Donald W. King
(This Opinion piece presents the opinions of the author. It does not necessarily reflect the views of D-Lib Magazine, its publisher, the Corporation for National Research Initiatives, or its sponsor.)
Open access through author payment has clearly caught the attention of a wide audience. While I am not specifically advocating this initiative, it appears to have real merit and warrants careful examination and testinga view I have not always held. However, I also believe it is counter-productive for author payment advocates to denigrate commercial publishers (and profit) and exclude them from the open access model. To do so diminishes the chance of success for the model in the long run. Rather than creating a divisive atmosphere among journal system participants, a more successful approach would be for all of them to join together to test the model. In the past commercial organizations such as online bibliographic vendors have made a tremendous contribution to science in partnership with non-profit and government database producers . The commercial sector has made too many contributions to science to dismiss the sector as being irrelevant and basically self-serving. This article is to argue for cooperation and to consider experiments to determine the strengths and weaknesses of various approaches to open access by author payment.
Some have likened the need for open access to the journal literature to extensive open access to the mapping of the human genome. Dr. Mark Walport, Director of the Wellcome Trust (a major Human Genome Project contributor) is quoted as follows regarding journal open access: "The model of the Human Genome Project need not be unique, and it is the principle of free access that we want to champion" . There is another side to the Human Genome phenomenon that is also akin to the current journal environment.
After the discovery of unique DNA patterns in 1984 and DNA sequencing in 1986, scientists gathered together to examine the possibility of sequencing the human genome and speculated that it would take 30 years and $3 billion to achieve this goal . Even though some scientists objected because it might divert funds from useful research, the U.S. Congress began supporting the Human Genome Project (HGP) in 1990, and pieces of the mapping began in 16 international laboratories with a 15-year timeline to finish. J. Craig Venter, a scientist at the National Institutes of Health suggested that advanced technologies be used to speed the process, but his idea was spurned. He then raised private sector capital to develop super computer programs and 300 special processing machines that would considerably reduce the time and costs. He intended to sell the raw DNA product to pharmaceutical companies and universities. This approach outraged the public HGP scientists who believed that this information was a public good and should be freely available to anyone (using the Internet). It was generally recognized that the Public HGP needed the machines as well and, after much disagreement and vilification of Venter despite his considerable contribution, the machines were provided to the Public HGP at a cost of $300,000 per machine.
This led to an "arms race" between the Public HGP and Venter's company. While both sides wanted to achieve a common and worthwhile goal, both sides had potential motives that were not necessarily praiseworthy. President Clinton, seeing the counter-productive nature of the competition, got involved and brought in an arbitrator. Since neither side wanted to be seen in public together, the two leaders met in secret in the arbitrator's basement and worked out a compromise. A draft of a plan was co-announced at the White House in 2000, and later that year final results were cooperatively revealed. Some 4,000 persons contributed to the project with positive results that demonstrated that public and private sectors can cooperate to achieve a common good. While the compromise was by no means perfect for either side of the controversy, science and humanity will surely benefit . An analogy to this story is that the hostility toward Venter and his company is similar to that directed by some toward commercial publishers. Even though arguments still linger concerning the genome technology, perhaps something can be learned by all journal system participants from this experience, which is that both the public and private sectors can contribute to the public good.
Author payment is not new. In the 1970s, a statistical sample of articles and a survey of their principal author showed that over half of all science articles written by U.S. authors required some form of author payment and in some fields nearly all articles involved author payment . Only four percent of articles involved submission fees (as under the OA model). The most common requirement was page charges, which took into account the range of size of manuscripts submitted (38% of all articles published) and further revision fees (3% of articles). Some publishers asked authors to pay for copies of reprints or preprints that could be distributed by them (20% of articles). Altogether, the average cost when an author payment was made was about $750 in current dollars. Most of the payment in the U.S. was covered by the Federal government by grant to the authors' organizations or as the authors' employer (56% of payment), 26 percent of payment was by other funders, and 18 percent by authors themselves (i.e., out-of-pocket).
