PRINCIPLES FOR A NEW SYSTEM OF PUBLISHING FOR SCIENCE
David E. Shulenburger
Provost University of Kansas

The conviction is now widespread that the existing system of scholarly communication through traditional journals is breaking down. The most visible symptom of the breakdown is the inordinately rapid price inflation of journals and the resulting subscription cancellations by libraries. Below I delve into the economic reasons for the breakdown, outline the suggested remedies and convey a report on the elements of a proposed replacement system of scholarly communications arising out of a March 2000 meeting in Tempe, Arizona. The Peculiar Market for Scholarly CommunicationsThe function of price is to ration the available quantity of a good such that it ends up in the hands of those who have the ability and desire to purchase it. Market economies, by definition, use the market to set prices and to distribute goods. Rising prices for a good preclude the purchase of that good for some, but provide an incentive for producers to produce more. Ultimately, production of the good increases and price is reduced by competition and supply to the cost of producing additional units of the good.

To economists, scholarly journals and journal articles are "goods." In recent years their prices have risen significantly and the quantity of journal articles has also risen. According to the Association of Research Libraries, serials subscription prices in U.S. dollars rose 207% between 1986 and 1999. During the same period, Ulrich’s International Periodicals Directory’s listing of titles increased by 55%. Is this increase in quantity-supplied evidence of a market moving toward equilibrium? Are we about to see falling prices in response to increasing supply? Perhaps. But if this is so I expect that such price behavior will occur in the long run, a period when, as Lord John Maynard Keynes observed, "We are all dead."

What keeps this market from having prices that respond to increased supply is its very nature. It is highly segmented, with the most prestigious segments being nearly completely price inelastic. These segments include those top few journals in each discipline that scholars and practitioners understand to be the repositories of the most carefully refereed literature. Publishing in one of these few journals on a repeated basis signals that an established scholar is among the elite; publishing a single article identifies a young scholar on a trajectory to join that elite. No reputable academic research institution in which the discipline is studied would seriously contemplate dropping a subscription to one of these journals.

It is not easy for a journal to enter this elite group. Because these journals publish what is acknowledged to be the best material in a field, new scholarship is sent there first and appears in journals in lower tiers only if it is first rejected by those in the top. Journals can rise into this top tier but it takes much time to develop top tier cachet. Generally, the route to the top is through the emergence of a new sub field that is not initially published in the established top journals; for example, the emergence of econometrics spawned journals that are now in the top tier of economics journals. Importantly, it is not head to head competition that generally leads to the emergence of a top tier journal but expansion of the field.

Inelastic demand permits one to raise prices with subscription quantity falling less rapidly than price rises. Profit margins can be increased when prices are in this range and increased profit margins can then be maintained. According to Brendan Wiley, the top four commercial publishers of scholarly journals, Wolters Kluwer, Reed Elsevier, Wiley and Plenum had average net profit margins of 18.8% in 1997, compared with the U.S. Standard and Poor’s periodical industry net profit margin of 5%. Further, looking and focusing on only one of the four, Wiley observes: "According to Reed Elsevier's annual report, the operating margin of the scientific segment ran at 40.28% (1997), 41.77% (1996), and 39.66% (1995) as a percentage of sales. As a point of comparison, Microsoft's operating income as a percentage of sales was 45.17% for 1997, 35.50% in 1996, and 34.33% in 1995." Clearly, the leaders in the commercial segment of the industry have been able to develop abnormally high profits and maintain them overtime, an artifact consistent with relatively price inelastic markets.

But increased prices also can result from cost increases. Do price increases in this industry reflect cost increases? Undoubtedly some do. But if one examines an industry utilizing roughly the same technology to produce its product and finds that price increases are far less than in the serials industry, much doubt is cast on the thesis that price increases follow cost. According to the Association of Research Libraries, monograph prices were increased only 65% during the 1986 to 1999 period while the serials industry increased prices by 207%. When one examines profit margin and price performance data from a related industry, it appears that ability to set price, rather than cost, is driving price.

Recent research confirms that control over larger number of journal titles affects journal pricing. Mark McCabe concludes, "Briefly, our results for journals sold by commercial publishers indicate that prices are indeed positively related to firm portfolio size, and that mergers result in significant price increases." The many successful attempts by large commercial publishers of scholarly journals to acquire individual journal titles and to acquire other commercial publishers is consistent with the thesis that firm behavior is directed toward maintaining inelastic prices.

I am the provost of a research university, the University of Kansas, and chair of the Council on Academic Affairs of the National Association of State Universities and Land Grant Colleges, a council made up of the provosts of over 200 research universities. My concern and that of my colleagues is the effect of the price system on allocation of access to the research literature. At member libraries of the Association for Research Libraries, the number of serials subscriptions dropped 6% from 1986 to 1999, while the number of new monographs purchased dropped 26%. Thus, less scholarly material is being subscribed to in absolute terms. Furthermore, since the amount of scholarly material extant has increased, as the 55% increase in serials titles cited above demonstrates, faculty at Association for Research Libraries institutions have access to a reduced relative proportion of the scholarly literature than in 1986.

Since scholarly literature is the building block for new scientific discoveries, a prima facie case can easily be made that reduced access to the research literature reduces the potential for scientific progress. However, does reduced access through libraries equate to reduced access to the research literature altogether? There is abundant reason to doubt that the equation holds. For example, James Langer, the president of the American Physical Society, states that some of his colleagues, especially the string theorists, "for research purposes don’t need print journals at all." His colleagues rely on the Los Alamos server, a repository of unrefereed manuscripts. A common experience we all share is membership on list serves that present manuscripts reporting what colleagues in our own areas are doing. A search of any university website turns up manuscripts on many topics. This so-called "gray literature" provides access to scientists to much that is not owned or catalogued by our libraries.

But while the gray literature may provide access to material unavailable in one’s university library, it is not an acceptable substitute for library access to the scientific literature. The gray literature generally is not adequately indexed and archived such that one can systematically search or rely upon it as a repository of scientific information in the future. One can easily make the case that relying upon it as a source of transmitting and preserving the scientific literature will create great difficulties or impossibilities in accessing that literature in the future. (The author has cited some of the gray literature of the scholarly communications movement in footnotes. While web sites for this literature are accessible now, no assurance can be given that they will remain accessible 75 years into the future.) In addition, much of the gray literature is accessible only to those who are in certain social or professional networks. This system is hardly conducive to access to the literature by new scholars and scholars in the third world who are not privileged to have such access on their desktops.

Using price to ration access has resulted, in my view, in an inappropriate loss of access to the literature. Rationing by price is inappropriate when use of a good by one individual does not reduce its availability or utility to other individuals. Such is clearly the case with scholarly literature. Unlike physical goods, access to and knowledge of the scholarly literature may well increase the societal value of the good. Economists call such goods "public goods." An example of such a public good is the polio vaccine where increased access and use improves the lives of all those in the community. With goods whose public benefit is great, market societies often choose to either subsidize access or to make them available for free. Market- rationed access limits availability and reduces benefit.

It is especially curious that we have fallen into market-rationed access to scholarly communication because the public generally funds the production of the research that is reported in scholarly journals. The ultimate funding source may be a publicly-funded research grant, a taxpayer-funded institution, student-paid tuition, etc. The research costs are funded by the public, the results of that research are written by the scholars that produced the research, submitted to a journal and refereed by other scholars, and then given in total to a journal that edits it, publishes the manuscript in a journal, and then charges whatever price the market will bear for that research. Note that the journal is not constrained to charge for the value that it added to the article by the publishing process; it is able to charge for the full value of the article even though it added only a fraction to its value.

Logic of this sort led a group of Association of American Universities’ provosts to declare that the following principle should apply to scholarly literature: "Common ownership of goods… holds that research and scholarship are products of social collaborations and are assigned ultimately to the community." Because of this conviction and the generally held conclusion that it is inappropriate to ration scholarly literature by price, many proposals have been made and many efforts are underway in the United States to regain access to the literature for the community. Below I briefly describe some of those proposals and efforts. Proposed Remedies

• Increasing Library Budgets. While many institutions are doing this, few are succeeding in keeping up with cost increases; I know of none that is keeping up with the pace of additions to the literature. Increasing budgets is doomed to failure as a strategy for increasing access to scholarly literature if employed on a large scale because demand for key segments of the scholarly literature is price inelastic and the supply of top ranked journals in each field is also price inelastic. Increased budgets for library material will simply cause prices for scholarly journals to increase even more rapidly. It is clear that larger budgets will not cure this crisis.

• Preprint Servers. The most famous server is at Los Alamos and serves the physics community. Its cost for making the literature available to the community is only $1 per article. While it is a source of unrefereed material, it is very widely used by the community. Unfortunately, the widespread availability of literature through this source has not led libraries to drop subscriptions to physics journals. Thus, it has not reduced the libraries’ cost of accessing the literature.

• Open Archives Initiative. This initiative aims to create archives that are accessible and interoperable in order to efficiently distribute scholarly information. The effort grows out of e-print servers like the one in Los Alamos. At various sites, software that will enable institutions to place scholarly works on local servers and to have that material accessible worldwide as part of a virtual collection is in beta testing. The specifications for the system arose out of the Santa Fe convention, a meeting of knowledgeable parties held especially to establish such specifications.

• Minimal Refereeing Servers. The U.S. National Institutes of Health have put in place a resource called PubMed Central. One part of PubMed Central is a free electronic distribution source available to journals that choose to use it. The other part was a proposed minimal refereed service, which was supposed to make articles available to the biological/medical community in a manner similar to that of the Los Alamos server. The medical community in particular found unacceptable the distribution of any lightly refereed or unrefereed literature, especially if anyone could infer from the distribution source that it was "approved" by a government agency.

• SPARC (Scholarly Publishing and Academic Resources Coalition). This coalition aims to provide alternatives to the commercial press and has spawned a number of ventures, among them are journal titles that directly compete with those published by large commercial houses. Other SPARC ventures attempt to keep the scholarly literature in society hands and make it more easily and usefully available to scientists. SPARC’s members provide a market for the products they spawn and help assure them of success. I sit on the Board of Directors of an innovative SPARC effort called BioOne. BioOne is an electronic collection of approximately 40 whole organism biology journals that will be made available electronically in a single, searchable package in March of 2001.

• Antitrust Activity. U.S. antitrust authorities have yet to initiate a major action against an anticompetitive move by large commercial journal publishers. The difficulty is the criteria used in pursuing cases. Generally, no publisher holds a proportion of the market sufficient to trigger action when it acquires another journal. The Association of Research Libraries has asked that the Justice Department investigate the proposed acquisition by Elsevier of a portion of Harcourt General, including Academic Press. I note that European antitrust authorities have successfully opposed some combinations.

• Decoupling. This is a proposal by a group of Association of American Universities’ provosts that disciplinary societies set up boards to referee manuscripts so that refereeing could be dissociated from publication. The logic is that faculty evaluation groups that decide issues of tenure, promotion, salary, etc., attach great prestige to authors whose papers have been accepted by top tier journals. The ability of a journal to transmit such value to those who publish in its pages in turn creates great demand for that journal. Providing a method of refereeing papers separate from the journal review process would break this chain and reduce the ability of the journal to create value for itself by refereeing faculty work. Many U.S. academics fear that decoupling would result in monopoly review processes by societies that might stifle creativity and reduce scientific progress. At this point no society has established a decoupled review process.

• Buying Cooperatives. Many library coalitions exist in the U.S. and elsewhere that aim to use the monopsony-like power of the associated libraries to offset the monopoly-like power of commercial publishers. Generally, such cooperatives are successful in moderating price increases or softening restrictive use clauses. Unfortunately the cooperatives’ members do not have the freedom to tell the suppliers of journals like Brain Research they will not buy the journal if their conditions are not met as such journals must be part of selected research library collections. Without this ultimate bargaining power, the cooperatives’ impact on prices is severely limited.

• NEAR (National Electronic Article Repository). NEAR is a proposal by the author of this paper that a publicly accessible repository be established into which all articles would be placed 90 days after journal publication. This proposal aims to limit the exclusive ownership of manuscripts by the journal to a 90-day period. Because those manuscripts would be freely available 90 days after publication, a journal's ability to raise prices would be limited as libraries could simply wait 90 days and have the articles for free. This is a conservative effort aimed at providing publishers sufficient revenue to cover the cost of refereeing, editing and publishing the journal but returning the remainder of value to the public.

