Electronic Scientific, Technical, and Medical Journal Publishing and Its Implications: Proceedings of a Symposium (2004)

Jeffrey MacKie-Mason, University of Michigan

Scholarly publishing in any medium requires substantial resources beyond the content creation costs. In the early 1990s, when digital publishing started to become a serious possibility, there was some confusion about this in some corners of the scientific community. There was a sense that publishers did not really do anything. Everything was done by authors, and publishers were valueless middle people who distributed the material to readers. As soon as the Web emerged in 1994, it began to be thought that the publishers could disappear.

That, of course, simply is not true. Publishers are providers of value-added products and services. They perform some very crucial functions by producing documents in different media, doing the editorial and design work, marketing the material, and getting readers connected to writers, and so forth. All of those functions involve costs. Even a not-for-profit publisher has to recover its costs and have some sort of reserve. The for-profit firms need to get a return on their investment.

This session, therefore, is devoted to a discussion of sources and types of revenue, ways of raising revenue, and different business models, particularly in a world where digital publishing is becoming much more the norm. The defining question is: How does a publisher organize delivery and rights management across modes of access so as to recover production costs and induce ongoing investment in development? As discussed in the first panel, new technologies are emerging all the time; new forms of delivery, new forms of access, and new types of services are being provided. That requires an ongoing stream of investment for things that once were thought of as one-time costs.

What are the sources of value? Revenue is simply a transfer of value between parties. To address a business model, particularly to start thinking outside the box and think about changing frameworks, we need to think about who values the publications or the information delivery in the first place. Who might be willing to share some of that value back to the publishers and the distributors?—the teachers, public and private researchers, students, practitioners, general public, and so forth.

Another question is, Where does the publisher collect that value? What is the best way to organize the extraction of value from some of the stakeholders and transfer it to some of the providers and distributors? Is it through the universities?

We heard that the traditional publishers are focusing increasingly on library subscription models as their primary mechanism of collecting revenues. In some cases it is from the practitioners, for instance the medical practitioners who subscribe to the New England Journal of Medicine. There has also been much discussion about the delivery of scientific materials to developing countries, and a question about whether or not those should be delivered free as part of an expansion of access to knowledge around the world or whether developing countries should be contributing as well in some fashion.

Who are the providers and who are the potential toll collectors? There are a number of different services that go into the publishing process, and the publisher does not necessarily provide all of them. Are those services going to be provided as a bundle by a single service provider, or will there be multiple providers providing overlapping services who will need to recover their costs, and provide different aspects of access to information content?

The business models depend on how the information is accessed and what the information is. Again, in a digital world, we no longer need a single standard mode, a journal. We can think about breaking up the information in many different ways, and repackaging and distributing it in different combinations. This often is referred to as the unbundling and rebundling phenomenon. The corpus of information can be broken apart into its constituent pieces, particularly if it has been nicely organized, with a good markup language and tagging system, and then repackaged in other ways. So, we can sell it by

the view, by multiple views, or by time periods of viewing, among many other approaches.

There also can be charges for printing and file usage, as opposed to merely viewing usage. You could allow people to keep local electronic copies for rapid access and local archiving, or there could only be server copies. It is a question of whether you can search the full text or the image only. Bernard Rous mentioned that this is an issue that the ACM has faced as a membership organization. Which services should it provide to the general public, and which enhanced services should it provide to its members in order to induce them to remain members?

What is the role for government revenue? Much of what we are talking about—scientific, technical, and medical information and scholarly research—is information that benefits the general public either directly or indirectly, far beyond the community of scientists and scholars who are using it. There is a public interest in the dissemination of knowledge, in addition to its creation. Of course, governments in most countries are one of the main, if not the main, sources of funding to create the knowledge in the first place—to fund the research. With some exceptions, however, they have not been one of the main providers or direct disseminators of that information. Ultimately, at some level, governments are paying for it through subsidies to universities to pay for the libraries. But there is a question about whether the government should be directly intervening and providing access, or providing the revenues necessary to do the publishing.

There is value in the content itself, as well as in the value-added services that publishers, disseminators, aggregators, and distributors provide. There is thus a question about how different business models might succeed at supporting both aspects of the process, both content creation and distribution, getting incentives to generate the knowledge in the first place and getting reasonable incentives to provide high-quality publication, dissemination, indexing, and abstracting services. Having a particular business model that may address some of those needs, may not address others.

Finally, the overriding question for this entire symposium—much less this session—is: What impact is the digital publishing world going to have on science itself, on the scientific enterprise? For instance, how are the different business models, the different ways that access is provided and structured, going to affect the quality and productivity of science, collaboration at a distance, access for developing countries, the professional review and career process, peer review, and other aspects of scientific research?

Brian Crawford, John Wiley & Sons

John Wiley & Sons is a global, independent publisher, established in 1807. Wiley has three major areas of publishing today: scientific, technical, and medical (STM) journals, higher education materials, and professional and trade information. It is a fairly diversified publishing portfolio, with about $1 billion in annual revenues. The publisher has 400 STM journals online on its InterScience platform, which was established in 1997. It now contains about 2 million pages of information in the sciences, technology, and medicine.

Wiley provides open access to its tables of content and abstracts, as a general principle. That is so that its information can be found easily on the Web, either through secondary information services or directly by users who are browsing for such information and coming in directly to the service.

Each journal has its own home page. The reason for that business model is that Wiley publishes on behalf of many learned societies. Wiley wants even those journals that it owns to maintain their own distinct identity online. It also sees the advantage of a rich integration of links to and from related material. HighWire certainly led the way with this, with its toll-free linking capability that they demonstrated with professional societies. Other publishers quickly followed suit. The establishment of

the cross-referencing service was quite visionary. And, of course, all publishers in the biomedical area benefit tremendously by the linkage provided by PubMed. So, the first aspect of the business model is being found on the Web.

Wiley has about 250,000 articles across those 400 journals, more than 100 of which are in biomedicine and linked to PubMed. It also publishes laboratory protocols on the Web. It has about 12,000 contributions from major reference works that have been placed online. It also has been moving its books online, about 300 books per year, from a parallel path of print publication.

Wiley’s online platform has about 400,000 users who have chosen to register. Registration is optional, and many users do not choose to identify themselves. The service gets about 3.6 million article or chapter views per month, and more than 12 million user sessions per year. The second message about Wiley’s business model is that it wants very much to drive usage and to encourage pathways for that use.

The next important aspect of Wiley’s business model is that it is very much a customer- or reader-pays model of delivering the content. As a result, the publisher places a great emphasis on sales staff worldwide—staff that it did not use in the conventional print environment, where it was marketing via direct promotion to scientists. Libraries customarily took print subscriptions, which did not require sales representatives to meet with them to negotiate electronic licenses. It is an added cost consideration, but important in driving the kind of revenue model necessary to securing the financial success of electronic publishing.

As Wiley embarked upon the development of licenses for its electronic journals, it did so very much in consultation with its major institutional customers. This resulted in flexible sales options. It did not want to make a commitment to one preferred option early in its electronic publishing, so it provided a menu of print and online options.

