Scholarly Associations and the Economic Viability of Open Access Publishing: Willinsky: JoDI

Scholarly Associations and the Economic Viability of Open Access Publishing

John Willinsky
University of British Columbia, Vancouver, Canada
Email: john.willinsky@ubc.ca

Abstract

The paper considers a number of economic issues that scholarly associations are confronting in moving their journals online, with a focus on the possible viability of an open access or free-to-read format. It explores the current content overlap between subscription-based and open access sources, and considers how these redundancies favor open access publishing and indexing. It utilizes the tax returns for 20 US non-profit scholarly associations to analyze current publishing revenues against costs, arguing that the associations could make up the loss of revenue posed by the open access publishing model through cost savings and other revenue sources, while serving their membership better through the increased readership in an era of declining subscriptions. While the decision to publish journals in an open access format is by no means simply an economic one, the viability of open access publishing warrants serious consideration by scholarly associations that are currently determining what this new medium may mean for the circulation of knowledge.

1 Introduction

After a decade of Internet scholarly publishing, roughly two-thirds of the current academic journals have gone online, even as they continue to publish in print, while more than 1000 peer-reviewed journals are going it alone in digital form (Tenopir and King 2001). In moving online, scholarly publishing appears to have taken the next in a long line of steps to increase the circulation of this particular form of knowledge. Yet rather than imagine that advances in knowledge naturally unfold with each new communication technology, it is well to realize that with each step, significant choices made by key players during the formative period of the technology do much to shape the future of each publishing medium. Just so, in this transitional and economically unsustainable period, as journals continue to publish in print and move into electronic formats, critical decisions on the economics of this public good are now being made.

While most journals have carried forward some version of print's subscription model (with pay-per-view and site licenses variations) into this new medium, some are pursuing a new model for the dissemination of knowledge known as open access. Open access journals, which may currently make up 10-20 percent of online journals by some estimates, provide free access to their entire contents.[1] Open access publishing makes a public virtue of the global computer network through which the research is distributed. In addition to open access journals, there are open access indexes, which are also discussed in this article, as well as open access eprint archives, in which researchers deposit copies of their work which may be published in fee-based journals (Hitchcock et al. 2002; Pinfield 2003). What has driven this development of open access publishing is not only the availability of new technologies, but a desire among researchers and scholarly associations to bring some relief to the decades-long "serials crisis" which has eroded library access to journals as a result of increasing subscription costs.[2] The open access model is also seen as a way to improve public, educational, and political impact of research (Willinsky 2002a, Willinsky 2003). It is part, then, of a larger movement to defend the public sphere within the Internet, ensuring that it serves the democratic "right to know" when it comes to governance, health, and other areas of public concern.

When it comes to deciding the place of open access publishing in the future of scholarly publishing, the scholarly associations are obviously key players. They have long been at the heart of academic journal publishing, and they are concerned with the public face (and funding) of the various disciplines which they represent. They now face critical decisions on how best to use this digital medium to further the scholarly interests of their members. At this point, some associations have begun offering various forms of open access to the electronic editions of their journals.[3] Others have continued to charge fees for their online journals and yet others have turned their journals over to commercial publishers, which have already invested heavily in electronic delivery systems.[4]

To assist these scholarly associations and their members in making the critical decisions that lie ahead for the shape of this online knowledge economy for the research literature, I explore economic facets of the current academic publishing situation, with an eye to the arguments in favor of open access journal publishing. For at this point, I would argue, there are a good number of contradictory and redundant economies at work in the circulation of this knowledge. While it is undoubtedly tempting to simply continue business as usual, with subscriptions and licensing agreements providing the scholarly associations with a welcome source of income, the question I want to ask is whether this best serves what should be the first consideration of every scholarly society, namely to advance the professional well being of its members. The journal does that by providing members and others working in a given field with access to a reputable and rigorous forum for advancing knowledge.

Restricting access to the journals, through subscription fees, even combined with delayed open access, six months to a year after publication, reduces the readership and level of citations that would be otherwise be available through open access publishing, as has been established by at least one substantial study (Lawrence 2001). For similar reasons, as I have argued elsewhere, this fee-based model now runs contrary to the spirit of copyright law, which is designed to protect the interests of both creators and public, which are both served by the widest possible readership (Willinsky 2002b). And beyond matters of reputation-building (with its direct salary benefits within the university job market) lies the humanitarian interests of the members of the scholarly association, as they think of their work as a public good, and as they would not knowingly deny a colleague in the developing world access to their work if that access could be readily obtained.

The scholarly association has, then, to put the question to its membership: Is this organization devoted to maintaining its current revenue levels or is it devoted to serving the professional interests of its members in fostering the greater development and circulation of knowledge? So it is that I wish to consider why it may no longer make the sense that it once did in a print culture for scholarly associations to charge subscription fees for its journals, when alternative funding models exist and still others await development, based on the principle that open access to research furthers its role as a public good.

2 Redundant Economies

First, consider the current transitional state of the publishing economy. As it is divided between print and electronic forms, a good deal of overlap and redundancy exists, some of which reflects the interests of authors in providing open access to their work. The argument will grow thick with dollars and sense, but the numbers are only meant to lend some precision to the broader principles of a system in a state of change, a system that is thus open to alternative ways of circulating knowledge. The American Astronomical Society provides an excellent example. It publishes three journals which, taken together, contributed $5,834,020 in revenue to this non-profit society's total budget of $8,255,845 in 1999 (see Table 1).

The society's two principal journals, the Astronomical Journal and the Astrophysical Journal, are published by the University of Chicago Press, while the Society itself publishes the Bulletin of the American Astronomical Society.[5] Membership of the AAS costs $110 annually, and while that provides a free copy of Physics Today and a number of newsletters, it offers no more than a greatly reduced rate of $50 for electronic access to the two journals and bulletin.[6] In this transition period of overlapping publishing media, the AAS journals demonstrate scholarship's current double economy, not only as they publish in print and electronic formats, but as the content of the journals exists in proprietary and open access forms.