• Author Boycott. As of April 1, 2001, more than 13,000 scientists have signed the following journal boycott pledge: " To encourage the publishers of our journals to support this endeavor, we pledge that, beginning in September, 2001, we will publish in, edit or review for, and personally subscribe to, only those scholarly and scientific journals that have agreed to grant unrestricted free distribution rights to any and all original research reports that they have published, through PubMed Central and similar online public resources, within 6 months of their initial publication date." This effort incorporates elements from the PubMed Central and the NEAR efforts.

• Control of Societies by their Members. As faculty have become aware of the complex and serious problems confronting scholarly communications, they have begun to insist that their societies retain ownership of journals rather than selling them to commercial publishers and that the journals be operated and priced such that all journal revenue be used to support only the journal. Actions of this sort may keep the problem from growing worse, but will not cause prices of commercially owned journals to be reduced.

This impressive array of thought and activity reflects the discontent with the current scholarly communications system. Each of the activities/proposals has much individual merit. But one must be mindful of the words of H.L. Mencken: "There is always an easy solution to every human problem - neat, plausible, and wrong." While any or all of them may have roles to play in the future of scholarly communications, they do not now form a system on which scholars can rely for distribution and preservation of scholarly communication. The effort to develop the parameters of such a system is described below. Development of Principles for Emerging Systems of Scholarly CommunicationsBecause of the widespread recognition of the severity of the problem, the Association of American Universities, The Association of Research Libraries, and the Merrill Center for Advanced Studies of the University of Kansas convened a gathering of knowledgeable university presidents, provosts, librarians, representatives of scholarly associations, librarians, heads of university presses and representatives from the ARL and AAU to attempt to establish principles that should govern a new system of scholarly communications. Only representatives of the academy were invited, as our aim was to describe the conditions creating a scholarly communications system that would serve the academy.

The gathering produced the following principles that were agreed to unanimously by the participants. The Association of American Universities, the Council on Academic Affairs of the National Association of State Universities and Land Grant Colleges and the Association of Research Libraries have endorsed it for discussion on university campuses during this academic year. Many vigorous discussions have already been held and there is much support for the principles in the U.S. academy.

The Principles certainly are applicable beyond the boundaries of the United States. In fact, as the scholarly communications system is worldwide, solutions must be also. I am very pleased for the opportunity to report the Principles to this conference in the hope that they will lead to development of a new worldwide scholarly communication system. The verbatim principles and accompanying text as adopted by the signatories are reproduced here.

Principles for Emerging Systems of Scholarly Publishing

May 10, 2000

The following set of principles was agreed to by the undersigned individuals as a result of a meeting held in Tempe, Arizona, on March 2-4, 2000. Sponsored by the Association of American Universities, the Association of Research Libraries, and the Merrill Advanced Studies Center of the University of Kansas, the meeting was held to facilitate discussion among the various academic stakeholders in the scholarly publishing process and to build consensus on a set of principles that could guide the transformation of the scholarly publishing system.

The creation, dissemination, and application of new knowledge are fundamental to the development of an informed citizenry and a healthy global economy. Institutions of higher education exist to fulfill these functions. From the lab to the classroom to industry to the public, the advancement of knowledge through research and teaching is an invaluable contribution made by higher education to the public good. Scholarly publishing is the process through which newly discovered knowledge is refined, certified, distributed to, and preserved for researchers, professors, students and the public.

The current system of scholarly publishing has become too costly for the academic community to sustain. The increasing volume and costs of scholarly publications, particularly in science, technology, and medicine (STM), are making it impossible for libraries and their institutions to support the collection needs of their current and future faculty and students. Moreover, the pressure on library budgets from STM journal prices has contributed to the difficulty of academic publishers in the humanities and social sciences, primarily scholarly societies and university presses, to publish specialized monograph-length work or to find the funds to invest in the migration to digital publishing systems. Numerous studies, conferences, and roundtable discussions over the past decade have analyzed the underlying causes and recommended solutions to the scholarly publishing crisis. Many new publishing models have emerged. A lack of consensus and concerted action by the academic community, however, continues to allow the escalation of prices and volume.

The participants in the Tempe conference came together with the hope of building consensus on a set of principles that would inform the design and evaluation of new systems of scholarly publishing. The goal was to provide guidance while leaving open to creativity and market forces the actual development of such systems. The following set of principles is the result of their discussions. While the principles and their explanations reflect a North American perspective, the participants recognize that the advancement of knowledge and scholarly publishing are international enterprises. While the academic community in North America may agree on collective action, international discussion and support will be needed for the success of any new systems.

The participants encourage broad discussion and endorsement of these principles by institutions of higher education, scholars, scholarly societies, and scholarly publishers. Endorsement carries with it the commitment to implement local actions that will bring institutions of higher education closer to the goal of providing access to all relevant published research across all disciplines to all faculty by way of systems that ensure dependable management and affordable access to information over time.

• The cost to the academy of published research should be contained so that access to relevant research publications for faculty and students can be maintained and even expanded. Members of the university community should collaborate to develop strategies that further this end. Faculty participation is essential to the success of this process.

With the creation, dissemination, and application of new knowledge central to their mission, institutions of higher education must work to create systems that will provide affordable access to all relevant published scholarship across all disciplines for researchers, teachers, and the broader public. To do this, faculty, university administrators and professional societies must work together to create the systems that will contain, and in some cases, reduce substantially the costs of scholarly publishing. Since every faculty member should have access to all the relevant published research in her/his area, it is imperative that we find ways to bring down the cost to accommodate the expanding volume of publication within available budgets. The business arrangements of the journals for which faculty write, edit, and review must become a major focus of contributors, editors and readers if scholarly publication is to become affordable again.

Containing costs might be accomplished over time within the current configuration of scholarly communication through the effective use of technology to streamline publishing functions, while increasing access and value. Such systems have been developed within the not-for-profit community by Stanford University's HighWire Press and The Johns Hopkins University's Project Muse; other efforts, such as BioOne, are being facilitated by SPARC, the Scholarly Publishing and Academic Resources Coalition. One could also envision systems that would build peer-review and abstracting and indexing functions on discipline- or institution-based e-print services. Such a system is being promoted by the Open Archives initiative, an effort that strives for compatibility among e-print services. Cost-containment should also continue through library consortial purchasing of electronic resources, a strategy that appears to be effective in lowering the unit costs of electronic information. Whatever the solution(s), cost must be made to fit within available budgets or the system will fail to provide the information to scholars that they need.  

• Electronic capabilities should be used, among other things, to: provide wide access to scholarship, encourage interdisciplinary research, and enhance inter-operability and search ability. Development of common standards will be particularly important in the electronic environment.  

  With the growing volume of scholarly research, it is increasingly difficult to uncover all of the relevant material published on a given subject. As more scholarship becomes available in digital form, this problem can be surmounted through powerful search systems, provided that commercial, technical and legal constraints do not prohibit such searches. Searching, navigation, and linking across titles and across disciplines is essential since many disciplines have multiple titles that serve them and many problems have multidisciplinary aspects that may lead a researcher to publications in fields as diverse as microbiology, law, economics, and internal medicine. The development of standards is critical to the implementation of cornfield searching and navigation. In addition, given the importance of older literature to the advancement of new knowledge, retrospective works should be digitized and made accessible online.  

• Scholarly publications must be archived in a secure manner so as to remain permanently available and, in the case of electronic works, a permanent identifier for citation and linking should be provided.  

  The advancement of knowledge is dependent on access to prior scholarship. While research libraries, with significant support from the National Endowment for the Humanities, have made significant progress in preserving print publications, there is still a large proportion of unique printed material yet to be treated and a number of additional formats, such as videotapes, sound recordings, and film, whose preservation needs have yet to be addressed in any significant way. Electronic publishing adds yet another set of complex issues to the archiving and preservation of scholarly works. With libraries no longer owning copies and with the fragility of the electronic media, questions of what should be archived by whom and how are critical issues that need to be addressed. Despite many unanswered questions and unknown costs, archiving and preserving scholarly publications in all media are critical to any credible system of scholarly publication.  

• The system of scholarly publication must continue to include processes for evaluating the quality of scholarly work and every publication should provide the reader with information about evaluation the work has undergone.  

  The academic community relies on the judgment of peers when assessing the quality of faculty work. While core archival journals are expected to preserve the peer-review process, the scholarly community recognizes that the exact nature and methodology of quality assessment varies by discipline. Any evolving system of scholarly publication should allow for an evaluation process to take place as appropriate and should provide a transparent mechanism that informs the reader - an expert, a student, the public - of the nature of the evaluation the work has undergone in its various versions. This recommendation recognizes the development of discipline- or institution-based collections of articles which may go through different stages of review and where neither the hierarchy of existing journals nor the reputation of the publisher may exist as a signature of quality assessment.  

• The academic community embraces the concepts of copyright and fair use and seeks a balance in the interest of owners and users in the digital environment. Universities, colleges, and especially their faculties should manage copyright and its limitations and exceptions in a manner that assures the faculty access to and use of their own published works in their research and teaching.  

  The role of copyright is central to the academic community's mission of advancing knowledge. Members of the community are both creators and consumers of scholarly publications. As creators, faculty depend on copyright to protect the integrity of their work and on fair use to be able to use and incorporate the works of others with attribution in their own work. By tradition, faculty have transferred without direct compensation all of their copyrights to journal publishers in return for the wide distribution of their work. In some cases this tradition has resulted in the need for faculty to seek permission and pay a fee to use their own work in their research and teaching. If the academic community is to achieve its mission of advancing knowledge, it is critical that faculty authors retain the rights to use their own works in their teaching and in subsequent publications. Widespread adoption of university policies requiring faculty to retain such rights could provide individual faculty with the bargaining power to negotiate such agreements with publishers.

  While this document concentrates on copyright and fair use of scholarly works, the importance of copyright and fair use go well beyond the scholarly publishing system. It is imperative that the academic community monitor and critically examine any new license arrangements or proposed legislation (whether it be copyright amendments or any body of law affecting intellectual property directly or indirectly) and take appropriate action to make sure that such arrangements or legislation do not upset the balance between owners' rights and users' exceptions to them that has been achieved in copyright law with its provisions for fair use and library and educational exemptions.  

• In negotiating publishing agreements, faculty should assign the rights to their work in a manner that promotes the ready use of their work and choose journals that support the goal of making scholarly publications available at reasonable cost.

By judiciously assigning the rights to their work, faculty members can help assure that scholarship remains affordably available to the community. In the publication process, faculty can choose to publish in journals whose access and pricing policies make their work easily and affordably available. All faculty members should know the cost of journals to libraries and should consider refraining from submitting their work and assigning copyright to expensive journals when high quality inexpensive publication outlets are available. In fields where alternatives do not exist, universities and scholarly societies should work with faculty to develop such outlets.  

• The time from submission to publication should be reduced in a manner consistent with the requirements for quality control.  

  In rapidly evolving fields, lags of 12 months or more mean that scholarly history rather than cutting-edge research is the subject of publication. If published scholarship is to be a useful building block, it is imperative that the lag between submission and publication be shortened as much as possible for each field. While a number of factors contribute to the lag - peer review, author's changes, back and forth with editors - and are important to the quality of the final work, technology should be exploited to speed up the process where possible. For example, some journals have already designed systems that select reviewers based on workload and availability. In addition, a number of disciplines depend on e-print systems for quick distribution of their work.  

• To assure quality and reduce proliferation of publications, the evaluation of faculty should place a greater emphasis on quality of publications and a reduced emphasis on quantity.  

  While a fundamental factor contributing to the rapid increase in the volume of published research is the rapid expansion of knowledge, the academic credentialing system encourages faculty to publish some work that may add little to the body of knowledge. In the spirit of creating an environment that reduces emphasis on quantity across the system and frees faculty time for more valuable endeavors, faculty in research institutions should base their evaluation of colleagues on the quality of and contribution made by a small, fixed number of published works, allowing the review to emphasize quality. This de-emphasis of quantitative measures could moderate the rate of increase in new titles and numbers of articles published. Some universities have already modified faculty evaluation in this manner and federal granting agencies, such as the NIH, have implemented policies to limit the number of articles cited in the grant application process.  

• In electronic as well as print environments, scholars and students should be assured privacy with regard to their use of materials.  