Wiley now uses basic and enhanced access licenses. Basic means title-by-title access, with some concurrent user restrictions. This is an option that is often suitable for the smaller institution or department. Its enhanced access license does not require that an institution take all Wiley titles that are available electronically. The institution can choose, but it is establishing a license for a larger body of work, with no concurrent user restrictions and with additional benefits such as negotiated price caps.

The publisher determined that individual and society member access privileges had to be instituted in parallel with its institutional sales models, because it also caters to an individual audience, unlike some other publishers who spoke in the first panel session. Its individual member audience is actually holding steady, if not growing, particularly for some of the smaller societies it deals with.

Wiley has been licensing access to journals online since 1999, which has resulted in strong revenue growth. That came as a result of its sales effort worldwide, but also because of the receptivity of the customer base, primarily the institutional library base, to making that switch to electronic resources. This is not an e-only shift though—it is very much a parallel purchase situation. For the most part, institutions are choosing to retain their print subscriptions until there is a reliable, permanent digital archiving solution.

The other part of the business model is an emphasis on direct relationships with customers. The business model has shifted away from selling solely through subscription agents as intermediaries. Wiley now often deals directly with its customers. It views its customers as partners, rather than simply as targets. It uses librarians as a channel to end users. Customer training is another mechanism to promote usage, something that all publishers need to do more. Another important aspect is a focus on customer loyalty to sustain business growth. Wiley also provides password access for its society members and individuals, recognizing the roaming nature of its professional customers.

Finally, Wiley constantly emphasizes investment in new features and enhancements. Some of the most recent ones include content alerts to apprise its audience of what is being published; delivery of content to mobile-edition platforms; and publishing online in advance of print publication.

In order to move beyond a pure subscription model of access, Wiley has developed a feature called Article Select. This is the sale of digital tokens to its licensed customers, enabling them to distribute those tokens to other users within their institutions to access content from journals that they do not subscribe to. The use of those tokens then can be examined to determine what journals might be added to that institution's collection on a subscription basis. This capability has now been expanded into a purer pay-per-view offering.

Many not-for-profit organizations are still in the early stages of offering bundled journal licenses to their institutional customers, whereas commercial publishers have undertaken much more aggressive licensing in recent years. This has caused some debate about this so-called big deal.

Wiley’s enhanced access licenses now include more than 1,000 institutional or consortia customers. That is not those who sign the license; it is those institutions that are part of such consortia. It constitutes about 70 percent of its e-license business. Wiley’s basic access license customers indicated that there are many institutions that are not prepared to make the move to the big deal. More than 3,000 of its customers are still taking the title-by-title electronic access option, which represents 30 percent of its e-license business.

A quick review of the respective license terms is useful to help make the distinction between the business models. The basic access license tends to be to a single site, one geographically contiguous building or campus. There is activated e-access for each subscribed title, and it is for one concurrent user per print subscription if the institution holds a print subscription.

The enhanced access license tends more to be multisite, multinational, and suitable for large consortia, with unlimited concurrent users. There is access to all subscribed current content as well as back content, and privileges such as Article Select, roaming access, and monthly usage reports. Wiley also offers license terms for individuals, what it calls a personal access license, providing individual subscribers with electronic access on a subscription basis; society member access, with members being given access to their subscriptions on terms on which they agree with their societies; and complimentary access to the publisher’s chief editors and members of its editorial boards.

Wiley offers free access, free from the time of publication, to developing countries. For its biomedical journals, it does this through the Health InterNetwork Access to Research Initiative (HINARI) program of the World Health Organization (WHO), in order to get content into the hands of investigators in parts of the world where they truly cannot afford such information.

Wiley's participation in HINARI, like that of other publishers, is viewed primarily as philanthropy, reaching out to provide information where it is very much needed in order to provide research support and medical care in struggling parts of the world, and not as way of seeding a market. The parts of the world that have been served initially by WHO through HINARI are not seen as being markets in the near or even medium term.

Wiley has chosen to maintain the paid-access model described above, and rejected what may be called the author-paid model that is being advocated by proponents of more open access to the literature as an alternative. The reason it has not elected to go the latter route, and why most of the society partners that it publishes with are quite chary of that, is perhaps best articulated by Professor Hal Varian, an economist at the University of California at Berkeley. Professor Varian has pointed out that any economic system tends to favor the entity that pays. Given that the bedrock of scientific communication is the peer-reviewed literature for original work, Wiley’s view is that any system that charges the author or a sponsor of the author in order for that author to be published is, in terms of the economic system, going to favor the author's desire to become published. There is nothing wrong with that, but one has to accept the consequences. What has evolved in scientific communication for the most part is a system where the reader pays, or an agent for the reader is paying, because that naturally introduces a selective filter that reassures any entity wanting to scrutinize the validity and the value of the work that it has passed muster.

It does not mean that peer review cannot exist in and support a system where the author is paying. However, one can offer a provocative example of why this is incompatible with the sort of publishing that Wiley does, certainly in biomedicine. That is, one never hears of anyone proposing that a pharmaceutical firm should fund the Food and Drug Administration to evaluate a new drug for approval, because this would be incompatible with the system of checks and balances necessary to prevent bad drugs from getting out on the market. Similarly, we have to be careful that bad science does not get out on the market. That does not mean that work that has been peer reviewed as a result of author support cannot be valid. It does mean, however, that the value of the information would be altered by such a mechanism.

When we talk about value of information, we tend to debate how much the sale value of the information is worth, or about the cost of actually producing that information. In undertaking any study at the National Academies that would look at the value of scientific information, however, one should not assess the value of information on the basis of the cost to produce it. Instead, the value of the information is its value as a tool, as a productivity multiplier in society. One could argue that scientific information, when it is peer reviewed and selectively filtered, has tremendous value to the author, but also to society. That kind of selectively filtered information is that productivity multiplier. Wiley, therefore, has very much supported a customer-pays model for the business because the company believes it ultimately enhances the value of scientific information for those who should value it most.

Joseph Esposito, SRI Consulting⁸

These are heady times for publishers. There is probably more interesting activity in the industry right now than at any time since the paperback revolution following World War II. Unlike the paperback revolution, however, which allowed publishers to grow enormously and to become highly profitable, the current state of affairs is often thought to present more challenges than opportunities for publishers. Indeed, it is not uncommon to hear people say that publishers are a historical artifact, who will be disintermediated by the Internet. Or, we hear that copyright is robbery and that the model for scholarly communications in the future is the open-source software movement, which has given Microsoft a challenger in the form of Linux. When you look at some of the more intriguing projects organized around

intellectual communities, such as the DSpace project that is being developed by Hewlett-Packard at Massachusetts Institute of Technology, even a commercial publisher has to concede that critics of traditional publishing may have a point.

At the same time, commercial publishing employs a great number of people, and it is unlikely that they will all simply go away because a librarian in Nebraska has learned to manage a Web server. It may be useful to consider how publishers see their own activities evolving and what the environment for scholarly communications is likely to look like in 5 or 10 years.