Take David Rusin's paper, "The Expected Properties of Dark Lenses", for example. It was published in a recent issue of the Astrophysical Journal. He submitted it to the journal in November 2001, and on having it accepted in February 2002 he then posted a copy to the arXiv.org eprint archive. After taking the paper through the editorial process with Astrophysical Journal, which was to publish the paper in June, he then updated the arXiv.org version, on May 2 with the comment: "Final version, minor corrections, 18 pages, ApJ June 20 2002". What this meant was that the paper was simultaneously available in both the University of Chicago Press' Astrophysical Journal (in both print and electronic forms) and in arXiv.org. Further complicating this doubling up of scholarly publishing is the fact that most AAS members would have access to Rusin's article not only through their reduced membership rate but also through the university library's subscription to the journal.

Nor does that exhaust current redundancies in scholarly publishing. Indexing of the research literature is no less divided between open access and fee-based sources, creating a series of overlapping services. On the open access side, arXiv.org indexes Rusin's paper, and it can be located using an Open Archives Initiative search engine, such as OIAster. Rusin's article has also been indexed by NASA's Astrophysics Data System, which is an open access service that, in the case of Rusin's paper, directs readers to the Astrophysical Journal site at the University of Chicago Press. A third open access indexing service, NEC ResearchIndex, lists papers that cite a given article, as well as listing works that it cites. While ResearchIndex carries other Rusin papers, it does not yet possess "The Expected Properties of Dark Lenses". The latest development in open access citation indexing is Citebase, which although still in the "experimental demonstration" stage, indexes Rusin's paper, providing a graph of the article's citation history, as well as a "top five" and complete listings of articles that cite Rusin's article and are co-cited with it (Hitchcock et al. 2002). Among the subscription-based indexing services, ISI Web of Science lists "The Expected Properties of Dark Lenses", featuring the abstract and a list of works cited by Rusin's paper, and a listing of the works that cite his paper and are linked to their own entries in Web of Science. [7]

The overlap of indexing services is not, in itself, undesirable for reader or author. However, given that a service like the Web of Science can cost well over $100,000 annually, incentives exist for exploring more efficient and affordable ways of providing indexed access to the research literature (Willinsky and Wolfson 2001). Although indexing is not something which scholarly associations have typically taken on, with some exceptions such as the American Chemical Society and its Chemical Abstracts, online publishing changes the situation. It is now possible to build autonomous indexing into the publishing process by adhering to, for example, the Open Archives Initiative Protocol for Metadata Harvesting. This protocol makes it possible for an association's journals to be automatically and immediately indexed by having authors fill in the necessary information (including title, author, abstract, subjects, method, etc.) on submitting their work, with the resulting "metadata" automatically harvested by an Open Archives search engine for indexing and search purposes.

In summary, the American Astronomical Society's electronic publishing strategy with the University of Chicago Press has obviously been compromised by the presence of open access publishing of at least some of the content. The Society's members and subscribers are paying for an exclusive service which they are no longer getting on an exclusive basis, even as its member-contributors are deciding that it is in their best interests, in some instances, to seek the broadest possible audience to their work by filing a copy of their published paper in an open access archive for others to freely consult. The library would seem to bear the brunt of this redundancy, as it subscribes to journals and indexing services, even as the viability of open access to the research literature and its indexing is being demonstrated.

In light of current developments for journal publishing, the scholarly associations have two critical factors to consider in thinking about the viability of open access publishing. The first is that free or discounted subscriptions to the association's journals will prove less and less an incentive to membership in the association. The second is that there are both actual and potential means of covering the association's impending loss of library subscription fees with open access publishing, and thus protecting the financial standing of the association.

3 Membership Subscription Benefits

The principal benefit of joining most scholarly associations has been the "free" or discounted subscription that comes with it. This benefit has always been somewhat compromised, for most members, by the availability of the journal in the member's university library. To now have an electronic version of the journal available at home or the office only further reduces the value of this membership benefit. For example, at my own institution, the University of British Columbia, the library provides faculty members with online desktop access to journals from 13 of the 20 scholarly societies listed in Table 1. The number is bound to increase as more associations provide electronic editions of their journals. Although most associations will continue to publish print versions of their main journals for some time, it is surely only a matter of time, given the greater ease of searching, the completeness of archives, the cheaper costs -  advantages for both libraries and researchers - before this dual publication mode no longer makes any sense. At that point, having a membership subscription will no longer have any meaning for those who have access to a substantial research library, whether the association's journals are open access or not.>[8]

Table 1. Annual publications costs and revenues of a selection of American scholarly associations, in 1999 or 2000


Scholarly Association 
(N=20)
Membership
Total Revenue
A
Publication Revenuea
B Royalties
C
Publication Costs
(A + B) - C
Revenue
over Costs
Academy of Political Science+*
10,000
679,894 
12,128
16,434 
265,878
(237,316)
African Studies Association
1,600
482,482 
41,676 
429,131 
(287,455)
American Anthropological Association+*
6,000
4,680,764 
637,950 
6,679 
790,133 
(145,504)
American Astronomical Society*
2,400
8,255,845 
6,055,176^
5,834,020 
221,156
American Education Research Association+*
28,300
5,462,350 
591,011 
27,744 
823,538 
(204,783)
American Economic Association*
18,000
4,501,541
1,685,640
0
3,974,715
(2,289,075)
American Fed. for Medical Research
3,500
904,399 
73,084^ 
70,139 
231,401 
(88,178)
American Historical Association*
7,000
3,377,258 
721,115 
22,464 
1,059,900 
(316,321)
American Political Science Association
10,500
5,994,628 
126,740 
119,906 
853,206 
(606,560)
American Psychological Society+*
8,300
2,248,227 
211,648^ 
328,765 
(117,117)
American Society for Cell Biology*
4,400
4,718,681 
1,228,161^ 
121,767 
1,195,934 
153,994
American Soc. for Information Science*
3,300
1,111,006 
153,314^ 
213,923 
(60,609)
American Soc. of Human Genetics
2,700
3,203,602 
1,251,617^ 
66,831 
1,120,170 
198,278
American Studies Association*
3,300
653,727 
194,683^ 
691 
201,262 
(5,888)
Cognitive Science Society+*
1,200
109,687
3,496
1,892
83,152
(77,764)
History of Science Society+
3,000 c
99,825 
201,500^ 
141,007 
60,493
Inter. Assoc. for Feminist Economics+
500
0
0^
6,872
71,718
(64,846)
Linguistic Society of America*
7,000
709,520 
5,319 
29,801 
228,526 
(193,406)
Microscopy Society of America+
700
1,058,897 
74,187^ 
94,314 
(20,127)
National Reading Conference 
1,000
302,964 
76,809 
29,888
84,557 
22,140
Radiation Research Society+*
1,100
919,855 
497,085 
347,595 
149,490
Averages
5,878
2,362,777 
659,159
24,815
874,897 
(190,924)