  The digital environment, in particular, makes it very easy to obtain data on users and use patterns, information that can have great marketing appeal. It is incumbent on the academic community to assure the privacy of individual users with regard to their use of scholarly publications or other source materials made available through our institutions, consistent with state and federal laws.

Signatories to
Principles for Emerging Systems of Scholarly Publishing

Introduction

Are fundamental shifts taking place in how we scientists behave, organise ourselves, communicate and interact as a result of electronic technologies? Is it true that we are witnessing "a paradigm shift from the traditional, old fashioned print publication system, directly descended from 15th century Gutenberg, to a pioneering, telematic-based system of science communication" (Brown 1998)? Is this new powerful system really aimed at meeting the needs of everybody, including the traditionally "information- poor"? The answers to such questions may have a crucial bearing on the scientific enterprise and, more broadly, on the future of our relations with society. However, often the debate is conducted on parochial grounds, from the narrow standpoints of the individual stakeholders, and the ensuing conclusions can hardly serve the best interests of the scientific community, nor of society and culture at large. By gathering a broader selection of these stakeholders, the present conference provides an invaluable forum for the much needed collective reflection and debate on these matters.

It is certainly too early to arrive at definitive conclusions, as many important changes proceed at such a rapid pace. The first indications from some studies, however, suggest that a number of widely held assumptions about the nature and prospects of "virtual society" in general need to be revisited (see e.g. Woolgar 2000) and, as we intend to show here, this applies in particular to some of the promises and assumptions made about electronic publishing in science. Electronic technologies seem not always to affect us to the extent or in the direction we imagined or we were promised. For one thing, popular declarations that technologies will substitute for major social functions, contrast with the finding that they more often augment existing practice. The possible directions of the internet, in particular, are recognized to be strongly shaped by path dependencies in technology, accumulation of economic and political power and durability of social and cultural traditions. These kinds of counter-finding should allow us to objectively distinguish between the promise and the reality of life in the modern communications era, and they ought to help us make more realistic assessments of the advantages and potentialities of the new technologies in the world of science.

1. The paperless and the wireless society

Both Alice and Elmont are homeless in Washington, DC.
Alice approaches Elmont who is sitting on a park bench, to ask him if he'd like
to go for some soup. He tells her to be quiet as he is scrolling through his email.
There is a pause, and then Alice asks, 'Don't you need a computer for that?'
Elmont replies, 'Possibly. Researchers still aren't sure.'

Garry Trudeau's Doonesbury comic strip, February 1995.

Science fiction continues to inform future thinking. We live in a future- oriented society, as Sally Wyatt (2000) notes: "just as winners get to write (past) history, present-day actors attempt to write tomorrow's history, largely as a way of attempting to secure their current positions". This is particularly true for the internet, which unleashes all sorts of futuristic fantasy and speculation. But of course, history still matters because it makes some futures more plausible than others.

Not more than 20 years ago (Lancaster 1982), the paperless society was promised to us - to scientists, in particular - for the end of last century. And indeed in 1997 - barely four years ago- -there was suddenly a multitude of options for electronic journal literature from primary and secondary publishers, subscription agents and aggregators. Over the previous decade many libraries in industrialized countries, and a few in the rest of the world, had chosen to subscribe to one or more abstracting and indexing (A&I) services, which included the full text of articles in the database (either on CD-ROM or via the Web), thereby simplifying the process of providing their patrons with the journal literature. Progress in providing the full electronic journals, however, had been slower during that period; it had been noted that although electronic products might offer more information and articles than a library can acquire through print subscriptions, libraries seeking to replace their print subscriptions with electronic alternatives needed to consider that the options are not necessarily equivalent (Ogburn 1997).

Still today, the electronic and print versions of journals are not necessarily equivalent, and there are good reasons for making them different. Collection oriented academic research librarians are thus faced with offering the convenience of access to the content, at the possible expense of an archival copy for future reference. According to a fairly recent study (Vézina and Sévigny 1999) most actors in the world of scholarly documentation (authors, editors, librarians and readers) seem to agree that the printed copy is still useful and should be kept for a long period of time (if not forever), whilst the electronic version has become essential and should be used also to develop new services for end-users.

To this one should add other obstacles and constraints that make of electronic archiving and preservation of scientific material a still unresolved issue, such as: the non-existence of some relevant titles in electronic form (e.g. Brown 1998), the lack of technical support and reliable electronic infrastructure, and the uncertainty faced by libraries and end-users about future access - even to previously paid subscriptions. Under such circumstances, the transition from the paper to the digital world sounds as hardly realistic.

In addition, there are day-to-day practical obstacles: libraries seem to resent having a mix of titles forced on them in a bundle which includes the publishers' main titles, and having to cope with a variety of search protocols and authentication systems required by different distributors. So we see that even the big commercial publishers offer e-mail alerting services that send journal tables of contents directly to a reader's PC, "to provide the very latest information on soon-to-be-published papers and thus allow readers to reserve issues in their library" (Elsevier 2001).

There is, furthermore, some debate about the future proliferation of interface devices: a world in which the computer remains the dominant interface is seen as a North American one; whereas a world with a multiplicity of access devices looks more European, and a world with an uncertainty as to access at all is still the dominant perspective in most other countries around the globe. This raises different questions: What kinds of services will work well on what kinds of devices, at what costs, and for whom? Content is increasingly becoming tied in with access means; how will it be offered and/or sold?

And then, at a time when paper still does not show signs of disappearing, comes the new technological generation, that of the mobile or cellular phones and wireless communications based on packet switching, with its alluring promise of a worldwide instantaneous information distribution that recognizes no geographical or physical barriers. "People are going to stop carrying around things like laptops", it is said (Marks 2000); more and more devices are going to fit in your pocket. Indeed, we should not be surprised to see some laptop owners shifting to i-mode communicators pretty soon. Yet even for the community of scientists and scholars, who admittedly have been quite privileged in terms of modern communications, it is difficult to imagine that such expensive wireless communications will be made worldwide available in the near future. And, one should further ask: What about those who never had a laptop - or who have lived in an essentially paperless world? Would they be expected to double-leapfrog into this new paper- and wireless society?

Although many of the aforementioned constraints may be just transitory, it is becoming clear (once more) that new technologies tend to supplement rather than replace or correct existing practices and organisations. Against the common assumption that electronic technologies supersede existing activities and arrangements, research shows that the new technologies tend to be used in addition to, and alongside, old ones. The much boasted "paperless office" seems more illusory now than a few years back.

ii) Worldwide access to scientific knowledge and information

The web environment introduces in principle a new dimension in scientific communication and information, by functioning as a giant network that lends itself to linking people, documents and data, individually and in groups. Web-based community services are offering a plethora of appealing facilities. Scholars are becoming attracted by the idea of accessing information from their own desktop on a twenty-four hour basis, with instant local printout, immediate updating of information, modern navigational facilities and a general saving in time.

There is talk of the "empowerment" of the scientist as end user (Brown 1998), whilst the librarians' role is seen increasingly as that of a support service for the most efficient way of providing users with the information they want. There is no need to buy all the content of a particular journal because only a small proportion of the articles is required by every scientist - to whom the library budget could probably even be devolved, as part of this empowerment process.

Surfing the net with the purpose of identifying information-rich sources of data has become indeed a growing feature of end-user behaviour. However, this is not to say scientists are fully using the advantages of the electronic tools and technologies to communicate scientific knowledge (aside from the use of personal e-mail communication for this purpose); the electronic means serve rather for distribution and access to otherwise print publications (preprints, papers and reprints) and to data banks. In fact, there is still little clarity about the extent to which scientific knowledge - as distinct from scientific information and data - can be communicated by such means. The experiences with long-distance education and communication with other sectors will probably provide useful insight into these matters.

Further, it is not certain that the net with all its services that are to a large extent public at present (including the use of email, which is by far the most frequent activity via the net) will always be there for free. The signs are that the current rate of straightforward rapid expansion of internet usage will not continue (Woolgar 2000), and there are in fact some early indications of fatigue and disenchantment (El País 2000). Large numbers of young people - including students - have stopped using the internet, in those countries where usage has been more intensive; it has ceased being a novelty and people look for less virtual ways to communicate, read, play, or go shopping. Of course, those former users who can afford it, may "come back to" the internet, perhaps with the third-generation mobiles. But the fact is that on the whole, the internet has a commercial orientation and is mostly used for the search of merchandise, trips, entertainment and general information; its success is tightly linked to the expected commercial profits, which lately have shown clear signs of dwindling.

On the other side, there are still those who have never used the internet: at present (January 2001), whilst 90% of internauts are in industrialized countries, and 57% of them in North America (US and Canada), barely 1% are in Africa and the Middle East. For scientists and scholars, the distribution of percentages does not seem to be substantially different.

Table I gives a rough idea of the situation in Africa regarding internet facilities, both in the only five countries having in 1998 over 500 000 telephone lines, and in the one country which had less than 10,000 telephone lines (ISA 1998). Still in 1999, Liberia and Sudan reported no internet hosts, and another ten countries around the world had only one host (ISC 2000), whilst in the USA there were over 53 million of them.

We should not forget that internet connection still requires a telephone line, and at least 80% of the world population does not have access to one. In dozens of countries there is less than one line for every 100 inhabitants, and most of these countries are in Africa. In Cameroon, Mozambique and Tanzania, there was in 1998 only one institution with email access, with no connection to the library. At a university in Algeria with only one email terminal, staff members were expected to type their message on diskette and give them to the system administrator to be sent. Often in these countries the access modem is extremely slow and the server is very busy, which makes it difficult to download material from the internet.

Financial constraints on widespread use of the internet are evident. In several countries in Africa and Asia, university authorities give restricted net access to heads of department or senior staff via shared terminals "because dial-up access is expensive" (Lund 1998). Communications in countries in Africa with a very low average income level can be even more expensive than in rich countries, with local call charges costing more than US$4.-/hr in Guinea, Mauritania and Sierra Leone (ISA 1998).

Political and other external events may also have a disastrous effect on communication facilities in some developing countries: for instance, during the political crisis in 1997 the internet system in Sierra Leone University became completely inoperable because the system's operators had fled to nearby countries.

The sentence so often pronounced by aid agencies and their consultants, to the effect that "computers and telecommunications... provide an unprecedented opportunity for rapidly narrowing the gap between the information rich and the information poor" (e.g. Lancaster 1984), is a persistent fallacy. The most we can expect is that they narrow the gap "between the information rich in the West and their less well-informed equivalents in the Third World" (Olden 1987), allowing the latter to become part of the globalized information elite. In some countries in Africa, where literate people have always been an elite, computer-literates and those they serve will be a super-elite.

In addition, one should bear in mind that often the use of high technology in less developed countries produces new forms of dependence rather than enlarging independent capacity. Which is not to mean that Africans and people from other developing countries do not need the new information technologies: they do, but of the kind that local information professionals and users can understand, maintain, nurture, develop and ultimately internalize (Tiamiyu 1989). This applies in particular to the use of the new electronic technologies in science; we shall come back to this point under section 5.

3. Dropping journal subscription costs

Last but not least, there are those who, even being connected, cannot afford to pay for the information, and who apparently constitute a growing group.

Indeed, for many scholars electronic publication has failed to address the problem of accessibility: one of its promises, lower costs (irrespective of who has to pay them in the end), has simply not happened. There are strong indications, in fact, that consumers - scholars, their libraries, and their institutions - are paying for the development of electronic versions of scholarly information (Create Change 2000).

It is useful to recall that as late as the 1960s, scholarly communication was still part of an academic tradition. It was a manageable system, in which academics reported their discoveries, scholarly societies added value by vetting publications for quality, editing and publishing them, and libraries did most of the disseminating of the results. Then came a period of enormous growth in the scholarly and educational environment, first and foremost in industrialized countries and later, to a lesser extent, in most of the rest of the world. The pressure on faculty to publish and to get research support through grants also increased, and the quantity of research grew beyond the capacity of the scholarly publication system.

Commercial publishers then entered the scene, and began to absorb an increasing share of scholarly publication. With this stimulus, existing journals expanded and new journals were formed to accommodate the growing quantity of research in increasingly specialized areas. At the same time, information was starting to be viewed as a commodity that could produce profits for commercial publishers. Not-for-profit publishers began to see commercial publication as a way to manage precarious finances and weak publishing infrastructures and get rid of high overhead operations while still retaining quality control. Faculty were happy to see their chances to win tenure and promotion enhanced through publication and through appointment to editorial boards. The whole sense of journal publishing changed as scientists' interest shifted away from the readers' end and closer to the authors' end.