From a commercial publisher's point of view, the primary thing to be said about the academic research publishing market is that it is mature, which means that there is only modest growth in revenue, making it hard for new players to enter the market. There are conventional ways to deal with a mature market, and we should expect that publishers will try them all. We should expect, for example, to see the consolidation among publishers continue right up to the point of antitrust action. It is highly unlikely that any scientific discipline will have more than two or three information providers in the years ahead. Some observers believe that there will only be two or three information providers for the entire STM market, covering all areas of study. However, this would result in fewer operating synergies across disciplines.

The form that consolidation will take may change, however. In addition to the already familiar phenomenon of the big companies buying up the smaller ones, we should look for the smaller companies to link together in an attempt to become big companies. We also should expect to see big companies establish distribution arrangements with smaller companies by, for example, putting the content of the smaller companies onto the electronic platforms of the larger companies. This provides critical mass, which can be used to shoulder aside the offerings of organizations with a small number of publications. It is not necessary for Reed Elsevier to buy Sage Communications if they can cut a deal to deliver Sage's publications on the Elsevier platform, but the impact on the marketplace is the same.

A mature market will also intensify the downward pressure on prices. This is especially true in academic libraries, where the open-access movement was created in part to put publishers under pricing pressure. Publishers will likely counter this by targeting the market share of other publishers, rather than looking for significant increases in library budgets. In other words, this is essentially a struggle for market share. This is already happening. For example, Reed Elsevier is now offering libraries access to previously unsubscribed journals, not by charging for each journal separately, but by simply insisting in an increase in total expenditures over the prior year. Previous speakers at this symposium referred to this practice as super-sizing, or the big deal. In other contexts, this practice is generally called bundling or tying. Bundling will have the effect of greatly increasing the number of Reed Elsevier publications available through particular libraries, at the expense of having less well positioned publishers lose those customers entirely.

We also should expect commercial publishers to seek so-called downstream value migration and to target competitors that for various reasons are thought to be vulnerable. Currently, publishers sit in the value chain between authors and wholesalers, and wholesalers have direct relationships with librarians. By moving downstream, more publishers will attempt to disintermediate the wholesalers and reap the wholesalers' marginal revenue. Critical mass is necessary for this, which is another impetus to consolidation. Disintermediation strategies that do not provide significant new value to end users are probably ineffective, but that does not mean publishers will not try it.

What makes one competitor weak and another one strong is a matter of debate, but we are likely to see commercial publishers target the not-for-profits more aggressively in the future. The reason for this is that, rightly or wrongly, the not-for-profits are perceived to be slower to respond and less capable of defending themselves, though they often have substantial goodwill in the marketplace. There are many prospective targets, but one might name just two: the American Chemical Society (ACS) and the Online Computer Library Center (OCLC). The ACS is a target because of its sheer size. The OCLC became a target when it acquired the assets of NetLibrary, making its strategic direction appear to be increasingly like that of a competitor.

Meta-content is interesting in that it sidesteps the copyright issue. Content about content can take the form of bibliographies, indexes, and, most important of all, search engines. Publishers are keenly aware of open-access publications and are looking for ways to make money from them. One way to do this is to create search engines for open-access content. Even better is to integrate open-access content with proprietary content for search purposes, as this will tend to bring even users of open content to proprietary services because the breadth of materials included will be much greater. In effect, open-access provides publishers with lower costs for content development even as it makes charging for access fees possible. From an economic point of view, copyright may not be necessary for supporting publishing profits.

The real economic response to open access, however, is likely to be in the creation of Web services, a dynamic substitute for the publication of fixed content in hard copy. In a Web service, a publisher provides online software that manipulates or processes data that are uploaded to it by a user. The user creates the content and then pays the service provider for the online processing. Think of what it would be like to do your taxes online and you will get the idea. Copyright is irrelevant for models like this, but the economic potential is very great.

One final strategy, like many of the other ideas here, is already playing out in the marketplace. It is customer diversification. If the academic channel is mature, publishers will seek new sales channels. The most likely one, because of its size and creditworthiness, is sales to businesses. We should expect to see more publishers trying to modify their offerings for this channel. Since businesses, unlike research universities, are profit-oriented, the kinds of publications they are most likely to purchase speak to an immediate business need. This will tend to shift capital investment away from pure research publications toward applied research and engineering.

Looking out 5 to 10 years from now in the world of scholarly communications, we are likely to see an ample number of open-access publications coexisting with proprietary ones, and we will witness ingenious publishing strategies designed to extract economic gain even in the absence of copyright. The toll collector may have switched positions, but he still has his hand out.

Wendy Lougee, University of Minnesota

This presentation addresses journal costs and value, and the kind of tension that libraries feel as the intermediary between the publishers and the consumers of publications. These tensions may suggest

new models, new ways of looking at revenues. One aspect that has not yet been talked about, but that constitutes a behavioral constraint that should be taken into account, is the longstanding convention of the scholarly journal, going back three centuries. Journals gradually accrued the function of providing a place to document discovery, to disseminate ideas, and also to codify prestige. That kind of three-century convention is not going to be easy to change.

Libraries are the intermediaries between two types of economies. They buy the content in a market economy, and users create and use content in a so-called gift economy. Libraries are often caught in the middle. Libraries have witnessed decades of dramatic journal price increases, with average annual increases of 8 or 9 percent for the past 5 years. It is a very inelastic market, in that as the prices have gone up, the libraries have not necessarily withdrawn or canceled the journal subscriptions. Data over the 15 years ending in 2001 show that journal prices went up 215 percent, yet libraries canceled only about 5.1 percent of their subscriptions. The rates of increase are such that if you look at the Higher Education Price Index, the rate of journal price increases or library materials now outstrips utilities and health care costs. This is certainly cause for concern.

Michael Keller earlier cited last year's report by Morgan Stanley on the STM market. It serves as a primer for investors, shedding light on STM publishing strategies. The report highlighted Reed Elsevier's profit margin of 37 percent for core titles and 22 percent for the medical titles that it recently acquired from Harcourt.

The Morgan Stanley report further discussed three trends worth future attention. First noted was the cyclical slowdown in industry growth due to research library budget cuts, coupled with a downturn in investments. A second trend noted the benefits of scale increasingly accruing to large players. It highlighted, in particular, the kind of bundled portfolio of journals that has come to be known as the big deal, as has already been mentioned. And, finally, it noted that the profit margins will expand for those publishers who have a successful online platform. Interestingly, the report revealed that libraries spend an average of $1.50 on staff and operating costs for every dollar they spend on acquiring content. It speculated that the saving for libraries in moving to electronic-only would be somewhere on the order of 16 percent.

The libraries have experienced a different range of models than publishers, some of which have been discussed at this symposium. These various models frequently come with a requirement that the library sustain its commitment to a certain level of spending, and they give the library very little predictability or control over costs. As the publishers are experiencing these cyclical effects, there is a tendency to stay the course with the expectation that the level of commitment by subscribers will remain stable.

This past year we have seen two trends reflecting an interesting shift. The first is that the publisher strategy of the big deal—the multiyear, all-titles approach—has begun to unravel. We have seen a number of major libraries withdraw from that model, largely fueled by budget reductions and by the recognition that many titles in the bundle simply are not used. The libraries are beginning to push for much more finely tuned licensing models whereby they can select the content that they want—not only at the title level, but in terms of other kind of bundling, much like the tokens that Brian Crawford described.