Note: Data drawn from Internal Revenue Service Form 990 for 1999 or 2000, available from GuideStar.org. Membership numbers are an approximation based on total membership dues divided by individual member fee levels (not shown).

a Publication Revenue does not include membership dues, while Total Revenue does.
b Combines journal and conference expenses.
c Includes institution and individual memberships.
+ Data is from year 2000 IRS Form 900; all other associations have data drawn from 1999 IRS Form 900.
* Association's journal(s) available online through the University of British Columbia Library.
^ Association's journal(s) published by university press or commercial publisher.
 

This needs to be seen as part of a longer-term change in reading habits of researchers. Scientists now do a third of their reading in electronic form, according to Tenopir and King (2001), drawing on a broader range of journals than a decade ago, when they would typically read through much of a single journal. Tenopir and King report that "the average number of personal subscriptions per scientist has roughly halved over 20 years".[9] This online search and browse method, rather than sticking with a few well-read journals, also prevailed in the survey of a thousand-plus scholars by the Association of Learned and Scholarly Publishers, just as it found that these scholars hoped that journals could be free in the future (ALPSP 2002). The scholarly association will clearly have to develop a new basis for attracting and retaining members, especially given the more widespread phenomenon of declining memberships in organizations of all sorts (Putman 2000).

Scholarly associations offer other benefits to members, such as the scholarly conference, workshops, and other events (which generate revenue), as well as conducting lobbying efforts, bestowing awards, publishing prestigious journals and offering an opportunity to serve the profession in notable ways. Participation in scholarly societies enables faculty members to establish a professional community and exercise leadership within that community. There is more to a scholarly association than subscription benefits, as was found by the Stanford e-journal user survey, eJUSt (2002), which discovered, with over 10,000 participants, that the "most popular reason for joining societies was to support the society's mission, but the second and the third most frequent motivations given were economic benefits - receiving journals free or discounted with memberships and attending conferences at a reduced rate".

While scholarly associations that opt for open access publishing are unlikely to be able to increase membership fees to cover the loss of publishing revenues and memberships, they may be able to maintain current fees (typically between $60-200), on the grounds that they are furthering their mission through open access publishing, by increasing journal readership, while providing public access to the benefit of students and scholars in developing countries, as well as other interested readers. The subscription model of membership does not have much of a future, no matter which way the association turns with the economics of electronic publishing, while open access publishing promotes the association's larger goals. That still leaves the question of how to make up for the lost revenues that comes with open access publishing.

4 Publishing Revenues and Costs

The scholarly associations have long been able to count on the revenue generated by library and other institutional subscriptions. It will take some understanding of the association publishing revenues and costs to see how the association could possibly publish their journals in an open access format without suffering a substantial financial loss. To that end, I gathered financial information from the tax forms of 20 US scholarly associations, distributed across a range of disciplines, with this data supplemented by email correspondence with the financial officers of a number of associations (Table 1).[10]

The publishing revenues and costs reported on the tax forms may incorporate the sale of books and other materials, as noted on the forms. With journals the costs include support of the editors, the editorial offices, copy-editing, composition, printing, and distribution, as John J. Siegfried, Secretary-Treasurer of the American Economic Association, explained in an email to me. Some associations use a commercial publisher to produce their journals, with the fees paid to the publisher constituting publishing costs.

For example, the International Association for Feminist Economics has a contract with Taylor and Francis to publish Feminist Economics. Taylor and Francis charges the Association their "publishing costs" for producing the journal, minus 25 percent in return for the Association's endorsement and editing of the journal. The American Psychological Society, on the other hand, pays Blackwell a flat fee of $20 per member for publishing their two principal journals, Psychological Science and Current Directions in Psychological Science. Elsevier charges the Cognitive Science Society $36 per member for the journal Cognitive Science, which otherwise costs $140 annually for individuals and $335 for institutional subscriptions, while Kluwer charges $50 per member to the Association for Computing in the Humanities for their journal, Computing and the Humanities, with annual subscriptions otherwise $167 for individuals and $411 for institutions.

Some scholarly associations not only cover their publication costs but generate a profit  - up to 43 percent - by selling subscriptions to libraries and others.[11] This is not typically the case, however, and most end up subsidizing some portion of the cost of providing members with the journals. On average, the associations have subscription revenues of $659,159 (with an additional $24,815 in royalties) against total publishing costs of $874,897. This means that subscription sales of journals, outside of memberships, pays for three-quarters of the associations' publishing costs. Or in terms of the open access publishing model, the average society would have to reduce its publishing costs by 75 percent to compensate for the loss of library subscriptions. To leave its budget relatively unchanged in moving to open access publishing of its journals, an association would, on average, have to find a way to cover 75 percent of its current publishing costs, either by reducing those costs by that much or by finding an alternative source, other than library subscription fees, or through some combination of the two.

One notorious difficulty in calculating how much, if any, can be saved by moving completely to an electronic edition of a journal is trying to ascertain what it costs, given that the estimates vary so wildly.[12] Tenopir and King (2001) point out that electronic publishing, up to this point, has not meant a great saving for journals: "Electronic access avoids these costs [for printing and distribution], but has a substantial additional fixed cost - putting up full text on the web, staffing, software and other technology issues including design, functionality, searchability and speed". While printing and distribution costs are on a fixed, per item, basis, the costs for the software that replace them are absorbed by the developers in the open source software model that is recommended below.

