OA movement, to advocate for open access themselves later. To illustrate what I am suggesting, [2] represents an attempt of creating an Internet research guide that integrates both OA and other forms of learning sources, coupling that to non-destructive OA advertising."

There has been a modest flow from OA usage/consumption to OA provision, indeed that is part of the source of the existing 20% OA content. But it is clear that nowhere near enough OA content can be expected (within the visible future) via that slow and indirect route. Something far more substantive is needed to end the already absurdly long delay in reaching 100% OA, and that something is what authors themselves have said they would need in order to induce them to provide OA: a policy of requiring it by their employers and/or funders.

SM: "To optimally achieve open access in scholarly publishing, the scope of efforts should encompass every possible way of open access (not only scholarly publishing)."

OA is not OA-in-scholarly-publishing, it is OA-to-scholarly-publications. OA is no more a reform of scholarly publishing than it is a remedy for the serials crisis. It is a way for researchers to maximize the usage and impact of their own research publications (whether the golden way, by publishing them in OA journals, or the green way, by publishing them in non-OA journals but also self-archiving them in OA Archives; Harnad, Brody, Vallieres, Carr, Gingras, Oppenheim, Stamerjohanns & Hilf 2004).

SM: "I propose that the efforts in this direction be at all levels (e.g. for students, researchers, and practitioners). There are mutual interactions among those groups, so the more open access is appreciated among, for example, college students, the more it will be among researchers. One reason is that the students of today are the researchers of the near future. Further, creative traditional print books on open access are needed. The traditional books still have their audience that respect them and see them as almost the only authority in any topic. The open access movement needs to address such an audience."

If we wait for students to come of age in order to reach 100% OA we are waiting absurdly long, and needlessly losing a great deal of potential research impact and progress in the meanwhile. Researchers are happy for students and practitioners to read and use their research, but they are writing it mainly for their fellow researchers (that's what their careers and funding depend on): It is their fellow researchers who can build on their research, thereby extending its impact. Researchers will not be persuaded to provide OA by evidence of enhanced usage by students (or practitioners, in that minority of research fields where there even are practitioners at all!); after all, they are not yet even sufficiently persuaded by the evidence of enhanced usage and impact by and for their fellow-researchers! The persuasion will have to come from the same source that persuades them to publish in the first place (rather than put their research findings in a desk-drawer and move on to the next piece of curiosity-driven, solipsistic research): the persuasion will have to come from their employers and funders.

SM: "A proposal for re-engineering the open access movement by re-directing the current efforts of the OA advocates was presented. It is hoped that this proposal will raise a lot of discussions, ultimately resulting in substantial improvements"

If everyone who wishes to invest some time and effort into promoting OA were to invest it, first, into (1) self-archiving 100% of their own articles, and, second, into (2) persuading their own institution to adopt a policy of requiring self-archiving for 100% of its research article output, that time and effort would be immeasurably better spent than investing it in trying to enhance the functionality of the impoverished 20% OA content we have today.