Moreover, in the big deals, the libraries have encountered pricing that is very hard to defend to their user community. At the University of Minnesota, the library worked with Reed Elsevier to move away from its bundled deal, but found it hard to explain and rationalize the end result: A reduction from 750 subscribed titles to 650 titles, and a move from two formats—print and electronic—to electronic only, resulted in a higher per-title cost.

A second disturbing event this year, which also has been alluded to by several speakers at this symposium, is the demise of some of the other intermediaries, the serial vendors. There was at least one major incident of a serial vendor going bankrupt, resulting in a potential loss of $73 million in library subscription payments. This raises the question of where the functions of procurement and distribution will go. These are functions that have to be sustained somewhere. The likelihood of increased revenues for libraries in the near term (particularly increases that match inflation in journal prices) is low. A recent informal survey conducted within the Association of Research Libraries suggested that nearly half of respondents expected cuts in some areas and the prospects were high for further budget reductions in

the coming fiscal year. Library budgets, a major source of revenue for publishers, are obviously stressed. The volatility in the publisher marketplace will probably continue, as will the push from the library community for the more finely grained models that allow them to make some choices.

It is also important to look at the changes in format and content, how these changes might affect the existing service models, and the resulting costs for libraries.

There are data indicating that much more intensive use is likely to come with electronic content. There are also recent studies of university users nationwide that have revealed an overwhelming preference for electronic format. Nearly half of all faculty in most disciplines reported they use online materials for the majority of their work. In the biological and physical sciences and mathematics the percentage of information needs met through online resources exceeded 60 percent. Yet interestingly, despite that preference, other studies looking at the perceptions of convenience and ease of use show a dramatic gap in terms of how the library performs in delivering that content. Users are citing evidence of an inability to really control the content, to navigate it well, or to deal with the myriad different platforms and channels that it comes through. Libraries increasingly are challenged to deal with these shortcomings—that is, they are challenged to develop systems to penetrate the so-called deep Web and federate the content in a meaningful way.

An experiment done by the University of California system looked at the differences between print and electronic use by putting the print versions in storage in order to force the use of the electronic versions and to see what would happen. They found that the relatively modest number of requests to retrieve the print versions from storage was by and large because the electronic versions failed to have some attribute or content that was solely in the print versions, or where the fidelity of some of the representation was not as strong. This result was similar to the data collected in the PEAK study at the University of Michigan (in the late 1990s), which provided access to all the Reed Elsevier journals to 12 different institutions and found that 30 percent of the use was for information beyond the text of the articles. This was at a time when many of the publishers were not converting or indexing the whole journal. Some of the major publishers still have been unable to migrate all aspects of a journal (e.g., the “front matter”) into the electronic realm.

These issues of journal integrity are important to understand, because they obviously impact how quickly we can migrate to that electronic-only nirvana. They are also important when we think about the kind of unbundling that might be possible and when we look at different parts of a journal and how they are presented.

There is one last point about how content is changing. Perhaps it is a subtle trend, but we are seeing a shift from our concept of publication as product to the notion of publication as process. There are a number of examples where online publications take on an ongoing, dynamic form as users interact with content and comment. For libraries, which are in the business of managing copyrighted, fixed works, that presents a real challenge. It is a challenge, too, to the STM publishing sector in terms of pricing, how to develop models that support something that is not fixed or well bounded.

There is a marvelous book by anthropologist Sharon Traweek about going to live among the physicists. The analysis (called Beamtimes and Lifetimes⁹) sheds light on how the physics community migrated into the e-print form. It describes how the early, informal sharing of information (now represented in e-prints) functions to document what is going on, as opposed to the formal publication that occurs months later, which is seen as a ledger to give credits and debits to fellow scientists.

As the divide between content and publication begins to occur, what is the role of the libraries? How do they manage and sustain formal publications and more informal processes? What influence do they have in affecting change with publishers, or with scholars and researchers for that matter?

The Morgan Stanley report suggests the potential for reduced operating costs for libraries—no journal check-in, no binding, no claiming lost issues. However, the counter-argument might be that the necessary infrastructure to support the investment in that content, to federate it appropriately, to ensure its longevity, and to archive it is going to be an increased expenditure on the library side. What we are likely to see is a shift of any savings to supporting the additional infrastructure. It underscores how critical it is to have community attention to this issue of infrastructure, to interoperability, and the kinds of protocols that will allow that federation to happen.

What influence might libraries have? It is rare that collective action is robust and sustained. A recent exception occurred a few years ago when Nature proposed to embargo content for institutional subscribers. Economist Ted Bergstrom has suggested an analogy with the economic construct of coordination games. He has a wonderful parable where he talks about the annual meeting of anarchists. They are used to going to the same hotel, but all of a sudden the hotel decides to raise its price, and they do not know how to coordinate action to change the venue for their meeting. Like the anarchists, librarians and scholars may be stuck in the context where it is impossible to agree on or organize collective action. And, when they do attempt coordinated action, they are accused of collusion.

Libraries do have a role, however, in seeding and developing alternative, competitive publishing models. Libraries have a fundamental understanding about content, its use, and users. There are a number of examples, of which HighWire Press is certainly one. There also is an increasing incidence of institutions, such as Cornell and the University of Michigan, starting incubator services, production services to help small publishers move to electronic publishing. These projects represent a move by the libraries away from their traditional role of providing access to information toward facilitating production of information, and it may help them reconceive the relative position they may have in the STM information sector.

To summarize, libraries play an interesting role in the middle in a time of shifting costs and resource constraints. The near-term focus for libraries is likely to fix on achieving a more finely tuned approach to acquisitions, while at the same time developing the necessary infrastructure for the future.

Patrick Brown, Stanford University

The Public Library of Science (PLoS) is a new scientific publisher based on an open-access business model, sometimes also called a midwife model. Bruce Alberts and James Duderstadt have already pointed out that scientific publication has traditionally been viewed as a public good by the scientists in universities, the sponsors of research, and the citizens whose tax dollars pay for much of the research that gets published. The goal of scientific publication from the perspective of these stakeholders has always been to make the results of scientific research as widely available and useable as possible. None of the important constituencies are benefited by limiting access or the freedom to use the published scientific information. This is the operating premise.

Before the Internet, the hard reality was that there was no choice but to charge the readers and users of scientific publications, because the most efficient way of distributing this information was by printing it on paper and delivering it in boats and trucks. Every potential user thus represented an incremental expense for the publisher. So, of course, any business model that did not take that into account was doomed to lose money in proportion to the popularity and success of its publications. That system, however, had the unfortunate consequence of limiting distribution to individuals and institutions that were willing and able to pay, and this was accepted as a necessary evil. With the advent of the Internet, that model is no longer a necessary evil.

The traditional business model also had another indirect consequence that was more subtle, but equally unfortunate. It was based on selling research articles. This gave rise to the mindset that the content of scientific journals was valuable property that the publishers were somehow entitled to own, control, and sell for a profit. This is a really big psychological barrier that we now have to overcome.