But already in the late 1970s there were the first signs that this scholarly communication system had to be maintained at a high cost, and by the mid-1980s cancellations of journals and reductions in other kinds of library purchases took place in many libraries. A chronic cycle started: cancellations caused higher prices and higher prices drove more libraries to cancel more journals. The dramatic increases in journal costs of scholarly publishing have in fact reduced scholars' access to essential research resources all over the world. Libraries simply cannot afford to keep up with the costs for commercially published journals, which are typically three to seven times as high as society or not-for-profit journals. A few recent figures (most of them taken from Create Change 2000) may serve to illustrate the situation.

Though the number of journals published worldwide has doubled in the last fifteen years, North American academic research libraries actually reduced their journal acquisitions, on average, by over 6% during the same period. During the twelve-year period ending in 1999, the unit cost of library subscriptions to scholarly journals increased by 9.0% per year (leading to a cumulative of 206%, in contrast to the 52% increase in USA consumer prices). Interlibrary borrowing among North American research libraries shot up 122.2% for the period 1989-1999 (8.3% annually); this increase suggests that scholars are having to depend on access from other libraries more and more.

Similarly, a recent report from Australia documents a 43.7% decline in total journal subscriptions for 38 university libraries between 1993 and 1998. During one recent year, 24 Canadian libraries cancelled $4.34 million (Canadian) in journal subscriptions.

The central library system of UNAM in Mexico, one of the largest university libraries in Latin America, has been forced to cut journal subscriptions, and book acquisitions by an even larger percentage, despite a considerable budget that has increased year by year. Librarians have been subject to pressure to buy electronic subscriptions in addition to the corresponding print versions, plus subscriptions to new electronic-only titles. It soon turned out that the titles offered initially for free or at low cost became expensive. Yet the library system has made an effort to maintain subscriptions to as many titles as possible, as academic staff and students are expecting to benefit from the conveniences offered by digital journal libraries and databases. For the time being, the ongoing changes in these services, merging of companies, appearance of new aggregators, etc., are resulting in frequent deficiencies in the provision of services to the end-user, despite the high costs paid.

Scholarly communication has become indeed an important business for commercial publishers, and mergers cause prices to rise even higher as competition decreases. Since 1995, eleven of the most expensive STM (science, technology and medicine) journals have shown a yearly increase of 10.6%; a library subscribing to just these eleven titles currently pays out $86,000 per year. The margin for commercial scholarly publishers is far higher than the median for all publishers, which is around 5% per year. The scientific and medical divisions of these companies seem to be the most profitable, with one company's science division operating with a 35-40% margin. Only the arts and humanities, where non-commercial publishers still dominate, regularly posted gains of less than 10% per year during the past decade.

There is a perception among some librarians that publishers of scientific journals could readily reduce their subscription prices by selling advertising. However, as remarked by Tagler (2000), advertisers speak with their spending power, and also their funds are limited. The majority of scholarly research journals have low circulation levels in the range of 400 to 1500, with an international spread of readers and with most of the subscriptions sold to libraries; hence their little chance of having a significant advertising revenue stream. Although there is still new ground to be covered in exploring the possibilities of web advertising, there are no strong reasons to believe that it will help to substantially reduce subscription prices of electronic journals having a limited readership.

The other side of the coin is no less dramatic: most learned societies have faced steady declines in library subscriptions during the past twenty years or more, at a rate of roughly 3% per year (Langer 2000). This is starting to force scholarly societies to revise their editorial policies and to reevaluate the roles of their journals and of the societies themselves.

This situation has its cruel ironies: universities pay to subsidize faculty research and then pay huge sums of money to buy the results back in the form of published content. A group of French specialists (Seminaire Villeurbanne 1997) argues that "we should answer ourselves the following, according to the nature of the text (commercial or not): Does the author write with the idea of selling his text? Is there a potential market ready to buy his text? If the answers are in the negative, it would be convenient for him to either pay for the publication, or post it as a preprint, or distribute it by electronic means to the potentially interested persons."

Probably much effort and money could be saved if our scientists and their institutions undertook a serious revision of the purpose of publishing. The scholarly and scientific communities are starting to respond in a more or less organized way to the present critical situation, not by undertaking such revision but at least by subjecting a variety of alternatives to discussion or promoting their implementation, such as (Create Change 2000; see also Shulenburger 2001):

• Creating competition to commercially published journals
• Creating more venues for scholars to communicate outside of commercial publications
• Supporting the self-publishing programs of scholarly societies
• Fostering alternatives to the current practice of surrendering intellectual property rights to publishers
• Fostering changes in the faculty peer review system that will promote greater availability of scholarly research: these changes might include movement away from quantity and toward quality as a criterion for tenure and promotion.

In this connection, it is estimated (Steele 1998) that the diversion of just a small proportion of the annual purchasing power of the libraries of universities in industrialized countries could create and fund on a recurrent basis the necessary academic electronic information/article data banks. A related concern has led to important efforts to provide free access to literature and data, such as the well-known Pub Med system in the medical field, or the Public Library of Science (PLS 2001), a project launched with the support of scholars and editors for on-line public libraries of science to provide unrestricted free access to the integrated and interlinked archives of scientific research.

4. Organizing the information chaos

There are of course other arguments in addition to the financial ones, for undertaking a serious revision of the purpose of publishing. Before the emergence of the internet, in the late 60's, there was a widespread concern among scientists and editors caused by the rapid (incorrectly called exponential) growth of the number of papers and journal titles, and a few attempts were made to bring some order and rationality into the picture by merging certain titles and cancelling others. Then, however, came the apparently limitless possibilities opened up by the modern technologies, which somehow interfered with those good intentions. Against many odds, scientific journals have continued to proliferate - at a much greater pace than the number of scientists-, beyond any planning and without responding to any strategy. Do we know, does anybody know what scientific journals are there for? Have science historians, sociologists of science, information scientists, reached any clear conclusion - or has at least any user group evaluated scientific periodical use systematically over a long period of time?

Under the present financial strain, referred to in the previous section, differences in values among key actors in the scholarly publication system have become more clearly manifest. Stead (1997) refers specifically to the different views held by scientists and librarians - not to speak of commercial publishers - of the role that journals play in the process of scholarly communication: whilst scientific societies' stated top priority is to preserve the archival record or research in their disciplines, a more cynical view is that these journals are principally a service to scientists in their quest for tenure, promotion and grants.

Indeed, a large number of papers are published as a result of the authors' necessity to be seen to be active and productive, as opposed to their genuine desire to communicate new, original and useful research findings. There can be little doubt that the primary stimulus behind this excessive publication remains the pressure to publish experienced by academics. A study based on a survey of over 200 journal editors in the areas of geology, physics, sociology and information science (Anderson 1997) found a high percentage of editors reporting fragmented publications (84%) in the physical sciences and duplicate articles (81%) in the social sciences - despite guidelines for authors normally warning against submittal of excessive articles. Studies as this one show that excessive publication is an issue of current and, apparently, growing concern within the editorial process.

It is largely admitted that huge numbers of articles published in science journals are probably never read by anyone, yet scientific societies seem to care little about this. Consider the following statements of the American Chemical Society:

• Journal publishers have had to increase the number of pages published and add new titles,
• Science journals, embodying the permanent archive of science, must not respond to the vagaries of the marketplace,
• Any loss of subscriptions causes a proportionate increase in the cost of the remaining subscriptions,
• If science journals fail, a vital archival resource may be lost, and
• For scientific society publishers... the mandate remains clear: the archives of science must contain all the quality research within the discipline that is validated for the record by peer-review and publication.

Whatever their reasons for wanting all this research published, scientists and librarians are at cross purposes; the latter have no business buying articles that their students and faculty do not need. One could ask, with Ginsparg (1996), "do publishers add so much essential 'added value' that we should all be willing to pay big bucks for it?"

One could further ask: Must every paper be treated equally? Just as libraries decide which publications are important enough to be held locally and which can be made available through interlibrary loan or electronic access, so would scientific societies and institutions do well to decide what to publish and what to maintain some other way.

In today's liberal economy, the market tailors products to what the research libraries can (hardly) afford. But who says that these products are adequate? Researchers, librarians and scientific societies should work jointly to develop better and more cost-effective systems of scholarly communication, supported by the modern electronic tools, so as to guarantee a satisfactory response to this question. For a large part of the international physics community, in particular, the electronic preprint archives are a partial but important step in this direction. Some researchers, notably in the fields of string theory and condensed matter, argue that they do not read refereed journals any longer for research purposes, since everything they need to know is published in the respective e-print archive (Langer 2000).

Other such preprint servers have been put in place, but it is not yet evident that the same solution - which some qualify as being "subversive" (Harnad 1997) - will work in all scholarly disciplines. Even among physicists, many believe refereeing is a good thing because manuscripts are improved and bad papers are filtered out. Yet the archives have introduced a healthy degree of flexibility and openness that certainly has been lacking in the environment of established journals.

What will further happen to the organization of scientific material as we continue to migrate to the internet? According to Pullinger (1998) the "journal" in the future will be the following, all at once:

• an information store of scientific data,
• an articulated editorial policy,
• a means of accessing information,
• a long-term archive so that scientists of the future will still have access to this information,

and this type of journal will live on because it is at the core of the conduct of science; it gives content to preprint servers, data bases, and more traditional products, depending on what the scientists want. The various goals of scientific publishing, namely: communication of the latest research, an archive of information and data, a record of scientific endeavour, claiming precedence in discovery, and career development, could all be improved, Pullinger argues, if they were not part of a journal. One thing electronic publishing has already begun to do is to unbundle these items. But how far will it go?

Scientists (especially in the exact and natural sciences) do not publish journal articles to communicate with disciplinary colleagues; personal communication, e-mail, e-prints, telephone, fax and meetings are much more important for this purpose. Also, they do not regularly read journal issues, which offer a miscellanea of papers of little direct interest to any single reader. Nevertheless, scholarly literature is a specific kind of cultural good that must be organized and stored for future purposes. It may be that eventually, some fraction of the presently little used material will prove of substantial import - just as has repeatedly happened in the past (Abel 1999).

Meanwhile, as noted earlier, librarians still face a seemingly relentless growth in the number of new and increasingly specialized journals. They are charged exorbitant rates on the grounds that they are read by many people (Chapman and Webster 1993). It is libraries, pressed by their academic colleagues, which provide the bulk of the circulation for journals and assured profit for the publisher.

Recommendations from academic staff, evidence from user studies, bibliometric analyses, subscription costs, are of course among the important factors in assessing the value of scholarly and scientific journals (Anderson 1997); but all these are not sufficient in themselves. A revision of academic organization and evaluation practices and research policies should contribute to devise more rational scholarly communications systems and orient their future development. These are important questions which require strategic consideration and a long-term approach.

As observed in the Introduction, in addition to supplementing existing "real world" activities the new technologies can actually stimulate more of them. Just as internet versions of virtual museums tend to generate more visitors to actual museums, it turns out that the on-line publication of academic articles leads to an increase in subscriptions (Woolgar 2000) and in the number of journal titles. At the same time, the new electronic technologies promise to make organisation more real in that they afford, in principle, the means of tracking and recording interactions and patterns of communications and relationships. However, this does not seem to be happening yet: whilst new electronic information providers are extending their information coverage, librarians face a confusing and bewildering scene of electronic subscriptions offers and packages, and scholars need to spend time on the web to research vast amounts of data and information on-line, offered in a large variety of formats and services. It seems that the electronic technologies have progressed at a much greater pace than the information sciences applied to publishing and communication.

We recall that from early studies of cross-citing amongst journals in different disciplines (Narin et al. 1972), a series of models of the interrelationship of these journals was developed; the separate disciplines appeared to relate to each other in an orderly manner, as had been noted earlier by de Solla Price (1969) and Brooks (1971), who referred to the "full interwoven fabric of science". Within disciplines, the journals were seen to form fully transitive hierarchies with very few relational conflicts. It was further observed (Carpenter and Narin 1973), again using cross-citing, that the journals can be grouped into clusters that are easily identified as sub-disciplinary subject areas. To the extent that they reflect the mosaic of (accepted) scientific knowledge, such citation analyses can be useful tools for the structuring of literature, provided they are used along with other scales to obtain useful or meaningful results (Garfield 1977).