The policy of author payment from Federal grant or contract was not without some dispute, although an attempt was made to set an author payment policy. A Federal Council for Science & Technology (FCST) policy stated that Federal payment of page charges is recommended under the following conditions: (1) research papers report work supported by the government, (2) charges are levied impartially on all research papers published by the journal, (3) payment of such charges is in no sense a condition for acceptance of manuscripts by the journal, and (4) the journals involved are not operated for profit. In hindsight, the latter condition may have contributed to the decline of author payment. In order to avoid payment, an increasing number of authors began submitting their manuscripts to commercial and other non-payment journals. While page charges were not the most important reason given in studies at that time for choosing a journal, page charges were a factor nevertheless. The competition for manuscripts resulted in many society and other non-profit journals losing revenue from declining page charges and thereby having to increase their subscription prices to libraries. I don't see how these conditions are changed by the re-emergence of author payment.
If author payment is to be examined realistically, valid metrics must be used to assess its merits. While observation and comparison of price trends and impact measures among types of publishers are useful, there are several other important indicators that one should keep in mind. The rekindling of author payment has highlighted the importance of three other indicators: the publishers' cost of processing articles, circulation, and the processing cost per subscription. The article processing cost has always been an important indicator, but largely ignored in the literature because publishers have generally been unwilling to reveal their costs. In this commentary, by author processing costs I include not only direct costs but also allocation of support costs such as overhead, system infrastructure, financing, and so on that must also be recovered through author payment revenue in order to break even. These costs, of course, together with overhead and profit are partially used to determine subscription prices and license fees, but article processing costs are not the same as prices. There are two publishing services that authors and readers (or their libraries) buy: (1) the value added to article information content and (2) means of distribution or access to the information. Comparison of publishers or journals should be made on the basis of article processing costs and the article attributes provided and not solely on subscription price or size derivations of it.
Circulation is the dominant variable that determines price up to a critical mass of about 5,000 library subscribers and the circulation of most science journals is below that amount. The price necessary for journals to break even can range from $100 to well over $1,000 per subscription depending on their journal circulation and size . For example, ignoring non-article processing costs (e.g., tables of contents, letters, etc.) and assuming a typical size journal of 150 articles per year, an article processing cost of $3,000 per article (not atypical for traditional science publishers), a $40 per subscription cost of reproduction and mailing, and circulation of 500, 1,000 and 5,000 the subscription price must be at least $940, $490 and $130 respectively merely to break even (subtract $40 from each if distributed electronically). The article processing cost and number of articles processed also affects price because the total amount of article processing and marginal cost of reproduction and mailing would increase if the number of articles went up. For example, if the number of articles published is 200, the respective prices to break even would be about $1,250, $650 and $170 (actually slightly less due to fixed costs associated with various sub-processes). While some citation impact indicators take size into account, circulation is rarely considered because reliable circulation data are not widely reported, particularly by commercial publishers.
Clearly, increased contribution from authors would reduce the high prices of many journals due to their low circulation and/or large size. The question, then, for author payment advocates is what realistic article processing costs are for journals providing different levels of attributes (i.e., quality, speed of publishing, special features) and what authors (or their funders) are willing to pay for this service and it's attributes. Open access publishers are currently charging between $500 to about $1,750 per article. Traditional publishers indicate the cost is from $2,000 to $4,000 with all costs included (i.e., no hidden subsidies). I believe that it may typically be in the $3,000 to $4,000 range depending on rejection rates, refereeing, editing, special graphics, formatting, special features and so on. Nevertheless, if author payment does become widely accepted, publishers will have to compete on a basis of what authors are willing to pay for such journal attributes as reputation, circulation (number and composition of readers), value-added features and speed of publishing.