What is clear, however, is that any reduction in publishing costs requires phasing out the print edition, and eliminating the expenses related to the handling of the associated paper manuscripts. With the print edition gone, related printing, distribution, photocopying, postage, courier and subscription management costs can also be dispensed with (Table 2).[13] Further savings can be achieved as more of the journal management is supported by the use of open source management systems which, for example, automate aspects of indexing (with the support of authors providing an initial set of metadata about their submission), filing, and archiving of manuscripts, as well as handling the clerical aspects of the correspondence with authors, reviewers and others.

Table 2. Savings with automated e-journal management systems, compared with traditional print journals
Stage
Agent
Automated and Assisted Journal Management 
Savings
Submission Author a) Online upload in variety of formats, including "camera-ready" tables and figures, as well as room for appendices, data, instruments, etc.
b) Templates to assist authors indexing their papers by asking them to provide appropriate metadata
Clerical time, copying, postage, courier
Submission  Editor a) Author notified of receipt of submission
b) Queuing for review
stationery, editor time
Peer Review Editor a) Maintain list of reviewers, interests, record
b) Contact selected reviewers with abstracts
d) Provide access to paper, and reminders
e) Tracking review progress (viewable by author)
 
Peer Review  Reviewer Management of review comments and marked copy  
Editor Review Editor Author notification, with reviews (complete or excerpts) and judgement  
Revisions Editor  a) Ready back and forth with manuscript
b) Re-circulate paper to reviewers, if needed
 
Editing  Copyeditor

Proofreader

a) Editor and author ready access to manuscript, re. queries
b) Preparation of manuscript for publishing
 
Layout Editor Revert to commercial software, such as MS Word, to convert word processed document to HTML. Including text, footnotes, references, appendices Printing services, time
Publishing  Editor a) Ability to include and order articles
b) Volume and number assignment
 
Distribution Editor Automated, email notification of contents for readers Postage, packaging, time
Indexing  Author / Reader a) Automated harvesting of author-supplied metadata for each article by Open Archives Initiative engine creating a distributed global index.
b) Citation indexes and tracking with hyperlinking of citations to sources and of article to materials in related databases, using metadata.*
Indexing services (purchased separately by library),
time
Interchange Readers

Authors

Posting of comments (with editor as moderator) for continuing open peer review, as well as online forum for continuing exchange on range of themes Not otherwise available
Archiving Host library and/or PKP Server maintenance and backup, software upgrading/migration Cataloguing storage

* See Public Knowledge Project (Willinsky and Wolfson 2001), ResearchIndex (Lawrence et al. 1999), Citebase (Hitchcock et al. 2002), and BibP (Cameron and Tatu 2000).
 

All of this is being achieved, to varying degrees, with a new generation of journal management systems. While commercial versions of this software have become common among the major publishers of journals, the US National Science Foundation (NSF) has supported the development costs of such systems, as well as the digitization of current archives, perhaps most notably in joining with the Department of Energy and the Los Alamos National Laboratory to provide $300,000 annually to arXiv.org (Kling et al. 2002).[14] The NSF has been further encouraged by the American Association for the Advancement of Science to "fund experiments intended to bolster alternative models of licensing and publication" with a goal of promoting "wide access to and the preservation of scientific information in a cost-effective way" (Frankel 2002, p. 25).

Over the last two years I have assembled a team to build an open source version of the necessary journal management and publishing software, which is now distributed free through the Public Knowledge Project at the University of British Columbia. It took roughly $60,000 in government and foundation funding, and the resulting Open Journal Systems can be downloaded to a Web server in a university library or scholarly association, where it can manage and publish journals tailored to the policies and standards of the editors, with the cost savings represented in Table 2.[15] Once installed, Open Journal Systems requires little technical knowledge on the part of the editors, as it uses templates for setting up the journal Web site to managing the review process. As an open source system, it has already attracted a small community of users and supporters who have assisted in its further development, and work is currently going on in Africa and India exploring its feasibility for developing publishing and research capacity there (Muthayan and Muinde 2003).[16] We look forward to it proving that it can reduce costs in the way planned.[17]

Whatever systems are deployed to support open access publishing, the critical economic factors are that

  1. the authors, reviewers, and editors already donate their time to making journals the leading source of scientific information
  2. universities contribute overheads to journals (while possessing the infrastructure, particularly in their libraries, for hosting and archiving the journals associated with their faculty)
  3. open source publishing software now exists that can greatly reduce the cost of running a journal

The one expense that escapes automated systems, and often goes beyond the volunteered time of editors, is copyediting and proof-reading, which may run to $10,000 annually for a typical quarterly journal. Yet the average figure for publishing costs (Table 1) that would remain in the scholarly association budgets, after projected reductions and alternative sources of revenue (see below), is $190,924. While the possibilities of drastic reductions in publishing costs are clearly in need of testing, the scholarly associations have sufficient grounds, I am arguing, to explore what it would take to make a go of open access publishing.

It may seem odd to suggest that the scholarly publishing enterprise can be run by a series of editors working from networked computers in their offices, airports, and Internet cafés. Yet it is important to see that this is not really draining the publishing system of money or support. Nor is it demanding a hyperskilling of journal editors, just so a few more people can have access to this literature. To take such a position would be to overlook just how much money and training have already been invested by editors, authors and reviewers over the last decade in acquiring increasingly sophisticated computers and the rudimentary skills to run such management systems (which include word-processing, Web browsing, and emailing). The technology of knowledge reproduction and diffusion has been transferred from the print shop to the desktops of the professorship. Authors, editors and reviewers taking on this additional step - authors, for example, are asked to upload a paper online instead of printing it out, photocopying it and mailing it out - will result in the immediate benefits, insofar as it makes open access publishing possible, of a greater circulation of research, adding to its value as both science and a public good.

5 Viability of Open Access

To sum up the argument so far, the current transition of scholarly journals, as they move from print to the online medium, is marked by emerging economies divided between open and fee-based access. The transition and sorting out of economic models is leading to redundancies in publishing and indexing which are not providing relief to libraries nor delivering the greater global access to knowledge that appears to be within reach. At the same time, the scholarly associations face a loss of value for their membership subscriptions to their journals, while the fee-based model means a reduced readership, in comparison with open access publishing, for the work of its members and others in the discipline.