Modern technologies considerably facilitate carrying out more comprehensive informetric analyses, in addition to allowing the development of new powerful tools and methods to organize and relate scientific information and documentation. Hyperlinks are already common within the internet culture, and the packaging of articles within journals may decline in importance in the near future (Brown 1998); the digital world and the internet offer new functionality and packaging possibilities, which hopefully will be used to build a more rational and organized publications system in science.

Again the example set by physics is worth mentioning here: the merger of the refereed APS journals with the e-print archive is expected to produce a powerful and cost-effective system with the help of the modern informatics tools. So good for physics as published in the APS journals; but, as noted before, the various scientific disciplines constitute a full interwoven fabric of science, which is continuously evolving, and it is therefore necessary to build flexible horizontal bridges that connect the large and increasing body of scientific literature in an orderly way. And this means linking intelligently not just among the literature produced today, but also with the literature of the past - including that which is not on the web or which is not published any more - because at this early stage there are already a number of electronic-only journal titles that have ceased to exist (Chartron 1999).

5. Worldwide opportunities for publication

Bill Gates, chief executive officer of by far the largest software manufacturer in the world, predicted in 1995 that "the information highway is going to break down barriers and may promote a world culture, or at least a sharing of cultural activities and values" (Gates 1995). More recently, Teledesic Corp., created by Gates and McCaw in 1990, announced: "On Day One of service - scheduled for 2004 - Teledesic will enable broad band connectivity for businesses, schools and individuals everywhere on the planet. It will accelerate the spread of knowledge throughout the world and facilitate improvements in education, health care and other crucial global issues" (Teledesic 2000). What is the "world culture" to be promoted? Whose knowledge will be spread? Who decides the 'crucial global issues'? The experience with the translation of the Encarta Encyclopaedia for different "markets" forced Bill Gates to admit that "reality can be subjective", that language is not the only relevant parameter because countries and cultures have different interests and interpretations.

In fact, information and communication, and the technologies used, are by their very nature cultural. As noted by Keniston (1998), "the content of software is determined not only by the language, but by deep, underlying, usually implicit and unacknowledged (because thought to be 'natural') assumptions inherent in the software itself. Software carries with it a view of the world, of people, of reality, of time, of the capabilities of users, which may or may not be compatible with any given and social context."

As is well known, the internet is overwhelmingly American-based, English- speaking, and Western-focused. In January 2000, roughly 73 % of the estimated 72.4 million host computers were in the US, 80 % in English-speaking nations and more than 90 % of the internet operated out of Western countries (ISC 2000). As to the on-line language populations, in December 2000 English dominated with 47.6 %, followed by Japanese with 9.6 % and Chinese with 7.6 % (GIS 2000). These distributions are very different from the distribution of languages around the world, which population-wise is dominated by Chinese in the first place, followed by English and further by Spanish in the third place with 6%. Of course, the distribution of languages in science is again very different, English being by far the dominating one. As an illustration of recent trends in the use of languages in science, Table II shows the breakout by language of papers published in the physical sciences.

The uneven use of local languages in science is of course not exclusive of the internet, but is being accentuated by it. More generally, the loss of linguistic and cultural diversity that is occurring among social systems, is exacerbated by technical systems. Among colleagues and peers there has been a gradual acceptance of the use of English as lingua franca and at least for the foreseeable future there is little reason to expect any change in this trend, as illustrated in Table II.

TABLE II. Percentage distribution of document languages, taken from Physics Abstracts.
Data for 1990-1 and 1995-6 (Cetto 1997), for 2000 (Physics Abstracts Online 2000).
The INSPEC database contains an average of 173,000 documents
for the biennium 1990-91, 178,500 for 1995-96, and 225,300 for the year 2000.

However, in addition to international communication among peers, there is communication with students, teachers, professionals, technicians, and with the broader public, which ideally should involve the local languages as they are best suited for a full comprehension. As noted by Einstein (1941), "The mental development of the individual and his way of forming concepts depend to a high degree upon language. This makes us realize to what extent the same language means the same mentality". Further, to be relevant a contribution to the growing body of doctrinal knowledge must be woven into the fabric of existing knowledge, and the writer must use the language conventions of his or her audience (Foucault 1972).

Precisely because countries and cultures have different interests and interpretations, as well as different needs and resources, the internet can become a helpful tool globally only to the extent that it is used worldwide for the dissemination of (locally produced) information. But the alleged 'worldwide' opportunities for publication in science are only very gradually being exploited here and there. There are no surprises: communities having a limited capacity to publish printed journals (besides having a limited access to information), can hardly show a strong capacity to publish electronically.

It is not necessary to recall that a small group of countries comprising 20 % of humanity account for over 90 % of the internationally recognized scientific production (see Table III). These countries are also the more technologically advanced and are therefore at great advantage when it comes to publishing and distributing the results of their scientific production. Scientists who are not connected to the net are excluded automatically from publishing in a growing number of journals. Further, the new ICT (information and communication technologies) products and applications are frequently designed in ignorance of developing countries' realities and fail to address the needs of the most disadvantaged sections of the community (UN 1998). As pointed out by Arunachalam (2000), the gulf in the levels of science and technology between the developed and the developing countries will tend to widen further with the rapid expansion of the internet in the West and the speedy transition to electronic publishing, and this can lead to increased brain drain and dependence on foreign aid of a different kind (knowledge imperialism).

 

Region or country	% publications	publ./GDP*	publ./popul.
Europe	37.5	165	424
CIS	3.7	125	76
Subsaharan Africa	0.7	29	-
North America	36.6	163	717
Latin America	1.8	21	22
Industrialized Asia	10.8	69	113
China	2.0	16	-
India	1.9	47	11
World	100.0	100.0	100.0

TABLE III. Breakdown of contributions to mainstream scientific publications,
by regions or large countries (Cetto 2001). *GDP = gross domestic product.

However, as was stated already in the 1996 Conference (EPS 1996), in some developing countries electronic publishing is not only seen as an opportunity but as a challenge, despite persistent problems of infrastructure, connectivity, resources, etc. Indeed, it has been recognized by scholars in these countries as an interesting and powerful tool to overcome some of the weaknesses of local journal publishing. A few examples are here described as an illustration.

In Brazil, out of a total of 3,630 scholarly journals having an ISSN register, 194 have been selected by their quality for governmental support and 80 of these existed in electronic format already in 1998; interestingly, 33 of these were electronic only (Goncalves da Silva 1998). When consulted about their plans to develop the electronic version of their journal, a majority of the other editors responded affirmatively (without thereby cancelling the paper version) but others gave priority to the updating of the printed version. In fact, Brazil has been a pioneer in Latin America with the development of BIREME, an information system for medical journals. Now it is developing ScieELO-Scientific Electronic Library Online, an electronic virtual library providing full-text access to a selection of Brazilian scholarly journals through indexes and search facilities. The pioneer application is the SciELO Brazil site, Chile is implementing the second application, and further recent advances are reported in a companion article (Packer 2001).

In Mexico, out of a total of 752 scholarly journals, 72 electronic titles have been identified, which seems a comparatively high percentage; however, not all of them provide full text on-line (Reyna 2000). Most of these are in the biomedical sciences (37 %), in contrast to the Brazilian case where the percentage in the social sciences is higher (over 40 %). When studied more carefully, these journals show a series of editorial deficiencies that could readily be solved with the development of an appropriate methodology. In addition, the National University (UNAM) recently launched an on-line full-text digital journal library with an initial number of 5 titles, which it is rapidly increasing.

Further relevant examples of a similar kind are Infomed, the well- established telematic health network of Cuba, with over six years of experience, and Imbiomed, the Mexican index of biomedical journals. This list is not exhaustive, and one can observe on a day-to-day basis that new journal titles and digital libraries continue to appear on the web, throughout the Continent. In fact, with the help of Latindex, an on-line regional information system for scholarly journals published in Iberoamerica and the Caribbean, it is possible to trace new titles as they appear on the internet. The Latindex Directory has compiled basic data on 10,000 serial titles (mostly paper-only) from 30 countries, and the second product, an on-line Catalogue covering a carefully built selection of journals, is under construction.

The most prominent initiatives for African journals are not yet based in Africa. African Journals Online, managed by INASP (www.inasp.org.uk/), aims to promote the awareness and use of African-published journals in the sciences by providing access to tables of contents on the Internet, backed by a document delivery service. It features fifty African-published scholarly journals (excluding South Africa). INASP investigates now the possibility of the programme being maintained in and from Africa. Electronic Publishing Trust for Development is a parallel initiative, that offers tables of contents, abstracts and full text in electronic format of nine African-published journals (issues from 1995-7 onwards) as part of the Bioline Service.

A further interesting initiative in Africa is the Digital-IST (Information in Science and Technology) project, aimed at introducing the personal computer, multimedia and internet in the Algerian publishing system (Bakelli 1999).

Participatory initiatives of this sort, aimed at increasing the publishing capacity and hence the diffusion and visibility of research produced in developing countries, were discussed in the Conference session devoted to the topic (EPS 2001), where it was considered important to foster national and regional partnership and alliances, as well as international support for these initiatives.

6. Science communication with the broader public

As far back as 1922, W. Ogburn, in his theory of cultural lag, observed that changes in material culture, i.e. "the applications of scientific discovery and the material products of technology", occur at a faster rate than do changes in the non-material, adaptive culture (e.g. norms, values, patterns of social organization), thereby causing mal - adjustment in this non-material culture. Many of the recent major changes are technologically driven; we are living in systems that have been significantly changed by the applications of science and technology. Yet our values and norms are still attuned to an earlier era, and our non- material culture has remained in general distant from science and technology. One consequence is our inability to properly face the many ethical issues arising from recent advances in science and technology.

It is therefore time to revise the relation between science, technology and society, so that we are in a condition to face the conflicts arising from this maladjustment and evolve smoothly in a world that is being so rapidly modified by technological circumstances. All important in this necessary revision is an appreciation of diversity as one of the factors by which our evolution takes place.

Local social context is always crucial for the successful implementation and use of new technologies. For the much needed communication between science and the public, it is important that the appropriate social networks, languages and symbols are in place, and that the modern technologies are used to facilitate a rapprochement between the actors of the scientific and technological changes, and other groups that have been mostly passive observers.

As scientists we have a special duty: it is our turn to take steps for this rapprochement. We have an extraordinary new tool to communicate science, and there is a huge public outside, that is growing in numbers and is also becoming more internet-literate; a public that has seen technology being incorporated into its material culture but no science being incorporated into its non-material counterpart. Putting aside some remarkable exceptions, the scientific community has not taken it to itself to communicate with the thousands or millions of people who day by day log into the net and navigate in search for information, or something to learn, or understand, or discover,... or simply to find amusement.

The medical (health) community has gone much farther in this task (with over 10,000 dedicated sites in the USA alone, offering an impressive range of services), and the biomedical community (including bioinformatics) is catching up. But most scientists still maintain as their high - almost unique - priority to communicate with their peers alone. A few scientific societies and associations for the advancement of science are notable exceptions. However, considering that the success of communication in science is strongly context-dependent, a much greater effort is required to communicate and to establish intelligent dialogues with the various sectors of society that demand or are in need of clear, reliable, transparent scientific information for a whole range of reasons.

To illustrate the point, a brief survey was made of the web pages dedicated to science in Spanish, using these two as key words (Cetto 2000). A total of 345,000 pages were found using the altavista search engine - in contrast to 15 million pages on science in English - most of them hosted in Spain, a few Latin American countries, and the USA. Out of the first 100 items in the list, only 29 were selected as meaningful, as the remaining ones did not offer any scientific content.

These 29 had the following features:

• There are a number of commercial portals with just one page on science
• The web pages containing some scientific material are difficult to classify, by their variety of services, contents and scope
• A number of them lack rigorous content or even provide incorrect information
• Many provide information on health only, although they appear as "science and health" pages
• The best pages are produced by scholarly societies or institutions, although there are also a few personal initiatives of good quality
• Services offered include: notes, news, games and experiments, answers to questions, access to journals, book reviews, biographies, directories of institutions, and relevant links
• Commercial sites tend to reproduce existing material, while scholarly ones include a certain amount of original material.