With the worldwide spread of the Internet, something fundamental has changed about the economics of scientific publication, not just the technical means of distribution. There is an absolutely

categorical change in the economics that is wonderful, because it makes Jefferson's ideal of the infinite, free dissemination of scientific ideas and discoveries a realistic possibility. What had been an impossible ideal in the pre-Internet era—to make the published information an open public resource—is now possible because the cost to the publisher no longer scales with the number of copies produced or with the number of potential readers of a publication. That means that we are not restricted to a business model that charges per access or per copy. In fact, it means that a business model that restricts the distribution and use of the published work is really working against the interest of science and society. It is unnecessary, anachronistic, and inefficient. This business model now stands in the way of the ideal that we have been working for all along, and that is now potentially available to us. If we do not need to charge for access, then we should not charge for it.

The PLoS will use an open-access business model. An open-access publication is one that meets the following two conditions. The first is that the copyright holder (either the author or the publisher, if the copyright has been transferred to the publisher) grants to the public a free, irrevocable, perpetual right of access to the work, and a license to copy, distribute, perform, and display the work and to make and distribute derivative works in any medium for any purpose.

The second condition is that not only do they grant that license, but they actually make it possible for people to access the work. They do that by making a complete version of the article and all supplemental materials available in some suitable standard electronic format, deposited immediately upon publication in at least one internationally recognized, independent online repository that is committed to open access. One example of such an open-access archive is PubMed Central, maintained by the National Library of Medicine. The PLoS plans to make everything that it publishes immediately and freely available online, through its own Web site, PubMed Central, and any other distributor or library server that wishes to house and distribute its journals.

Under this definition, “open access” does not mean just some free peeks or downloads from the publisher's Web site. It means that everything PLoS publishes can be downloaded in its entirety, manipulated, copied, redistributed, integrated with other data, and even resold by commercial publishers. The point is that the content is not the publisher's property. The publisher in this model plays the role of a midwife, an expert at delivery that gets paid for the service but does not expect to keep the baby.

The advantages of this kind of true open access are already very familiar to everyone in the life sciences. They are certainly not theoretical. In the life sciences there are two longstanding, amazingly successful experiments in open-access publication, GenBank and the Protein Structure Database. The success of the genome project, which is generally considered to be one of the great successes of recent times scientifically, and perhaps even its existence, is in no small part due to the fact that the world's entire library of published DNA sequences has been an open-access public resource for the past 20 years. If the sequences could be obtained only in the way that regular published work can be obtained, one article at a time under conditions set by the publisher, there would be no genome project. The incredible value of genome sequences would be enormously diminished.

More significant is the fact that open access is available for every new sequence, and it can be compared to every other sequence that has ever been published. The fact that the entire body of sequences can be downloaded, manipulated by anyone, and used as a raw material for a creative work has meant that thousands of individual investigators on their own initiative have taken up the challenge of developing new data-mining tools. It is such tools and the new databases that incorporate sequences, enriched by linking them to other information, that have really made the genome project the big success that it is. With open-access publication, there is absolutely no reason to think that we would not see at least as great a flowering of new, investigator-initiated research using that wealth of published material as a resource for creative work, and adding to its value.

Unlike the subscription-based model discussed by the previous speakers, to pay for the cost of publication the PLoS plans to charge the cost to authors and their sponsors. From the standpoint of business logic, this is by far the simplest and most natural model. It is natural, because the cost of online publication is scaled with the number of articles, not with the number of readers. It also makes perfect sense from the standpoint of institutions that pay for research. Their mission is to promote the production and dissemination of useful knowledge. From that perspective, publication is inseparable from the research that they fund.

The PLoS initially plans to charge about $1,500 for a published article, with no charge to authors who cannot afford to pay. That figure is based both on a ground-up analysis of the publisher’s cost at steady state and also on a survey of publishing costs from a number of publishers who were kind enough to open their books to PLoS.

Research sponsors should welcome this model, because for a fraction of 1 percent of the cost of the research itself, the results can be made truly available to all, not just to the fortunate few who are at the lead research institutions like Stanford that can afford to pay for site licenses. It can include the students and faculty of community colleges, high schools, and smaller institutions; scientists in poorer countries; and patients—and there are plenty of them—who want, on their own initiative, to read about the latest progress in research on their condition and are unable to pay exorbitant fees to read all the articles. This model therefore would greatly increase the benefits that are realized from the public research investment.

In the short term, of course, there will be an incremental expense, but in the long term, once the scientific community has made the transition to open-access publication, there will actually be savings to the major research sponsors because they are ultimately the ones who pay for most of the costs of the subscriptions in libraries and even for individual subscriptions. The Howard Hughes Medical Institute (HHMI) is one of the largest funders of scientific research in the life sciences, and it has already endorsed this model. The HHMI has agreed to provide budget supplements to its investigators, specifically to cover author charges for open-access publications.

Those of you in the publishing business should take this as a carrot. The money is available only for publishing through an open-access mechanism, as defined above. It is deliberately intended as a gesture of how important HHMI considers open access and as an incentive for the researchers to choose that system. At least some other major funders of research in the life sciences are taking similar steps, and this should be encouraged.

The PLoS is also going to produce printed editions of its journals that it will sell to institutions or individual subscribers at a price that is intended just to recover the cost of printing and distribution, everything downstream of producing the published digital document. This is estimated to cost on the order of $160 a year, but that will depend on many things, including the size and the distribution of the journal, as well as other factors. If there is a market for the printed journal, which there very well may be, then the print journal operation is not going to be selling the content, it will just be selling a physical form of the content that some people may find convenient.

There will not be any cross-subsidies between the open-access online publication and the basically break-even print publication operation. That makes perfect sense, because now that these two versions are being decoupled, the system is much more flexible. The PLoS does not care how well the print publication business does. The market will decide whether it is something people value.

In closing, it is useful to re-emphasize a point that was made by the first two speakers at this symposium—that scientific publication is seen as a public good. This ideal of scientific publication being managed as a public good, as an open public resource, is now within our reach. If we, as scientific publishers, can work together, and work with the universities and the institutional sponsors of research, we can make the transition from the anachronistic, restricted-access business model that we have now, to a system that is economically sustainable, much better for science and society, and that provides free and open access to everyone.

Pat Brown began with a comment for Brian Crawford. He thought that Mr. Crawford’s argument in favor of restricted access, namely, that it had an important role in maintaining the quality of publications and that an open-access publication model would somehow undermine quality standards in scientific publications, was specious. There are two grounds for saying that. First, all the studies that he knows of, and there have been a number of them that have looked at the relationship between the price and measures of quality of scientific journals—citations, their assessment by peers in the field, and so forth—have found, overall, a dramatic negative correlation between price and quality. The notion that users paying for journals somehow upholds high-quality standards is not supported by the data.

Second, every scientist, at least in the life sciences, knows that most of the journals that are regarded as third-tier journals of last resort, but which publish 99 percent of the articles, have no author charges but very high subscription costs. The premier journals, however, typically wind up charging authors in color charges and page charges something on the order of $1,000 or more. So, all the evidence that Pat Brown knows of actually tends to argue against the point Mr. Crawford made.