Yet to move to open access publishing, without suffering too severe a drop in existing scholarly association budgets, requires some way to make up 75 percent of current publishing costs, by reduction of those costs or by some other means. While new automated systems could substantially reduce those costs, the scholarly associations have other means of making the open access model economically viable. Crow and Goldstein (2003) present close to 20 possible sources of income and subsidies, including author fees, conference co-hosting and value-added services.

To take the author-fee model, BioMed Central, one of the few corporate publishers providing open access to its journal content, charges author fees ($500 for accepted articles) for what is currently "50+ websites, 60+ online research journals and 20+ review journals", with institutional memberships available to cover the author fees of faculty. The institutional members, now numbering more than 100, pay between $1500 to something beyond $7500 annually, depending on the size of the organization. Although BioMed Central is not yet a profitable venture, its author fees and institutional memberships, as well as Web site advertisements, have shifted the cost of running the journal to the individuals and institutions that use it to publish, or promote, their work.[18] While advertising revenue may not be viable for many disciplines, author fees might be thought of as reconstituting membership dues for a scholarly society (while allowing exemptions for authors from, for example, developing nations). Given the scale of BioMed Central publishing activities, this model also suggests the value of scholarly associations forming consortia, in conjunction perhaps with related independent journals, to manage and publish a good number of journals within a given field.[19]

This idea of developing institutional memberships and publishing consortia also suggests a second model, based on capturing a portion of the existing funds already set aside for these journals by the research libraries. After all, the libraries have the most to gain financially from journals that go open access. Yet rather than a windfall of free journals, they are looking only to contain their costs, as University of British Columbia head librarian Katherine Quinlan once explained to me. The Association of Research Libraries has been promoting a number of open access initiatives, including the "institutional repository", which follows in the footsteps of arXiv.org and the eprint archives.[20] Yet the institutional repository creates a parallel (open access) universe in relation to the world of fee-based journals, as we have seen with Rusin and the Astrophysical Journal. The repository relies on voluntary submissions and is unlikely to offer a complete record of the literature, just as it sustains the redundancy of the system. It compels libraries to continue subscribing to journals, which remain the source of peer review and a complete record of the literature. But is such a division of function necessary, between universities maintaining open access repositories and publishers maintaining peer-reviewed archives?

A cooperative venture formed among research libraries and scholarly associations could well provide the reviewing, publishing, indexing and archival processes that are vital to the health of the research enterprise. Such a cooperative would not be easily achieved, but it can be readily imagined building on what is already in place. It might be based, for example, on the famous 20-80 rule used by Institute for Scientific Information (ISI) in justifying the limited number of journals it includes in the Web of Science Citation Index (Garfield 2002, p. 13). ISI contends that 20 percent of the journals produce 80 percent of the citations, and that rule might well apply to universities, and be used as a basis for supporting the scholarly association publishing activities. That is, roughly 20 percent of the universities produce 80 percent of the most heavily cited articles and are the major beneficiaries of the academic knowledge economy, whether one looks at grants, citations, salaries, or other measures. Further, imagine that 400-500 research libraries worldwide (the top 140 of which in North American are already gathered under the umbrella of the Association of Research Libraries) form an alliance to support the publishing programs of scholarly associations at a rate based on perhaps 80 percent of the current subscription fees paid by those libraries to the associations. In return, the scholarly associations would publish their journals on an open access basis. The top research libraries would achieve immediate and long-term savings, while thousands of other institutions would have access to these journals for the first time. The incremental costs of this greater access are minimal, while the extra financial responsibility carried by the leading institutions needs to be figured in their longstanding claims to be making a greater contribution to the public good by virtue of their research productivity.

6 Conclusion

Open access publishing of e-journals is within reach for scholarly associations, whether through the savings realized by dropping the print editions, direct forms of support provided by research libraries, charging author fees and institutional memberships, or a combination of all three. While open access is an economically viable alternative to charging users and readers, the scholarly associations still have substantial issues to work out, in dropping print versions, curbing the loss of membership, and settling contractual obligations to corporate and university presses. If the scholarly associations are going to be able to make informed, if not enlightened, decisions, then these economic aspects need to be explored on an association-by-association basis, as do questions of intellectual quality, faculty members' careers, and responsibilities for contributing to a global public good. The associations need to understand the economic redundancies and general undermining of the commercial model already well underway. Compromises between the commercial and open access models may seem to resolve some of the redundancies. It is, after all, still open access if the journal archives are opened six months or a year after initial publication, or if open access is granted to developing countries.

Yet serious thought has to be given to the nature of the privilege of exclusive access afforded by subscription, which is being protected by this process. Scholarly associations need to consider that the general decline in faculty member subscriptions will accelerate with the greater access afforded by the digital editions provided by their research library, especially as those editions are integrated in larger systems of the related literature. By trying to protect the subscription privilege, if only with a six-month to a year lag time between subscription copy and open access, the associations are, in effect, doing a disservice to their authors and editors and reviewers by further postponing widest possible publication, and giving their members a dubious privilege that has already had its exclusivity seriously compromised.

During this transition period of both print and electronic publication - which makes open access far more difficult to sustain - the scholarly associations need to rethink their role and services, rather than holding on as long as possible to a publishing model which may well be passing. This is the time to bring the scholarly and economic, the ethical and intellectual, aspects before the membership, as the change in publishing mediums could well alter the nature of the scholarly association. Just as the individual subscription is a declining incentive, so many associations may need to think beyond the membership model and see themselves sustained by the services they provide through conference fees, author fees, and related benefits. BioMed Central has 200,000 registered users, representing a new model of membership and association forming around the sharing of knowledge. Another way forward, I am suggesting, is for the associations to work in greater partnership together, to build a critical mass for electronic publishing, as well as for garnering support from the institutions which their disciplinary leadership and marshalling of intellectual resources - whether through journals, conferences, or other means - so directly supports, namely, the universities and their research libraries. There is a need for disciplinary leadership around issues of access and ownership, not only in publishing, but in the sharing of data sets and related research databases to strengthen the quality of research and encourage the scientifically productive notion of an information commons against increasing efforts to privatize data (Reichman and Uhlir 2001).