Can commercial publishers compete in an author payment environment? I don't think they want to, but evidence suggests that they could. To begin with, their article processing costs appear to be competitive, although on the high side of my range quoted above ($3,000 to $4,000). Even though their costs appear to be competitive, two reasons for their high prices in the past were (a) that their average circulation was lower and (b) their size (number of articles) tended to be larger than other publishers' journals. Thus, the high prices are not just due to profit. The attributes of their journals also are competitive. Therefore, there is no reason for commercial publishers not to be included in an author payment journal system. On the contrary, there are reasons they should not be excluded, but rather encouraged to participate. The investment necessary to replace an existing commercial journal tends to be about $100,000 for start-up, capital requirements, future research and development, and operations. Thus, total investment to replace all commercial journals would be on the order of hundreds of millions of dollars, with many replacement start-up journals not surviving due to their experimental nature and the inexperience of those publishing them. I think it a mistake to dismiss commercial publishers. If they cannot compete in this environment, so be it. I suspect they will find a way to thrive, which should be much to the benefit of the open access concept.
The question then is: "Who should pay for article processing?" It doesn't make sense for authors themselves to pay, even though some authors have paid out of their own pockets in the past and, as readers, some currently help support journals through personal subscriptions. Authors from developing countries have objected to author payment because their research is not funded. Yet library subscriptions and licenses have been subsidized by some publishers and some journals in publishers' portfolios and articles within journals have always been subsidized by others. Thus, there is no reason some author payment will not be subsidized by other authors.
Some have suggested that universities should pay since they already expend so much in purchases of library collections and examples are given which compare potential author payment with library collection purchases, but with conflicting results. Even though some universities may pay more in collection purchases than they would in author payment, the approach ignores the fact that, while most articles are authored by university scientists, most reading is done by scientists located elsewhere and, as such, their libraries currently contribute to publisher revenue. Thus, universities would, in effect, partially subsidize other organization journal purchase and use by their staff.
It seems to me that research funders have a responsibility to subsidize article processing. A distinction is made here between the source of R&D funds and performers who conduct R&D from these funds. Of $265 billion in R&D funds in the U.S. (2000), only $30 billion is performed in universities and colleges even though about three-fourths of articles are authored (or co-authored) by their scientists. Over half of the university $30 billion is funded by the Federal government, about one-fourth of these funds are internally funded, and the rest from industry, foundations, etc. About another $10 billion is performed by other organizations which lead to articles.
There is a long history of the U.S. Federal government recognizing the importance of open access to research and technology results through the technical report literature. Most agencies made copies of these reports openly available without cost to organizations funded by them, and provided access to all others through the National Technical Information Service at a cost to the recipient of reproduction and mailing. This policy recognizes that the value of research results is highly dependent on the extent to which the information is used. Surveys of scientists in non-university environments over the past 25 years have clearly demonstrated the enormous and growing use, usefulness and value of science articles. In effect, results of about $40 billion in R&D funding for research performed in universities (and elsewhere) have produced an essential resource used in $225 billion of R&D effort performed elsewhere.
Some have argued in the past that university authors write largely to achieve tenure and advancement, as though this is inherently bad. While this is a motivation for some, I think advancement of science is a more powerful motivation. Furthermore, the discipline of writing about research in many instances is an integral part of the creative research process. Regardless, the body of journal literature, even with some poor articles, has more than proven itself by its use among authors and the many more scientists who do not write articles.
If open access by author payment is in fact an optimum model, there must be some way to demonstrate it other than by the trickle approach now taking place. To be realistic, all system participants must be involved in a comprehensive manner. One way that I think this can be done is initially through experimentation, or other approaches requiring primary funding agencies, publishers, libraries, and other participants to cooperate in a meaningful way. This may seem unrealistic, but just as with the genome experience, I think that all current participants have the goals of science as their ultimate mission. It will take an influential force to make it happen.
 Bourne, Charles P. and Trudi Bellardo Hahn. A History of Online Information Services: 1963 - 1976. 2003. Cambridge, Massachusetts. The MIT Press.
 "DNA: The Secret of Life." (2003) [DVD]: Windfall Films.
 King, Donald W., Dennis D. McDonald, and Nancy K. Roderer. Scientific Journals in the United States: Their Production, Use, and Economics. 1981. Stroudsburg, PA: Hutchinson Ross Publishing Company (a division of Academic Press).
 Tenopir, Carol and Donald W. King. Towards Electronic Journals: Realities for Scientists, Librarians and Publishers. 2000. Washington, D.C. SLA Publishing.
Copyright © 2004 Donald W. King