It is evident that whatever is published - and what is not published - on the web today, be it specialized research journals or other kinds of material for communication or popularization purposes, is not in general the result of any scientific information policy or of carefully examined and agreed strategies. Only recently have there been some official expressions of concern in this regard, notably from the Commission of the European Communities, according to which "the development of new and sustained forms of dialogue between researchers and other social operators should be encouraged" (CEC 2000). The new technologies open up a very valuable opportunity to establish different modes of communication not just with our partners but also with other stakeholders in the scientific enterprise. They invite us to revise the approach, the concepts and practices of 'science popularization', 'public understanding of science', etc. and to redesign them so as to establish the much needed real links - as opposed to virtual - between science and society.

Concluding remarks

Let us not forget that the technological innovations of the fifteenth century related to the Gutenberg press, that were essential for the development of scholarly journals, took place long before the first journals appeared. Many features of print communication that are now standard (such as alphabetical ordering, page numbering, peer reviewing and citing of previously published papers) were only gradually introduced along the centuries (Schaffner 1995). Initially, one can expect the electronic journal to replace the paper journal to the extent that it fulfills at least the basic functions of the latter, though in the longer run, when electronic journals have become the natural thing, the new technologies will possibly add new functions, eliminate some of the former ones, or carry them out in intrinsically different ways; eventually, then, the print journal might be displaced by the new "thing" which we now call electronic journal and which probably will be something different from what we now have in mind. In the present transient stage, we have to be patient and cope with a multiplicity of formats, interfaces, ill-defined objects that change or disappear, without really knowing where we are heading - but attempting at the same time to actively influence the outcome.

Certainly the new technologies will continue having an impact on the way scientists communicate amongst themselves and with other people. But, as has also been evident along the discussion in the present text, external conditions that determine the evolution of science, its structure, its institutional arrangements and its relation with society, are influencing the pace and the modalities of the transition to the electronic format. Amongst these conditions perhaps the most salient one is the present scenario of disparities and differences within and between countries, which should be seriously considered in any strategic discussion on the future of electronic publishing and communication in science.

Acknowledgment

The author wishes to thank the staff of the General Library Directorate as well as of the Institute of Physics library, both at UNAM, for helpful information and extensive document delivery. Thanks are given also to the organizers of the Conference for their support and to Graciela Muñoz for her patient revision of the manuscript.

References*

• Abel, R., Publ. Res. Quart., Summer (1999), 37-46.
• Anderson, P., Serials Rev., 23:2 (1997), 45-57.
• Arunachalam, S., in W. Krull, ed., Debates on Issues of our Common Future. Velbrück, Volkswagen Found, 2000.
• Bakelli, Y., INASP Workshop, Oxford, 10-12 May 1999, accessed in June 2000).
• Brooks, H., Science, 174:4004 (1971), 21.
• Brown, D.J., Learned Publishing, 11:3 (1998), 171- 178.
• Carpenter, M. and F. Narin, JASIS, Nov-Dec (1973).
• Cetto, A.M., Boletin. Soc. Mex. Fís., 11:4 (1997), 195-201.
• Cetto, A.M., Proceedings INES International Conference, Stockholm, 2001 (in press).
• Cetto, A.M., La comunicación científica más allá de los pares. CUIB-UNAM, 2000.
• Chapman, L. and F. Webster, Times Higher Ed. Suppl., 26 Febr. (1993), 17.
• Chartron, G., Docum. et biblioth., oct.-déc. (1999), 173.
• CEC 2000: Towards a European research area. COM (2000) 6, Commission of the European Communities, Brussels.
• Create Change accessed in December 2000.
•  Solla Price, D., Measuring the Size of Science. Tel Aviv: Israel Acad. Sci,, 1969.
• Einstein, A., in C. Seelig, ed., Ideas and Opinions by Albert Einstein. NY: Wing Books 1941, 336.
• El País (Spain), 24 December 2000, 24.
• Elsevier Science, 200.
• EPS 1996: Report, working group on developing countries, in D. Shaw and H. Moore, eds.,
  Proceedings of the ICSU Press/UNESCO Conference on Electronic Publishing in Science, Paris 1966, 184.
• EPS 2001: Report, breakout group on initiatives in developing and transitional countries. These proceedings.
• Foucault, M., The Discourse on Language. NY: Pantheon Books, 1972.
• Garfield, E., Citation frequency as a measure of research activity and performance. PA: ISI Press 1977, 406-8.
• Gates, B., The Road Ahead. NY: Viking 1995, 263.
• Ginsparg, P., quoted by G. Taubes, Science, 271 (1996), 767.
• GIS 2000: Global Internet Statistics.
• Goncalves da Silva, L.A., Ciencias de la Información, 29:1 (1998), 61-67.
• Harnad, S., 1997, The paper house of cards (and why it's taking so long to collapse).
• ISA 1998: Internet Statistics in Africa, cited in Lund 1998.
• ISC 2000: Internet Software Consortium: Internet domain survey.
• Keniston, K., Economic and Political Weekly, Mumbai, India, 17 Jan.1998; cited in R. Alan Hedley, Current Science, 79:5 (2000), 592.
• Lancaster, F.W., Information Research (1982), 206.
• Lancaster F.W., in B. El-Hadidy, ed., The infrastructure of an information society. Amsterdam: North-Holland, 1984, 417.
• Langer, J., Physics Today, August (2000), 35.
• Lund, H., Bridging the Gap? Commonwealth Higher Education Management Service, September 1998.
• Marks, P., New Scientist, 21 October (2000), 43.
• Narin, F., M. Carpenter and N.C. Berlt, J. Amer. Soc. Inf. Sci., Sept-Oct (1972), 323.
• Ogburn, J., Against the Grain, 8:6 (Dec. 1996-Jan. 1997), 70.
• Ogburn, W., Social Change with Respect to Culture and Original Nature. NY: B.W. Huebsch, 1922.
• Olden, A., J.Amer. Soc. Inf. Sci., 38:4 (1987), 298.
• Packer, A., these proceedings (2001).
• PLS: Public Library of Science, accessed in January 2001.
• Pullinger, D., accessed in 1998).
• Reyna, F.R., La publicación electrónica en México. DGB-UNAM, 2000.
• Schad, J.G., J. Acad. Librarianship, 23:5 (1997), 406- 408.
• Schaffner, A.C., Lect. Matemáticas, 16 (1995), 247-266, preprinted from Information Techn. and Libraries, Dec. 1994.
• Seminaire Villeurbanne: Une nouvelle donne pour les revues scientifiques? France, November 1997 accessed in January 2001.
• Shulenburger, D., these proceedings (2001).
• Steele, P., Information Management Report, August (1998), 1-4.
• Tagler, J., Ser. Librarian, 38:1,2 (2000), 187-197.
• Teledesic Corp. 2000 accessed in December 2000.
• Tiamiyu, A.A., J. Amer. Soc. Inf. Sci., 40:5 (1989), 325-328.
• UN Commission for Science and Technology, Knowledge Societies Report 1998.
• Vézina, M.-H., and M. Sévigny, Docum. et biblioth., oct.-déc. (1999), 161-172.
• Woolgar, S., Virtual Society? 1 July 2000.
• Wyatt, S., Future of the Internet. Report on the workshop 28 January 2000, accessed in February 2001).

The changing role and form of scientific journals.
Michael A. Keller
University Librarian, Director of Academic Information Resources,
Publisher of HighWire Press, Publisher of Stanford University Press,
Stanford University, Stanford, California 94305-6004, U.S.A.

This paper reports on some of the experiences and results of what might be termed on-line publishing trials underway since 1995. These were begun by a small number of distinguished and venturesome scholarly societies engaged in self-publishing, grown six years later to a community of over 60 such publishers. The catalyst for these experiences as well the provider of Internet publishing services was and still is HighWire Press, a department of the Stanford University Libraries. I ask your patience as all of this recital is based on the HighWire Press experiences.

The innovations covered here are these:

1.	Hyperlinking, including toll free linking
2.	Free back issues, free peer-reviewed manuscripts, and free pre-publication articles
3.	Cross journal searching
4.	Citation and other alerts
5.	Articles appearing only in electronic form
6.	Publishing articles when ready in several variants including publishing issues out of cycle
7.	Publishing long electronic versions of short paper versions of articles (ELPS)
8.	Including supplemental data, moving pictures in electronic editions
9.	Providing referees’ reports and other pre-publication communicationsbetween the author and those involved in the refereeing and editorial processes to readers
10.	Direct communication between readers and authors
11.	Private overnet for distant readers ~25 country nodes
12.	Development of new modes of scholarly communication – knowledge environments

The consequences and implications of these innovations from my perspective will be mentioned. I will mention in passing some of the numerous business models in use by publishers associated with HighWire Press. In addition, I will foreshadow the implementation of a few new features, including:

Concept searching

Aspects of a true digital archive

Retrospective conversion of backsets

Surveying scientists to understand how they use e-journals and print journals

To conclude, I will attempt a synthesis of all of this, fulfilling an aspect of the theme of the title of this contribution.

My topic requires an introduction and an explanatory preface.

By way of introduction, we should all note that Internet publishing of scholarly works has been underway really only for 6 or 7 years. While there were a few, solitary attempts to use the net for publishing prior to the widespread availability of Internet browsers, Internet publishing with the forms and functions now more or less common is less than a decade old. And while one might observe that not enough readers are able to get to Internet editions of scientific journals, it is true that growing numbers of readers are connecting to the Internet daily and growing numbers of them are reading scientific journals as soon as the Internet editions are released rather than awaiting the arrival of paper editions weeks after release. Since the first ICSU conference on Internet publishing in 1996, large numbers of scientific journals have come to the Internet and considerable progress has been made in providing scientists with new and better tools for information retrieval as well as communication of the results of their work. And yet, while much has been accomplished, much is left to be done.

By way of explanation, this paper concerns solely work done on behalf of the scholarly societies publishing Internet editions of journals through the services of HighWire Press. The examples and conclusions drawn are based on the experiences of those associated with HighWire. It should be noted, however, that some of these features are found in other publishers’ sites and with certain aggregators. The 65 publishers receiving services from HighWire each determine their own business models and are responsible for the marketing and sales of their journal subscriptions. For this reason, among others, HighWire is NOT an aggregator; it does not sell access to journals associated with it. HighWire does provide some marketing services, principally market intelligence and analysis, under certain circumstances to some of the societies associated with it, however.

HYPERLINKING & TOLL FREE LINKING

HighWire began hyperlinking in 1995 with links from cited references to Medline. Then, in 1996, at our first publishers’ conference, now a biannual event, there was vigorous argument about the implications for what became known as "toll free linking" from cited references to the digital version of those references. After some debate, some publishers decided to go ahead with the function, provoking a chorus of praise from readers. These positive responses encouraged all the other publishers but one to agree to engage toll free linking. Toll free linking is the feature allowing any reader of any article in any journal in the HighWire environment to read free any cited article in any other journal brought to the Internet through HighWire. Readers might be individual subscribers or they might be authorized members of an institution with a subscription to one or more journals associated with HighWire. Readers might also be anyone on the Internet reading one of the free journals, such as the British Medical Journal, or reading among the 220,000+ articles in the free back issues program. In addition to hyperlinking to full text articles, HighWire publishers also provide free links from cited references to meta-information in Medline, the Web of Science, and a variety of other files, such as Genbank. Presently there are over 2 million hyperlinks made in the HighWire journal set, of which over 500,000 involve free access to full-text articles, either through toll free linking or free back issues. Toll free linking has become a feature required of publishers wishing to receive HighWire’s services.