In his economic inquiries into scholarly publishing Machlup (1977, p. 217) noted that "in a wide sense of the phrase, any activity is 'economically viable' if its product is promoted to the ranks of public good and its cost is borne largely out of public funds, such as an actual or potential tax revenue". Scholarly inquiry is economically viable, in the first instance, as a public good - with its cost largely borne by public funds - and the scholarly publication of that research should be no less viable for the same reason, as colleagues edit, review, join in non-profit societies, to further the very work of that inquiry, with public support. With so much scholarly activity supported by public money, it is only natural to ask whether there is not now a way to distribute the resulting research in ways that make it open and available, as a global public good. As the American Association for the Advancement of Science would remind us, "users" of scientific research include "historians and philosophers, editors, consultants, students and educators, journalists, consumer advocacy groups, government regulators and policy makers, and members of the legal community, as well as that diverse group we refer to as 'the general reader'" (Frankel 2002, p. 8).

With print there was reason to make readers and libraries pay for elaborately published volumes, prepared in specialized print shops, well-bound on good paper, meant to stand as a permanent scholarly record of scientific and intellectual achievement. The Internet changes what it means to go public. Movements are afoot to extend the public domain within cyberspace, with the growth of e-government and open source software, even as groups such as Creative Commons and Public Knowledge feel it is imperative to place a check on the rampant privatization of this knowledge economy. New needs have been recognized for global access by a wider community. Now is the time for scholarly associations to ask themselves how best to use this new publishing medium, already integral to the scholarly process at every stage, to extend and advance the circulation and exchange of knowledge by a considerable step. The scholarly associations need to add to their publishing agendas, their annual conferences and other forums, the question of whether open access publishing is worth exploring, given that research and scholarship which their members work so hard to produce may now be made unequivocally part of a larger and revitalized public sphere.

Acknowledgements

The author wishes to acknowledge the assistance of Kevin Jamieson, Faith Shields and Anne White in the preparation of this article, as well as the support of the Social Sciences and Humanities Research Council of Canada.

References

Arms, W. Y. (2002) "Quality Control in Scholarly Publishing on the Web". Journal of Electronic Publishing, 8(1) http://www.press.umich.edu/jep/08-01/arms.html

Association of Learned and Scholarly Publishing (ALPSP) (2002) "Authors and Electronic Publishing - ALPSP Study". Scholarly Communication Report, 6(4), 6

Bergstrom, T. C. (2001) "Free Labor for Costly Journals". Journal of Economic Perspectives, 15(4), 183-98 http://www.econ.ucsb.edu/~tedb/Journals/jeprevised.pdf

Cameron, R. D. and Tatu. S. G. (2000) Bibliographic Protocol Level 1: Link Resolution and Metapage Retrieval Internet Draft
http://www.cs.sfu.ca/~cameron/bibp-revised.html

Crow, R. (2002) The Case for Institutional Repositories: A SPARC Position Paper, SPARC, July
http://www.arl.org/sparc/IR/ir.html

Crow, R. and Goldstein, H. (2003) Guide to Business Planning for Converting a Subscription-based Journal to Open Access, Open Society Institute, Edition 1.0, January
http://www.soros.org/openaccess/oajguides/business_converting.pdf

Doyle, M. (2001) "Peer-review's future in a world of open archives". Workshop on the Open Archives Initiative and Peer Review Journals in Europe, Geneva, March
http://agenda.cern.ch/askArchive.php?base=agenda&categ=a01193&id=a01193s4t8/transparencies (Powerpoint presentation)

eJUSt e-Journal User Study (2002) "Research Findings: Second Survey Highlights". Stanford University Libraries and Highwire Press, May
http://ejust.stanford.edu/findings2/2SurveyHighlights.html

Electronic Publishing Steering Committee, Cornell University (1998) Electronic Publishing Steering Committee Report on Electronic Publishing Strategies for Cornell University, October
http://www.library.cornell.edu/ulib/pubs/EPSCFinalReport1998.htm

Fisher, J. H. (1999) "Comparing Electronic Journals to Print Journals: Are there Savings?". In Technology and Scholarly Communication: The Institutional Context, edited by Richard Ekman and Richard E. Quandt (Berkeley: University of California Press), pp. 95-101
Also available in Scholarly Communication and Technology Conference Organized by The Andrew W. Mellon Foundation, Emory University, USA, April, 1997
http://www.arl.org/scomm/scat/fisher.html

Frankel, M. S. (Chair) (2002) Seizing the Moment: Scientists' Authorship Rights in the Digital Age, American Association for the Advancement of Science, Washington, DC
http://www.aaas.org/spp/sfrl/projects/epub/

Garfield, E. (2002) "Reflections of a Scientometrician/Communicator - From Current Contents to Citationology and Algorithmic Historiography". China Medical Board Symposium, Beijing, September http://www.garfield.library.upenn.edu/papers/reflectionsofascientometrician(china)092402.pdf

Ginsparg, P. (2001) "Creating a Global Knowledge Network". Second Joint ICSU Press - UNESCO Expert Conference on Electronic Publishing in Science, Paris, February
http://arxiv.org/blurb/pg01unesco.html

Hitchcock, S., Bergmark, D., Brody,T., Gutteridge, C., Carr, L., Hall, W., Lagoze, C. and Harnad, S. (2002) "Open Citation Linking: The Way Forward". D-Lib Magazine, 8 (10), October http://www.dlib.org/dlib/october02/hitchcock/10hitchcock.html

King, D. W. and Tenopir, C. (1998) "Economic Cost Models of Scientific Scholarly Journals". Proceedings of ICSU Press Workshop on Economics, real costs and benefits of electronic publishing in science - a technical study, Oxford, March-April
http://www.bodley.ox.ac.uk/icsu/kingppr.htm