Brian Crawford responded that Professor Brown may have extrapolated incorrectly from what he had said. He did not say there was a correlation between the price and the quality or value of the information. He was making the distinction between two different economic models—one where the customer or the user is paying, versus one where the author is paying. Dr. Crawford’s point was that there is an inherent bias in a system where the author pays, which ultimately would devalue that information in terms of its overall utility as a productivity multiplier.

There are two reasons for this. The first is that a less selective filter would be imposed. And second, going back to Hal Varian's argument, is that an economic system where the author pays is naturally going to favor the author. That means that any entity wanting to make decisions about that work needs to impose yet another filter at some cost, in order to determine how that work stacks up against others.

Pat Brown added that in his view, that second filter exists already, and there is no added cost for it. The factor that serves to maintain the quality of the work that authors submit for publication is not that they do not think they can ever get a paper published. One can always find a mediocre journal that will accept just about anything. Rather, it is that the authors know that sooner or later their peers are going to read what they write, and their reputation depends on it being good. That is ultimately what determines their career advancement, their status in the field, and so forth. So, that filter is already in place. It is not at the level of whether or not a paper can find its way into print.

Brian Crawford said he thought that most scientific publishers would say that selectivity is probably not exercised all that often by authors, because they seek to get their work out. Professor Brown did ask another question regarding the subvention of publishing costs by the payment of page charges or subvention for color reproduction charges. This is a good point, and an example of where some organizations have found an effective balance. They have done this even in a user-pays model, implementing certain charges that are passed on to the author, where the needs of the author are seen as unique and something that the author would want to pay in order to benefit from a service. Most organizations, however, do not make that a criterion for a decision of acceptance or rejection; it is merely a matter of presentation of the work.

Steven Berry, of the University of Chicago, said that Hal Varian's point that publishers would favor the author by charging the author is absolutely right, and that this is exactly what the scientific community would like to do. The public good that is produced by scientific research is very special. In contrast to other public goods, the value of scientific public goods increases with use, and it is specifically in the author's interest. It is very much a part of the motivation of the scientific paper author to have that paper as widespread as possible, more and more used, thereby enhancing the value to the community, as well as the interests of the author. Consequently, Brian Crawford’s quote of Hal Varian is very much an argument for placing the charges to the author, rather than to the user, who does not have a particular vested interest in increasing the use of each written piece of work.

Brian Crawford clarified his remarks further. It is absolutely correct that one needs to optimize two

variables. One variable is the dissemination, and the other is the filtering. His point was that the author-pays model moves the filter boundary, whereas publishers also have an interest in disseminating the work as broadly as possible. It is a matter of getting the balance right between those two considerations.

Professor Berry added that despite Mr. Crawford’s logic, the facts already indicate, as Pat Brown observed, that the quality of articles in the journals is not correlated with the users' charges; if anything, it is inversely correlated.

Donald King noted that one of the things that he has found in his collaborative research is that it costs authors on average about $7,000 to produce an article. Roughly one-third of the articles are funded by the federal agencies, about a third of them are funded exclusively within the universities, and the rest by industry.

There are two problems in this context. One is with the argument that Steven Berry was making. How will the funders of these authors be convinced that they need to pay an additional $1,500 above the $6,000 or $7,000 that they are already paying the authors to prepare those articles?

It is not an easy argument to make for all of those different sources of funds. For the government funding sources, such support can be written into the grants, although that has its own problems. About 75 percent of the readers of STM articles are outside the universities and the granting organizations, so they may not want to pay for page charges, especially if federal funding for research grants is reduced. Nevertheless, some good arguments can be made here. Where you are going to have even more difficulties, however, is with the people who are writing those articles, who are paid by the universities or industry. You will need to have different kinds of arguments for those three different constituencies. How are you going to get all of them to do it?

Pat Brown said he thought that the interests are similar for all three of those constituencies, namely, whatever motivated them in the first place to support the research and to encourage the authors to publish it. The motivation to pay an increment of less than 1 percent of the total research cost to make it much more valuable to the people who are supposed to be served by it is presumably the same for all three sectors that sponsor research.

Donald King continued that it may be necessary to go further in trying to understand the funding priorities and budget profiles in each sector. For example, one could look at what the universities are already purchasing through their subscriptions and get some notion of what the transfers of funds are. Once you have those data, you might be able to have even stronger arguments for the author-pays model.

Ed Barnas, journals manager for Cambridge University Press, said he has been involved in both scientific society and commercial journal publishing for about 30 years. One thing about the discussion of business models that has been ignored thus far is the question of the mission of the organization that is doing the publishing. That does have a significant impact on the business model selected.

For-profit, commercial publishers will seek to disseminate information in journals that are very attractive to the library market, the institutional market that provides them with the financial return on those journals. Not-for-profit, nonsociety publishers will seek to disseminate information also, but in a nonloss situation so that they can continue to stay in business and serve whatever their own mission is. Society publishers, however, have a significant investment and interest in serving their members, and in that regard, some of the comments about author support, page charges, and color charges may be seen in a different light.

When Mr. Barnas was working for the society publisher, or working with societies as a contract publisher, various societies would make use of the page charges as a way of supporting their member subscriptions, keeping the member rates low, so that they could provide more benefits to their members. A commercial publisher or a university press cannot really charge page charges, unless it is a journal they are publishing on behalf of a society because that is not viewed as appropriate. It would be seen as gouging. Color page charges might be acceptable, because they are a direct cost, but author page support for articles in non-society publications cannot be a viable part of current business models for non-society publishers.

Brian Crawford commented that he considered the points made by Mr. Barnas to be quite valid. One should not look at the customer-pays model, as opposed to the open-access model that has been proposed as an author-supported model, in absolute terms. Indeed, many professional societies have had a hybrid model, where they have benefited from subscriptions at the same time that they have used author page charges and other subventions to help support their publishing programs and to keep the costs to the customer down. That kind of hybrid model has been determined over time by market forces.

One could ask, Why are we debating now what the right business model is? Many of the business models that we have serve us well, and have been adopted over time.

Wendy Lougee raised the question of how publishers see the incorporation of the long-term preservation requirements in the context of the user-pays economic model, given the unknown future costs associated with that.

Brian Crawford noted that the focus of publishers in electronic publishing is mainly on the actual production of the work and its dissemination on the Web. They have not addressed the important issues that deal with long-term preservation, including the integrity of the information and its migration to new platforms. Reed Elsevier has begun a very sensible project with the Royal Library in the Netherlands, where they are looking at how to address those kinds of long-term preservation problems. They will be costly, and the costs will need to be shared between publishers and libraries, ideally with support from governments and other institutions.

Jeffrey MacKie-Mason asked Pat Brown to address that same question from the perspective of the PLoS. Professor Brown’s definition of open access included the immediate deposition of the publication with some organization committed to long-term access. However, there is no guarantee the organization will follow through on this commitment. One of the advantages of print publication in is that with enough copies produced, very long-term preservation is more or less guaranteed, which is not always the case with electronic publications. The PLoS will make print copies, so does it plan to deposit print copies in archival institutions as well?