FREE BACK ISSUES and other free content

Over two-thirds, or about 227,000, of the full-text articles in the journals published on the Internet through HighWire are free to one and all readers. Somewhat after the implementation of toll free linking, in one of the 1997 HighWire Publishers meetings, Associate Editor Robert Simoni, Editor Herbert Tabor, and Executive Director Charles Hancock, all of the Journal of Biological Chemistry, volunteered to provide free access to JBC articles in issues part of volumes more than a year old. This practice has spread to about 24 publishers working with HighWire and as mentioned earlier, now makes over 227,000 articles accessible free. The rationale for the practice is based first on the desire of the scholarly societies to return value to the academic and research communities hosting the scientists responsible for the articles. For the two dozen publishers engaged in this practice, statistics of use of articles read via HighWire provide the additional rationale that such free access does little or no damage to their income, either from subscriptions or sales of access to individual articles. Those statistics show that practically 100% of all articles are read within days of release on the Internet. After 3 months time, only 13% of the articles are read, and after 6 months and beyond only 7% of the articles are read. To put the rationale another way, many HighWire publishers, those in the free back issues program, have structured their budgets so that their costs are covered by subscription and other income received in the first year of publication of their journal issues. Since they received negligible income after the first year, they saw negligible reason to insist on that income. For publishers not in the free back issue program, the reason for not participating is primarily that there is significant income after the first year of publication and that income is necessary in their budgets. It is important to note that several journals are free from the moment of release including the British Medical Journal, the Journal of Clinical Investigation, Stem Cells, The Oncologist, and Advances in Physiology Education. And numerous other journals have loss leader periods or free trial periods resulting in significant free access to research articles as well as other content. This is the second largest collection of on-line articles presently; honors for first place goes to the NASA Astrophysics Data System with over 300,000 free full-text articles.

There is other free content, of course. Clinical Medicine NetPrints, a project of the British Medical Journal and HighWire’s parent organization, the Stanford University Libraries, provides about 35 preprints. And all of the articles represented through HighWire provide free access to table of contents and abstracts. Presently there are over 912,000 articles represented by abstracts on HighWire’s servers, with almost 400,000 of them presenting full-text in addition to the abstracts. Some publishers, the Journal of Biological Chemistry principal among them, provide free access to peer-reviewed, but not yet copy edited articles in a program called "Papers in Press." And Science Magazine makes certain articles free and occasionally entire issues free, as it did last week, out of cycle, with the issue on human genomics, resulting in a couple of days of six times the number of hits Science usually gets, that is over 6 million hits per day for a couple of days.

It is worth noting that virtually all of the publishers receiving HighWire’s services have special pricing for nationwide subscriptions and very favorable pricing for subscribers in the developing nations. These publishers want to make use of their Internet editions to support therapeutic practices in distant and difficult places as well as connect to colleague scientists and clinicians far from what formerly was regarded as the primary clientele for their journals.

CROSS JOURNAL SEARCHINGReaders can choose to search for: citations; keywords in titles, abstracts, or articles; authors; and issues or dates in one journal, in several journals of their choice, and across all journals. Results are returned in order of relevance. Numerous special features broaden or make narrower a search and Boolean operators are possible. However, we note that searching accounts for only about 5% of all interactions with the journals brought to the Internet with HighWire’s assistance. The vast majority of searches are simple ones involving a single author or a single keyword or phrase. We suspect that most of rest of that 5% is searches performed by librarians and other information professionals.

CITATION and OTHER ALERTS

One of the reasons that a higher rate of searching by individuals, by humans, is not apparent maybe the alerting features. Presently, we make the same parameters for searching available for alerting. Publishers offer alerting services to member-subscribers, not to readers gaining access by institutional subscriptions, thus reinforcing the ties that bind members to their scientific societies. We now have over 300,000 individuals specifying over 1.1 million alerts and over 100,000 new alerts are set up each month. The servers conduct the alerts weekly and send e-mail messages with "hot" links to the hits found on the basis of the requested alerts. Naturally, one of the most popular alerts is the citation alert. One always likes to know who is citing one’s own or one’s competitors’ papers. The alert wizard allows a reader to select one, some, or all of the journals in HighWire to be subjected to the requested search. The wizard then performs the searches just after the new issues of each selected title appear. It is this characteristic which might reduce the number of what might be called spontaneous searches.

ARTICLES APPEARING ONLY IN ELECTRONIC FORM

The American Pediatric Society decided early on to distinguish its Internet edition from the paper edition of its journal, PEDIATRICS, by offering some authors the opportunity to get their articles published swiftly after submission, but in the Internet edition only. The attraction to the authors is obvious. The publisher of PEDIATRICS, however, was and is experimenting with driving readers to its Internet edition. In the print edition of PEDIATRICS, the electronic-only articles are cited in the tables of contents and abstracts are provided along with a special pagination scheme to make clear when citing the article that it was an electronic-only one. After some initial concern on the part of authors, the electronic-only form of publication seems to appeal to a steady minority of authors seeking to place their articles in PEDIATRICS. The society believes the experiment to have been a success and is continuing the program. In a parallel experiment, the JOURNAL OF BIOLOGICAL CHEMISTRY offered their Internet edition unbundled from the print edition, a move that was a cause-celebre among librarians for a while because it was felt that this business model was meant to extract nearly twice the print subscription from them. The JBC was and is really trying to drive subscribers to the Internet edition. The BRITISH MEDICAL JOURNAL has another variant practice in which some news articles, labeled "news extras" never appear in the printed editions.

Publishing articles when ready in several variants including publishing issues out of cycle

The practice of publishing out of cycle was already mentioned earlier in the free back issues section. SCIENCE MAGAZINE is famous for this practice, first engaging it when the article by Richard Zare et al. was released in the summer of 1996 as the news embargo about the hypothesis that petrified or crystallized bacteria were found in a rock known to have originated on Mars was broken. The most recent example of this practice occurred last week with the release of the human genome issue by SCIENCE. SCIENCE is not engaged ordinarily in the free back issue program, preferring to release the occasional article or issue when the contents are of such significance to the general public to warrant distinction. One might think of the practice as a kind of attractant to readers, tantalizing them with the implications of other articles of interest behind the access control.

Several publishers working with HighWire publish articles in several variants under the terms of what the JOURNAL OF BIOLOGICAL CHEMISTRY calls its Papers in Press program. Successive versions of papers submitted and accepted after peer review appears, with the first and final version showing in the Internet issue, but only the final version in the paper edition. The PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES calls its similar program "Early Editions", offering authors more speedy publication of articles than waiting for simultaneous appearance of print and Internet editions. Eventually, the print edition includes the completely copy edited versions, of course.

Publishing long electronic versions of short paper versions of articles (ELPS)

In the FASEB JOURNAL, as well as others, short, summary versions of articles are published in the paper edition, but longer versions of the articles, more like the article as originally submitted by the author, are published in the Internet edition. Publishers are taking advantage of the possibility of expansive treatment without penalty of paper and postage costs through the Internet editions. Authors are gratified to see the longer version of their papers presented in the Internet editions, yet are happy as well to see their article in actual print. This practice is a sort of halfway house between electronic-only issues and print issues for conservative tenure committees. The BRITISH MEDICAL JOURNAL has published abridged news summaries in the printed edition, with more expansive versions in the Internet edition.

Including supplemental data, moving pictures in electronic editions

This practice began early in the Internet editions of SCIENCE MAGAZINE. Data supplements, moving pictures, and extra links to carefully selected "related sites" appeared with some articles. There is a convenient link to such articles on the SCIENCE MAGAZINE site. Many other journals, including the BRITISH MEDICAL JOURNAL and MOLECULAR BIOLOGY OF THE CELL include supplemental data. In all cases, the tables of contents of journals offering supplemental data indicate the fact of such extensions to the article. SCIENCE, in addition, maintains an index to such extensions under a sub-site labeled "beyond the printed page". Very often, the supplementary data are video clips or additional figures. Occasionally, the supplemental data is just that, data in a spreadsheet or other database presentation. This is, of course, another example of publishers taking advantage of the possibilities of the Internet to increase the amount of information flowing from authors through editors to readers. The limitations of print, among them the physical problems of pages and gatherings as well as the cost of color images, are overcome entirely by the capabilities of the Internet. The cost and production complexities prohibited the inclusion of moving pictures or sound in a CD-ROM supplement to a print edition of a journal, especially one with a monthly or weekly schedule; this sort of extra might have appeared as a kind of trick or gimmick in the days before easy Internet publishing, but now is rare but for the occasional appearance with a book, a one-off event with a long time horizon for accomplishment.

Providing referees’ reports and other pre-publication communications between the author and those involved in the refereeing and editorial processes to readers

The debate regarding who provides the essential element of refereeing and when in the publication of scholarly articles referred to in an earlier paper has fostered more theoretical and political commentary than actual experimentation in new schemes built according to the assertion that refereeing should be "liberated" from the publishers. The BRITISH MEDICAL JOURNAL has begun to provide referees’ reports as an experiment in order to open the process of accepting a paper for publication to the readership. The BMJ includes in practically every issue now the pre-publication history of at least a few articles. This practice along with the "rapid responses" feature in the Internet edition of the BMJ makes the "afterlife" of an article a more vivid one, making the reception of an article or editorial by readers willing to send a letter or comment to the editors. The BMJ is definitely showing the way for a more open system not only of peer review, but also of the involvement of readers in the publications themselves. The practice of documenting additions and corrections in the Internet edition rather than silently making emendations is another aspect of making the electronic editions more transparent to the readership and thus more credible to them as well.

Nota Bene:the previous nine features taken together by the 65 publishers receiving services from HighWire Press have prompted them to conclude in last October’s conference that their Internet editions are the editions of record. Certainly many of them had come to this conclusion some time earlier and the most adventurous of them, the publishers of the JOURNAL OF BIOLOGICAL CHEMISTRY, SCIENCE, BMJ, and the like, from the beginning were ready to a considerable extent for this change of belief and perception. The ramifications of this switch are numerous even if not all of them are yet revealed. One matter is highlighted, however, that of the digital archive, and our practices to accomplish the task of saving harmless the mother files comprising the editions of record. The question of access to the Internet issues as decisions are made to cancel current subscriptions arises as well. Several meanings of the term "archive" flourish regarding Internet editions and sites. Many commentators and service providers, including aggregators, refer to archives when really what is meant are sites with lots of content available for public access, really sorts of libraries and anthologies publicly accessible, whether for free or for fees. At the other end of the spectrum of meaning is the traditional one referring to an archive as a secured copy, most often unique, of a cultural record, one that is made available to readers and investigators under controlled circumstances and frequently through copies so that the originals are not damaged by use. It is this latter definition that we at HighWire and the Stanford University Libraries prefer and about which more later in this presentation.

Direct communication between readers and authors

In each article published in Internet editions at least one author has been designated the corresponding author, responsible for receiving and answering e-mail messages generated by readers by way of the "pop-up" e-mail feature common on all browsers. We have not yet extracted indicators from our logs to know the extent of use of this feature. However, traffic from the use of the feedback feature on all articles brought to the Internet via HighWire’s services is so extensive that several staff are responsible for receiving it, sifting the editorial questions from the editorial ones, forwarding the latter to editors and less frequently to authors, and answering or otherwise addressing the former. We may know more about this phenomenon as we perform the survey to be described later in this paper, but the indicators are that this feature is well enough used.

Private overnet for distant readers ~25 country nodesDue to HighWire’s posture as a service provide very responsive to its customers, scholarly societies as publishers, it was apparent as new customers were added that numerous distinctly different journals with distinctly differing business plans and access controls would result. The number and complexity of the business plans meant as well that the data bases providing access for institutional and individual subscribers would be numerous. Given as well that we had from the beginning several of the most cited journals with frequent new issues and large additions in those new additions to the databases, we knew from the beginning of HighWire Press that we had numerous unusual problems, principal among them was the need to provide for reliable and swift access without having mirror sites of HighWire’s constantly changing databases of content and subscribers. For two years we sought solutions to our problem from suppliers of private networks to corporations. None appeared to be either interested or capable of providing the service we needed. Then a start-up company, a spin-off of Cisco Systems called Digital Island, appeared on our horizon and an agreement providing for a private overnet with expandable bandwidth and nodes in about 25 countries now was written. Essentially, readers in countries featuring Digital Island nodes, mainly the E.C., industrialized east Asia, Australia, New Zealand, South Africa, Brazil, Mexico, and Russia, link to Digital Island’s nodes and then their interactions occur on a low latency, expandable bandwidth network with a direct fiber link to HighWire’s servers at Stanford. This has simplified operations, improved access for distant readers, and allowed HighWire to continue its elaborate technical architecture in service of all the features and functions publishers have specified. Others are involved with Digital Island and similar services, presumable as in the case of HighWire investing less in private overnets than would be the cost of mirror sites.