Kling, R., Spector, L. and McKim, G. (2002) "Locally controlled scholarly publishing via the internet: The Guild Model". Journal of Electronic Publishing, 8(1), August
http://www.press.umich.edu/jep/08-01/kling.html

Kyrillidou, M. and Young. M. (2002) "Research Library Trends". Association of Research Libraries, August
http://www.arl.org/stats/arlstat/01pub/intro.html
Also published in ARL Statistics 2000-01 (Washington, DC: Association of Research Libraries)

Lawrence, S. (2001) "Online or invisible?". Nature, 411 (6837), 521 http://www.neci.nec.com/~lawrence/papers/online-nature01/

Lawrence, S., Giles, C. L. and Bollacker, K. (1999) "Digital Libraries and Autonomous Citation Indexing". IEEE Computer, 32(6), 67-71
http://www.neci.nec.com/~lawrence/papers/aci-computer98/aci-computer99.html

Machlup, F. (1977) "Publishing Scholarly Books and Journals: Is It Economically Viable". Journal of Political Economy, 85(1), 217-225

Milstein, S. (2002) "Scholarly Reviews Through the Web". New York Times, August 12, p. B3.
http://www.nytimes.com/2002/08/12/technology/12NECO.html (free registration required)

Muthayan, S. and Muinde, F. (2003) "Public Knowledge Project Open Journal Systems". International Symposium on Open Access and the Public Domain in Digital Data and Information for Science, Paris

Patel, J. and Kumar, K. (2001) Libraries and librarianship in India (Westport, CN: Greenwood Press)

Pinfield, S. (2003) "Open Archives and UK Institutions: An Overview". D-Lib Magazine, 9(3), March http://www.dlib.org/dlib/march03/pinfield/03pinfield.html

Publisher Developments (2002) "E-journal pricing survey". Scholarly Communications Report, 6(5), 13

Publishers' Products (2002) "Integration of IDEAL with ScienceDirect". Scholarly Communications Report, 6(6), 3

Putman, R. D. (2000) Bowling Alone: The Collapse and Revival of American Community (New York: Simon & Schuster)

Reichman J. H. and Uhlir, P. F. (2001) "Promoting Public Good Uses of Scientific Data: A Contractually Reconstructed Commons for Science and Innovation". Conference on the Public Domain, Duke University, November, pp. 241-322
http://www.law.duke.edu/pd/papers/ReichmanandUhlir.pdf

Rosenberg, D. (1997) University Libraries in Africa (3 Volumes) (London, UK: International African Institute)

Tenopir, C. and King, T. (2001) "Lessons for the Future of Journals". Nature Web Debates: e-access, October 18 http://www.nature.com/nature/debates/e-access/Articles/tenopir.html

Willinsky, J. (2003) "Policymakers' Use of Online Academic Research". Education Policy Analysis Archives, 11(2), January
http://epaa.asu.edu/epaa/v11n2/

Willinsky, J. (2002a) "Education and Democracy: The Missing Link May Be Ours". Harvard Educational Review, 72(3), 367-392

Willinsky, J. (2002b) "Copyright Contradictions of Scholarly Publishing". First Monday, 7(11), November http://www.firstmonday.dk/issues/issue7_11/willinsky/index.html

Willinsky, J. and Wolfson, L. (2001) "The Indexing of Scholarly Journals: A Tipping Point for Publishing Reform?" Journal of Electronic Publishing, 7(2), December
http://www.press.umich.edu/jep/07-02/willinsky.html


Notes

[1]

For a list of open access journals, see Directory of Open Access Journals (http://www.doaj.org/).

[2] On the reduced access to serials, the Association of Research Libraries, representing the 124 top libraries in North America, reports a 5 % drop (1986-2001) in serials purchases among its members (Kyrillidou and Young 2002). Far more drastic drops in access to serials in the developing world are documented for Africa (Rosenberg 1997) and for India (Patel and Kumar 2001).

[3] On scholarly associations and open access publishing: the American Education Research Association (http://www.aera.net) offers free access to Educational Researcher alone among its six journals; the National Council of Teachers of English (http://www.ncte.org) provides access to all of its other journals, one year after publication; and the Institute of Physics (http://iop.org) provides the first 30 days after publication of free access to its 36 journals.

[4] On scholarly associations and commercialization: 10 of Elsevier's 13 new journals in 2001 were drawn from scholarly associations, and 10 of the 35 new titles offered by Sage Publications for 2002 represented "society contracts" (according to their respective Web sites). On publishers' digital expenditures, Elsevier declared in 2002 that it invested $30 million in its online system (Publishers' Products 2002). In terms of the consequences of moving a journal from a non-profit to a commercial publisher, as well as the increasing hold of commercial publishing, Bergstrom (2001, p. 2) found that among the top economics journals, non-profit publishers such as scholarly associations charge an average of $180 for library subscriptions, versus an average $1660 for commercial publishers. He also found that in 1960, the 300 economic journals of the time were almost entirely non-profit, while by 1980, half of the then 120 journals were published by commercial concerns, and by 2000, that proportion had risen to two-thirds of the 300 journals (p. 9-10).

[5] The Astronomical Journal annual subscription fee is $440 for both print and electronic versions, with the electronic version alone $350, while the thrice-monthly Astrophysical Journal has an annual subscription fee of $1525 for paper and electronic versions, with an additional $250 for its Supplement Series.

[6] One serious limitation of print journals becomes apparent when you consider that for research libraries in Africa, for example, the airfreight charge for American Astronomical Society journals is more than double the cost of a subscription. The Society does have a "journal donation" program, in which members donate their back issues, but that again requires the libraries to pay shipping costs.

[7] ResearchIndex has the distinct advantage over Web of Science of being (a) free, largely as a result of its experimental status within the NEC Research Institute and its "autonomous" ability to find and parse scientific articles on the Web (b) improving on citation indexing by providing 100 words of context for each citation of a given paper such as Rusin's, as well as a similar context for each citation, and (c) providing direct access to papers (Lawrence et al. 1999). ResearchIndex uses an autonomous search of the Web for locating scientific papers and abstracts (posted in PDF, Postscript and related formats), which naturally skews its sampling of the literature. Citebase offers extra advantages as well as (a) and (c), and works with papers found in Open Archives Initiative research databases, such as arXiv.org, which similarly skews its sampling of the literature. In each case, automated processing of papers is by no means as reliable as the handcrafted work of the Web of Science. Web of Science prides itself on correcting the large number of errors found in reference lists, but exclusions are based on its highly selective processing of 8600 journals, taking only 10-12% of the 2000 or so journals that apply each year for indexing, according to its Web site ( http://www.isinet.com/isi/hot/essays/selectionofmaterialforcoverage/199701.html).