Pat Brown responded that PLoS has a number of plans for providing archival stability for its information. What is important, however, is not only that somewhere there exists a permanently preserved copy of the information, but that it is permanently openly available to everyone. Of course, one would be hard put to find a more trusted and trustworthy archival repository than the National Library of Medicine, but the fact that the information is also going to be freely and widely distributed, and will exist in many institutional servers, provides another measure of assurance.

Wendy Lougee added that her question was focused more on the natural process involved in digital preservation—the durability of format, whether it will migrate, whether the publisher is doing all the right things to guarantee success—and then how to incorporate those requirements and their costs into the publishing business model. Will it be ensured through an escrow fund or folded into the base cost, and how will those costs be accounted for realistically when they are unknown in advance?

Pat Brown noted that it would be a terrible mistake for institutional repositories to agree to take on the job of archiving information without requiring the publishers to grant unrestricted open-access rights. To have them take on the burden of preserving information that the publishers still treat as their property, in his view is really an abuse.

Martin Blume, editor-in-chief of the American Physical Society, addressed the archiving issue, because his society has put all of its content going back to 1893 online. It is linked and searchable, and PDFs are also available. Current content is added as it becomes available. There will be costs in changing the format of this in the future. The society also recognizes the concerns of libraries and of the entire scientific community that this historical record might be lost. What would happen if the American Physical Society were to go under?

There is a full mirror site of the entire archive that is already accessible and tried at Cornell University, which is where The Physical Review originated back in 1893. The full content of that, from 1893 to 2001, is 378 linear feet of shelf space. The journals are also deposited at the Library of Congress. If the society is terminated, it has an agreement with the library that these holdings will be put in the public domain, freely available to anyone. This is the society’s primary effort for ensuring continued availability of its archived publications. It should be noted, however, that unlike the American Physical Society, Cornell is not an eleemosynary institution. The society, therefore, has provided the university with servers for keeping its digital publications available and also pays a fee.

Martin Blume also noted that the American Physical Society has an open-access model for one of its journals that it started five years ago—Physical Review Special Topics, Accelerators and Beams. It is a small journal, but it is available completely without access barriers. It is not, in fact, free because it does cost something to publish it. The costs are recovered by subscriptions through the sponsorship of 10 large particle accelerator laboratories around the world. It is also cosponsored by the European Physical Society's Interdisciplinary Group on Accelerators. At this point it is only a limited experiment because the American Physical Society has other expenses that it must recover, and it has to see whether these things work or not. The society cannot afford to bet the whole ballgame on the open-access approach because it still needs to recover its costs.

In a published debate in Nature on this topic, Dr. Blume did propose that the society would like to put everything online without access barriers. It could do this right now if every organization that now subscribes to its journals would make those subscriptions sponsorships, providing enough to recover its costs. Then it could be opened to everyone else. The risk would be that institutions might be tempted to decide that since all journals are available, they no longer have to pay anymore. The libraries would love that, but then the publisher would go under. If the society were to attempt this, and did not recover enough sponsorship in the future, it could have a “pledge week.” Every time a reader tried to download an article, there would first be a solicitation for contributions.

Doris Peter, from the Medical Letter, a not-for-profit tertiary publisher that publishes reviews of drugs approved by the Food and Drug Administration (FDA), expressed concern about pharmaceutical influence in publishing, and industry influence on publishing generally. She wanted to know if PLoS is going to accept sponsorships from industry.

She said that there appears to be some support for industry to be more involved in supporting STM journals. From the Medical Letter’s point of view, it purposely maintains its independence because it could not credibly publish independent reviews of drugs if it were to accept money from anyone that is associated with the pharmaceutical industry. In addition, Brian Crawford earlier cited a hypothetical situation in which the FDA might accept money from the pharmaceutical industry. In fact, the pharmaceutical industry does pay the FDA to review new drugs.

Pat Brown responded that he completely agreed with Ms. Peter’s point of view and that it would be a bad idea to accept such sponsorships. It is not something that the PLoS is contemplating doing.

Nick Cozzarelli, editor of the Proceedings of the National Academy of Sciences (PNAS), said that the idea of open access is a great one and he hopes that PNAS someday will be switching to that business model. The problem is that PNAS is a not-for-profit journal, with no endowment. The risks of going immediately to an open-access model would be too great.

His question to Pat Brown was whether he saw any paths or way stations between full open access and the current subscription-based model. Dr. Cozzarelli suggested that one possible approach might be for a publisher to give the author an option for a surcharge. The author could pay extra to have immediate open access to his or her article, whereas all the other articles of authors who did not pay the surcharge would be free in PNAS after six months. Are there any other ideas of some sort of possible intermediate path?

Pat Brown responded that he did not object to the surcharge idea in principle. The amount of the surcharge, however, must not be so high as to be a disincentive. It is not a crazy model, although he does not think it is the best way to publish. He realized the difficulty of making that transition. The PLoS had to get a large grant from the Moore Foundation to buffer the financial risk for its experiment. Without the grant, the PLoS could not have done it.

In addition, Martin Blume’s suggestion in the Nature debate was not a bad idea as a transitional approach. In fact, it was something Professor Brown would strongly urge the PNAS to consider, that is, to ask the institutional subscribers that now provide most of the revenues, mostly academic institutions that probably accept the philosophy that journals provide a public service, to continue to pay their subscription fees at the current rate for some interval of years to be specified, during which time the PNAS would make the transition to open access. With such a multiyear commitment of support, PNAS would have a stable revenue source that is not put at risk by making that transition. It can try to make the transition, at the end of which time it can determine whether it looks like it is going to be a self-sustaining model.

Professor Brown thought that would be worth doing. He would hope that the current subscribing institutions would not take the low road and try to undermine the process by saving themselves a little money. They could see that it is in their own best interest in the long run to encourage open access. He asked whether Dr. Blume’s society had actually proposed that to institutional subscribers, at least as a transitional approach.

Martin Blume responded that the American Physical Society is having some discussions with other U.S. organizations to start this. Of course, the situation is different in every country of the world. He added that this is not unlike other consortium agreements that are made, whereby well-off institutions offer to pay a little more, and then open access can be provided for all institutions. Unfortunately, he has found that even the big institutions have asked, “What's in it for us?” and have refused to go along with this approach. They are unwilling to help the smaller organizations. So, the society has made efforts along this line, but so far to no avail. Such an arrangement is not unlike what Ohio Link does, but they came up with some extra money that enabled the society to open its journals to the institutions in the entire state of Ohio. That is the kind of catalyst that is needed, a bit of extra money.

Pat Brown noted that it might be best to view this approach as transitional, because ultimately the sensible thing would be for the research sponsors to cover the publication costs as an essential part of their mission of promoting and disseminating research. In the short term it is necessary to catalyze the process. He asked if the libraries would be open to trying an openness approach as a temporary experiment.

Mr. Blume answered that the problem with the temporary approach is, What happens if it does not work? It is very hard to get people to resubscribe. Once a library has given up the subscription and used the money somewhere else, resubscribing becomes a new acquisition.