Development of new modes of scholarly communication – knowledge environments

Editors of SCIENCE MAGAZINE, the publisher of Island Press, a not for profit environmental house, and members of HighWire Press recognized several factors together pointing to a new direction to take in scholarly communication. The factors are the heavy flow of highly cited articles from publishers to readers through HighWire’s servers, the hyperlinking and searching features previously described, and the burgeoning sense of communities coming together at several levels – as publisher-customers of HighWire, as readers of multiple journals associated with HighWire with common themes or subjects, and as scientists making use of the features of the Internet editions to improve their information retrieval as well as their individual communication of research results via the journals. Outside of HighWire, the NSF was funding the first round of research in their digital libraries initiative and numerous other activities began to define what was meant by the term "digital libraries." A new sort of digital library, serving a narrowly defined topic, but for a well-defined global readership, with both information resources as well as information services took shape in several planning sessions. Eventually, the Pew Charitable Trust funded the development costs of the first knowledge environment, one devoted to cellular signal transduction. Pew’s interests were in maintaining and extending communities supported by their philanthropy through the use of communication embedded in the sort of narrowly defined knowledge environment we proposed to them. After a lengthy gestation period in which the American Association for the Advancement of Sciences took the editorial lead and HighWire developed the technical facilities, the STKE, signal transduction knowledge environment, was launched. Publishers of 47 journals permitted free access to articles in their journals on the subject though a feature known as the virtual journal. Essentially, the virtual journal on signal transduction is constructed by a smart software agent which recognizes terms relevant to signal transduction even if the term is never used in an article. Readers of the virtual journal call up the article from the original journal in the branded site of the original publication. Additional features of knowledge environments are: unique navigation devices, such as the graphical pathways in STKE; new content provided by the editorial group including review and perspective articles and protocols for performing experiments; threaded conversations on topics in the knowledge environment, some monitored and edited and others not; biographical information; news of events and available positions in the discipline; and personalization features such as citation and keyword alerting. After a free period of about 15 months, the STKE was launched a few weeks ago as a fee-based site; readers need not have subscriptions to any of the titles contributing to the virtual journal, but would be dependent upon pay-per-view access, if available, to go beyond abstracts to full-texts. So far several hundred subscribers have paid to continue their access to the knowledge environment.

Another knowledge environment, one entitled "Bone Key or Osteovision and devoted to bone, minerals, cartilage, and teeth for clinicians and researchers, has been launched by the International Bone and Mineral Society and supported by HighWire Press. It includes a selection of Internet sites of interest to patients along with most of the features of the STKE.

Others are in development, including one on cancer and another on the sciences of aging. Still more are in various planning and proposal stages, including one with the Island Press on aquaculture.

The point of the knowledge environments is the movement from the traditional characteristics of printed journals and even from at least the periodicity of the Internet editions of those journals. As we see these in the life and clinical sciences launch, we may be witnessing one of the many steps toward a vastly different universe of scholarly communication.

Some Developments Signaling Progress

Concept Searching

For some time, publishers working with HighWire have requested an addition to the searching feature in the form of concept or subject searching. Some journals, such as the BRITISH MEDICAL JOURNAL and SCIENCE magazine provide a sort of subject classification of articles they publish. This pre-selection of broad topics allows a kind of broad subject search. Some journals use this subject classification to describe collections of articles, encouraging readers to browse in such collections. However, true subject searching across multiple journals is made difficult by the numerous differing classifications and made less than useful by the necessary breadth of the subjects assigned by any individual journal’s editors. We have been investigating and are still investigating the use of semantic search engines to provide true concept searching on terms provided by readers as well as by using a sentence, paragraph, page, or an entire article to match in meaning to all other articles in a journal, in selected journals, or in all journals in the HighWire environment. We are on the cusp of adopting a semantic search engine and within a year of providing this sort of searching capacity. In a way, the smart agent we use to extract articles from multiple journals for the virtual journal in each knowledge environment is such a concept search engine, but it is too rigid and perversely not robust enough to be used by several hundred thousand readers. So, we are examining commercially developed concept search engines.

Aspects of the true digital archive

A single group, institution, or publisher working alone will not answer the question of what is a digital archive. Nor will the answer have a single clause or a unitary solution. Assertions of providing something called a digital archive are now being made and the spectrum of meaning of that term are as already described with the addition of commercial overtones in order to satisfy the desires of institutional subscribers, librarians, anxious to be sure of the long term survival of what most of us know to be the editions of record, the digital editions. Standards and principles are being written and enunciated. Pronouncements are being made. Foundations are supporting planning grants and new relationships are forming to satisfy the need. The several possibilities for assuring the survival of digital content are: migration of digital information from data format to data format ad infinitum; the emulation of prior states of databases, applications, and the data sets themselves; the virtual encapsulation of datasets, applications, and operating systems by some means not yet known making possible the recapitulation of a digital environment to reproduce the digital edition. Short-term digital preservation efforts now involve tape libraries and tape archives, the former on-site and immediately at hand and the latter at some distance and subject to different calamities than the local copies. Some retrospectively inclined prophets of the digital archive have suggested microfilm and paper copies of the digital editions. And there is a nano-laser technology in development offering the prospect of either analog copies of pages in micron dimensions or digital copies on non-oxidizing substrates, looking remarkably like CD-ROMs, but without the layers of plastic and foil. Many of us are concerned that none of the possibilities offer the prospect of saving all of the information at the end of links remote from the publisher’s products, e.g. hyperlinks to other Internet sites outside of any publisher’s digital community or service provider’s environment.

HighWire and Stanford are working on several fronts at once, briefly described here.

First, like all responsible Internet service providers, we maintain an on-site library of tapes of every source and production file as well as remote archives of the same.

Second, HighWire has migrated the entire database six times. Each time this was done to enable new features, but each time, the migration had the effect of moving the production files from one data format to another.

Third, with support from Sun Microsystems, the National Science Foundation, and recently the Andrew W. Mellon Foundation, we are developing an open source software package dubbed LOCKSS for "lots of copies keep stuff safe." This network caching software will permit institutions first to capture content authorized to them and second to compare as well as assemble the most complete version of the mother files when the mother files disappear for more than a defined and short period of time. When the mother files disappear, the various LOCKSS sites poll one another and mutually and redundantly assemble the most up-to-date and complete version of the disappeared site, making each LOCKSS cache visible to the local institution. Most of the HighWire publishers are in full support of this program. Any data format transmissible by the http protocol is potentially visible to LOCKSS. No, it is not like Napster.

Fourth, the Stanford University Libraries are developing, first as a plan and soon, we hope, as a prototype a true digital archive, known as The Dark Cave. We intend to operate a 100Tb digital archive in the prototype and a 1Pb digital archive in its first production version. It will use automatic metadata capture software, self-organize files, and permit access by intermediary copies of files on intermediary services only under terms negotiated with publishers until such times as the copyrights and other intellectual property regimes’ rules permit other sorts of access. The files will be accessible only to digital archivists in the Dark Cave and, with permission, to others as copies of the stored files. We are in agreement with a small number of North American and European libraries of comparable or larger size than Stanford’s to share results of similar projects and in principle as well as subject to permission of the owners of the files to make redundant copies of files ingested at each site. We believe that a small number of such huge digital archives, some in national libraries, some in private research institutions will evolve. As a last word on this topic, one should note that the computational life sciences with the necessity of generating huge files of data from living beings, especially humans, as well as the desire to perform feats of data mining across libraries of such files will find such masses of archived data very attractive in the search for new research findings and new therapies addressing issues at the cellular and even molecular level of our lives. It is vitally important to the future not just of our species, but of all species that we preserve the states of our knowledge reflected in such files for reference in the future as "progress" or "disaster" ensue.

Retrospective conversion of paper backsets to digital forms

Following the lead of the JSTOR, many publishers are converting their paper backsets to digital form, most often PDFs. Some are converting to SGML and even XML formats. Even with PDFs, hyperlinking forward and backward is possible, though not yet in widespread use. We will see much progress made in reviving use of forgotten, but still useful articles in the coming years. Remembering that articles more than six months old in the HighWire environment still enjoy a steady, but low percentage of use, we should expect to see similar results of older articles, particularly in the more "timeless" disciplines of the hard sciences. Some HighWire publishers are contemplating the publication of digital anthologies of "classic" articles, frequently mentioned or even mythologized, but rarely read, because they exist only in paper editions.

Surveying scientists to understand how they use e-journals and print journals

During these early years of Internet publishing, there have been numerous of the usual sorts of surveys conducted. That is to say, opinion surveys have predominated, ones collecting impressions after the fact of reading on-line. And most of those surveyed are librarians, institutional agents for masses of readers, rather than the scientists and other scholars themselves. Little or nothing has been reported on the extraction and analysis of data from logs of actual use of Internet editions of STM journals. With support from the Andrew W. Mellon Foundation, HighWire has begun a survey which will undertake both sorts of collection and analysis, the impressionistic, after-the-fact survey of readers, and the extraction and mining of data logs. To be sure, we have provided our publisher customers and in some cases institutions as customers of the publishers with data concerning the use of articles and issues. However, we have not had the time to prepare the macro-programming tools to deeply analyze actual use of hyperlinking across its numerous possibilities, use of e-mail and feedback by whole populations of readers, and so forth. This project will take 24 months to run. Already we have hired key staff and contracted with a professional survey research firm to conduct surveys on the web and interviews of individual scientists. We will report detailed results to HighWire publishers and aggregated results to the world at large when the survey is complete.

The Synthesis and conclusion In a few years, we have leapt forward from a mode of scholarly communication that was both passive and slow. The dependence of readers and publishers on postal services, even ones involving air delivery is gone. In addition, the manner in which scientists and librarians retrieved information based on the hard work of indexers and abstractors extracting slowly and codifying relentlessly the mass of printed information has changed too. We still need indexing and abstracting services, because we still have competitive boundaries between and among publishers preventing the fullest use of some of the new techniques. And despite claims to the contrary already made and soon to be made, the new modes of scholarly communication are considerably less costly than the old ones.

I propose to you that we only dimly perceive the role the new forms and features of this first stage of Internet publishing has had. We certainly see increased speed of delivery of articles and news to the scientific community, but we have not yet measured the results on science of the improvements. And because the rate of change in the new Internet environment of publishing has increased so much in the past few years, we may have run ahead of our ability to optimize systems, features, and forms to the humans who are simultaneously creators and consumers of the information conveyed by way of the net. Does the downturn of the markets investing in Internet technologies and companies signal a decreasing rate of change in scientific communication? No one knows.

What we have now is a new, if transitory form, of the old, made better by timely Internet delivery, but only to those privileged to have easy access to fast networks and good desk-top computers. We have more active modes of communicating reports of scientific research, with plenty of features promoting accurate and speedy retrieval of information now based on the vocabulary employed, but soon on the concepts implied as well as expressed. Readers can communicate with authors and editors more readily and sometimes their reception of a particular article or editorial will appear adjacent to the object of the commentary. Where formerly, all readers were subjected to the same crushing demand to comprehend someone else’s choices for subject and keyword indexing, now at least keyword and citation alerting can be personalized and automated.

We are beginning to see new modes of scientific communication and scientific information services. Some are based on global communities of need and interest, providing tailored information services based on commonalities, but still supporting differences within the narrow confines of a knowledge environment. We see as well the possibilities for broadening the grasp of individuals and groups of scientists through concept searching, prospectively making more interdisciplinary connections than ever before.

Thanks to the ever-increasing numbers of people connected to the Internet and the generous business practices of some, the badly tilted playing fields of science across the have / have-not boundaries are slowly righting, if not yet level. At least some measurable improvement of communication is visible as the disenfranchised by cost and by semi-catatonic modes of physical delivery is replaced by virtual delivery in much less time.

Independent of the adoption of Internet technologies for scientific communication, but seeking to take advantage of these new possibilities, some of those direly affected by the crisis in STM serial publishing are calling for the disintermediation of the chain of actors and events in scholarly communication. Others are opening the process of refereeing and encouraging readers to issue communiqués for publication as a partial antidote to the secrecy of refereeing in the ancien regime. To date and despite the calls to pull down the walls of the refereeing process, little change has been noticed, presumably because validation of results and the presumed privacy of the inside critics of submitted manuscripts are still widely believed to be crucial to the process of publication, even in the Internet age.

And, with some effort, we will understand better what use has actually been made of these transitory forms of Internet editions. The promises as well as the unknown results of having true digital archives may provoke even more scientific developments, some perhaps tragic or leading to tragic comprehension of past decisions and their effects on the future. Better science may result.