[8] Although associations may be tempted to deploy devices that ensure the library's electronic copy does not replace the member's copy - perhaps by limiting the number of users of an electronic edition at any one time, as some publishers do, or by providing access to additional content - my hope is that the associations will see the advantages of providing open access to their journals, not just for those scholars with access to substantial research libraries but to readers on a global basis.

[9] It is also worth noting, Machlup (1977, p. 224) found that the number of individual and institutional subscriptions declined between 1966 to 1977 in the handful of science and technology journals he was examining, only to be compensated by fee increases that caused revenues to increase between 76 and 167 percent. This non-subscription approach to search and select reveals cracks in the journal system's reliance on bundling of loosely related articles in a format dictated by the economics of periodical printing, sales and distribution.

[10] The best source of financial data for American non-profit organizations, such as scholarly associations, is GuideStar which is run by Philanthropic Research Inc, whose mission is "to revolutionize philanthropy and nonprofit practice with information". GuideStar provides a scanned version of the Internal Revenue Form 900 (Return of Organization Exempt from Income Tax), a form that declares itself "open to public inspection", that must be filed by non-profit organizations with a revenue of over $25,000. I did not include societies which have a large professional or non-academic component to their membership and mandate, such as the American Chemical Society, which has a revenue that exceeds $300 million.

[11] The American Astronomical Society, for example, clears $221,156 in subscription revenue above its publications costs, which is the largest "profit" among the sample of associations considered here (explained in part, perhaps, by membership fees not including free subscriptions). The Radiation Research Society, however, does better with its 43 percent "profit" on a $347,595 investment in publishing its monthly Radiation Research (annual subscription $615), while the History of Science Society has a 42 percent return on its two quarterlies, Isis (annual subscription $201) and Osiris (annual subscription $50.50) published through the University of Chicago Press.

[12] On the range of estimated costs for e-journal publishing: (1) Glass, editor of the online Education Review and the Education Policy Analysis Archives, in an email in February 2001 estimated his publishing costs as "zero, nada, no budget, no grad assistant, no secretary"; (2) Ginsparg (2001) estimated that arXiv.org costs for authors self-archiving their work are roughly $1-5 a paper, (3) the Chicago Journal of Theoretical Computer Science budgets $2070 per article (Fisher 1999), (4) King and Tenopir (1998) put the cost of electronic publishing at $368 per page or about $175,000 per year for an e-journal; and (5) the Electronic Publishing Committee at Cornell University (1998) estimated that it would cost $2,700,000 to establish an electronic publishing program.

[13] Annual postage costs of the Journal of the American College of Cardiology, for example, have been estimated at $60,000-$70,000 (Milstein 2002), while a publisher's rough estimate of administration costs of a single paper have been put at $500-$1000 (Doyle 2001, cited by Arms 2002).

[14] Commercial versions of this software, at a cost of typically $5,000 to $20,000, with processing fees, generally $12 to $50 a manuscript, are being used by 30 percent of journal publishers by one estimate (Milstein 2002).

[15] In their business plan for converting journals to open access, Crow and Goldstein (2003, p. 8), for example, identify the "incremental costs of converting" as "site design and technical development, including implementation of a user interface, file or database structure, access authentication system, back-up systems, etc.; licensing and implementing an editorial 'pre-press' workflow system; content formatting and metadata tagging; and web site hosting and storage". Each of these is handled by Open Journal Systems, except the hosting and storage, which we are encouraging research libraries and scholarly associations to absorb within their existing infrastructure.

[16] Public Knowledge Project, Open Journal Systems (http://pkp.ubc.ca/ojs).

[17] Peter Suber's Timeline of the Free Online Scholarship Movement http://www.earlham.edu/~peters/fos/timeline.htm

[18] In terms of revenue, a typical BioMed Central journal publishes roughly 40-50 articles a year, generating $20,000-$25,000 in author fees, which is similar to the open access New Journal of Physics which also charges $500 per article and has 17 societies supporting it. Advertising rates at BioMed Central run from $5 per "clickthrough" (every time someone clicks on an advert) to "supersize ads" that cost $100 per thousand impressions, which could amount to more than million dollars a year in revenue, given the site's more than a million viewings a month.

[19] The importance of enabling independent journals, by keeping entry barriers low for new publications of scholarly quality, has been identified by the American Association for the Advancement of Science report on scholarly publishing (Frankel 2002, p. 7).

[20] See MIT's DSpace digital repository (http://www.dspace.org), Eprints.org (http://software.eprints.org/) and institutional repositories (http://www.arl.org/scomm/open_access/index.html). The institutional repository is intended to house "digital collections capturing and preserving the intellectual output of a single or multi-university community" that "will form part of a global system of distributed, interoperable repositories that provides the foundation for a new disaggregated model of scholarly publishing" (Crow 2002, pp. 4, 6). In supporting institutional repositories, the Association of Research Libraries recognizes that "library programs and budgets will have to shift to support faculty open access publishing activities in order for the library to remain relevant to this significant constituency", and that this is "a natural extension of academic institutions' responsibility as generators of primary research seeking to preserve and leverage their constituents' intellectual assets" (Crow 2002, p. 20). As to the cost of institutional repositories, the current answer is consistent with estimates in this entire area: "Practically speaking, both development and operating costs can range from virtually no incremental costs (for institutions that reallocate resources) to hundreds of thousands of dollars (for institutions recognizing incremental systems and staff resources)" (Crow 2002, p. 28). As the social sciences and humanities, as well as a number of sciences, have not had a preprint culture, the regular use of eprint repositories can hardly be expected to come as easily to them.