Pat Brown added that the sponsoring institutions could be provided an incentive to maintain support. For example, if Stanford provides $5,000, or whatever the amount may be, then everyone affiliated with Stanford would have open-access publishing rights in the journal. Therefore, it is a competitive advantage for Stanford to offer this. If you treat it as kind of a credit pool that could be drawn on by authors from the subscribing institution, then even in the open-access model, they are getting some special benefits beyond what the nonpaying institutions are getting. It becomes an added incentive.

Bruce McHenry, of Discussion Systems, raised the issue that everybody is still working in the paradigm of the old paper model of publishing, where there is a lot of prepublication work, because a big print run is needed in order to economize on the costs of distribution. The questions about archival policy, editorial policy, and open access all change completely if one moves to a model of continuous improvement of the materials, or continuous publication, where all peers have an opportunity to adjust the prominence of newly developed pieces. In that model, the world changes completely; this issue is discussed in more detail in the last two sessions of the symposium.

Pat Brown noted that a prerequisite for the kind of innovation that Dr. McHenry has raised is that the materials that are being continually re-evaluated are an open public resource. It is hard to see how this kind of approach would work, as a practical matter, if you still have a system in which every publisher's body of work is in a separate, restricted repository.

Bruce McHenry commented further that there are certain domains outside academia where that is simply not viable, because people need to be able to earn a living for doing intellectual work. In that case, they cannot make it free and open right away. However, after some period of time, as the prices go down, foundations, government, and corporations may decide to purchase the materials and make them accessible within their group or publicly.

Professor MacKie-Mason raised a question posed by a Webcast participant who works for a large society publisher. Nearly 50 percent of this publisher’s total costs are attributable to editorial, peer-review, and production processes—items other than printing, paper, and distribution. Advertising associated with paper journals underwrites more than 50 percent of the total cost of operation. If there

were less advertising in the electronic-only model, then although it may save on some costs, it actually would represent a major lost source of revenue. Would electronic-only publishing be a viable business model if you were not to charge even more than publishers currently do for electronic access that is coupled with a print version?

Brian Crawford responded that for journals with broad member or individual circulation, most advertisers still see print as their means of reaching that audience. They are not yet ready to move away from print advertising to online-only in scholarly publications, although this may not be the case with other consumer-oriented publications. It would be very risky to go to an online-only strategy for STM publications, if the publisher currently relies on such advertising revenue.

Steven Berry raised the issue of the sponsor-supported publication model, in contrast to the author or the author's institution supporting the publication. He believes that many researchers like the former model because it is the sponsor of the research who shares with the author the interest in having the product disseminated as widely as possible.

When the National Institutes of Health (NIH) supports research, it is because it is going to be published and made into a public good. Consequently, to fulfill the goals of supporting the research at all, the sponsor assumes a responsibility to see that the material gets published. For this reason, the HHMI model that Pat Brown described is a very good one, and many scientists who publish research would like to see that kind of approach propagate. Is there a way to persuade the less well funded agencies, particularly the National Science Foundation and Department of Energy, to take on the responsibility to pay for publication? This would be in contrast to the present model, which puts the author payment responsibility into a grant in which the author has discretion whether to spend that money for publication or to spend it for support of another graduate student or some other research cost.

Pat Brown answered that despite his best efforts, he is not sure if there is a way to persuade them. It is an open question. However, one can make a strong argument that the publication costs amount to less than 1 percent of the research costs in biomedical research, (although he does not know the percentage in other disciplines). If that is true, if you had to take a 1 percent hit on other aspects of the budget to do it, it is a plausible argument that the return on that investment would be extremely high, as opposed to the 1 percent cut from other areas, because all the grantees and all the research that an agency is funding would be providing much freer access to a more extensive body of information. Again, the purpose of funding of research is not just to serve the immediate community of the grantees, it is the wider scientific community and the general public that should be much better served by the information.

At the same time, when this issue has been raised with respect to NIH funding, many NIH grantees have objected to such a proposal if it were to come at the expense of a 1 percent cut in research funding. They argue that there is not enough research funding to go around now; so it obviously is a controversial proposal.

Ted Campion, senior deputy editor at the New England Journal of Medicine, commented that open access makes eminent sense, especially for anything of archival value, with GenBank being the ultimate example. Scientific information should become freely open to all. At the same time, the value and the future of any publication are going to be determined mainly by what authors want to do, where authors want to publish. The NEJM is open to all authors; 4,000 scientific articles are submitted to the journal each year, with no charge for submissions and no page charges if the paper is accepted. However, the NEJM does reject many submissions. The role of a biomedical or scientific journal is to be critical and selective. That is in part why authors want to publish there. If authors or their sponsors have to pay, is this selectivity going to be compromised? How is the PLoS going to exercise the functions of peer review and of being selective? Is that part of its model?

Pat Brown agreed that the journals most attractive to authors tend to be the ones in which they have the least chance of having their papers accepted. The PLoS certainly has factored that into its development strategy. It intends to be very selective from the beginning—not selective just on the basis

of whether the article is good enough to be published somewhere else, but selective on the basis of publishing papers likely to be of interest to a very wide audience as well, precisely because the PLoS considers this important in terms of developing the journal identity and as a magnet for submissions.

If the point was that adding author charges would be a disincentive to authors to submit to the NEJM, that would be highly unlikely. Professor Brown has looked into author charges quite a bit and has paid plenty of them to a lot of journals that are very selective. It is not at all uncommon for such charges to be more than $2,000 per paper. There is a prototypical archival journal in the life sciences, for example, that does not charge for submissions, but it does charge for color. It is $750 for the first color figure and then $500 for the second. About 20 percent of all the papers that are published in that journal (in the several issues that Professor Brown has seen) had, on average, two color figures. The authors were willing to pay extra money to add color to their articles published in a completely archival journal. In the vast majority of those cases, the colors were primarily to make it look better and not because color was essential for the scientific content. The point is that an author will not balk at paying extra money after putting a great deal of work into a scientific paper for all the world to see.

If we go to a model where the institutions are covering page charges as an essential part of research, it will make it even less of a disincentive to authors. In Professor Brown’s view, the New England Journal of Medicine has nothing to worry about in terms of going to author charges, and such a change would be better for the community it is supposed to be serving.

Bob Simoni, associate editor of the Journal of Biological Chemistry (JBC), which is a publication of the American Society for Biochemistry and Molecular Biology, commented that the JBC charges authors page charges. On average, an article in the journal costs about $1,000, which includes both page and graphics charges. This has allowed the journal to keep its subscription price quite low.

The journal also considers itself to be an open-access journal, though it does not comply fully with the current PLoS definition. Two years ago, prompted by all of the discussion that came up with PLoS, the JBC decided to find some way to make all of its articles free online as soon as they are published. The journal started something that is called PIPS, or papers in press. On the day a manuscript is accepted, the paper is published on the Web in PDF. It is free to everyone, and it stays free forever. Anyone with Internet access can read freely every single JBC article that has been published. It has been enormously popular with the journal’s readers and authors. The fact that it is published on the day of acceptance reduces the time to publication by eight weeks on average, which is the time usually taken for copy editing and processing; so it has served everyone well.