HEATHER DALTERIO was manager of publications at the AAS. She is now with Elsevier Science in New York. h.dalterio@smtphost.elsevier.com
It is, of course, impossible to predict the form electronic publishing will eventually take. But certain general principles are emerging. The electronic future is approaching at breakneck speed, and many forward-looking publishers are preparing to make use of the new technology.
Many organizations are discovering the potential of the World Wide Web for the electronic dissemination of information. For years, computer networks were used mainly for electronic mail and the transfer of files. As recently as 1990 the Web was not yet in existence,[1] and e-mail was still somewhat difficult to use. Even the most optimistic experts could hardly foresee how rapidly the technology and the software would advance. Few people envisaged that within four years all the pieces necessary for a full-featured electronic publishing system would be in place.
Then, in 1992, Mosaic burst onto the network scene. Here was an easy-to-use, intuitive software tool for browsing offerings posted on the World Wide Web. Mosaic was developed by a team at the University of Illinois's National Center for Supercomputing Applications. For the first time, a workable tool was available that could deliver text and graphics in the same document over the Internet. And it was free of charge to the user. Finally we had a tool for displaying graphics and the special symbols and complicated equations needed for publications in the physical sciences.
With the necessary technology available, though in its infancy, the physics and astronomy communities are asking why the major publishers have not more rapidly adopted a system of electronic publishing. This perceived tardiness creates a tension between the scientists, who now have the capability of publishing their papers on the Internet, and the traditional publishers, who cannot move so rapidly to adopt new technology.
The publishers are moving so slowly for several reasons, among them the difficulty of pursuing innovation within established operations, the sheer size of the present publishing enterprise and the need to develop a well-planned system for distributing, exchanging and retrieving electronic information before launching a new project. In combination, these factors make it difficult for established publishers to be the pioneers in adopting new technologies. Fortunately, however, significant progress is being made.
Much of the tension between authors and publishers over electronic publishing stems from a fragmented understanding of what it takes to have a complete system. There is not even a general agreement as to what "electronic publishing" means. The various groups involved tend to focus on those aspects with which they are familiar. (See the list) Authors concentrate on electronic manuscript preparation and submission. Readers focus on retrieving information from the Internet. Librarians focus on the delivery of information to users, but they often overlook electronic archiving. Publishers concern themselves with the handling of electronic manuscripts, copyediting, formatting, typography and the production of versions suitable for delivery to the end users. They also worry about collecting the revenues needed to keep their operation financially viable.
Electronic publishing of scientific journals embodies all of these aspects. In the end it creates new capabilities that extend far beyond what the paper journals can provide. It is precisely these new capabilities that will make electronic publishing such a powerful tool for scientists.
Electronic tools provide search and retrieval abilities beyond anything that can be done with paper, but only if the system is set up to take advantage of the interlinked access capabilities of the Internet. Many different kinds of information can be effectively distributed electronically. Astronomer Andre Heck at the Strasbourg Observatory, which houses a worldwide repository of astronomical data, has encouraged the scientific community to go beyond traditional publishing and provide a wide variety of information in various forms.[2] As both a publishing astronomer and a database provider, Heck stresses the need to maintain the quality of information on the World Wide Web. It is not enough to prepare the input information with great care, he insists. It is also essential to maintain files and archives in a usable state once you have them.
The central feature of any electronic information-distribution system is the database of information, coded and organized in such a way that the desired information can be retrieved in a useful form. From our early experience at the American Astronomical Society, we have found that readers want papers and other data in various forms: Output to a screen is important for browsing and searching but is not well adapted for extensive reading; paper copies of selected information will definitely be required. Therefore the article has to be transmitted electronically to the reader's local printer. Tabular material will be needed in formats suitable both for reading and for manipulation by the reader's computer. Finally, it is clear from reader feedback that the community is not yet ready to abandon altogether the traditional typeset, printed and bound journal. For the most part, publishers will have to ensure that they can produce the familiar, printed volumes from the central database and that all versions a re identical.
The tools available for self-publishing on the World Wide Web are seductive. It is now possible for an individual author to add codes to a manuscript, convert equations and special characters into graphic images and display a readable copy of the paper on the Web. The principal tool that makes this possible is the Hypertext Markup Language. But it is important to understand why individual, hand-coded papers will not serve the needs of the scientific community as an effective publishing mechanism. Readers must be able to search for information and then have it available as part of a structured, searchable database. We know from reader feedback that electronic manuscripts have to be available in multiple forms. Readers want more than just displays on the screen. They also need paper copies printed locally, some stand-alone form such as CD-ROMs and, of course, the traditional typeset journal issues on the shelf.
All of this requires a significant amount of logical markup coding to identify various pieces of the manuscript. Hypertext Markup Coding, including the new version currently under development, is not sufficiently structured or versitile. To be effective, the markup has to conform strictly to a uniform set of standards; most authors are not willing to take the necessary trouble.
Even the best papers are of no use if they cannot be found when you need them. The research literature is of most value when you can search and recover articles of interest from a large fraction of the entire collection. Strict adherence to markup standards is absolutely necessary if such searches are to work. Heck emphasizes that precision and attention to details are extremely important. Our experience at the AAS confirms this. One cannot effectively search among single papers stored in hundreds of locations across the whole net. Being able to put their papers on the network may gratify authors. But the intended users of the information must first locate and retrieve a paper and, second, be assured of its quality and stability. A database maintained by authors gives no assurance that the version of a paper you see today is the same as the one you were reading last month. It will take good planning to create an electronic publishing system that will really serve the research community's needs.
One of the consequences of the traditional publishers' slowness to create an electronic system has been the increasing private electronic distribution of preprints and data. And scientific societies are beginning to provide fledgling electronic services. Particularly important has been the use of electronic delivery to provide rapid notification of new results through mailing lists and electronic preprint services.
Astronomers have been distributing information about early-type stars through the Be Star Newsletter.[3] This early-warning scheme for alerting the community to all new, relevant papers has become very popular within the small circle of interested astronomers. The Be Star Newsletter is hosted on a computer at Georgia State University and maintained by volunteers. The resources necessary to continue such a small-scale operation are modest. But one cannot maintain a robust system for long with volunteer editors and contributed computer and network resources. Nevertheless, such experiments are very useful in demonstrating new and different ways in which the network can serve the working scientist.
On a larger scale, most physicists and astronomers are by now aware of Paul Ginsparg's Los Alamos[4] preprint server, and the SLAC[5] preprint server at Stanford University. Both serve a useful purpose, particularly for the high-energy theory community. They may be a model of how to serve users by providing rapid distribution of preliminary information. The SLAC system in particular has all the critical components of a full electronic publishing system except peer review and copyediting. It now has a large collection of documents and a search capability for abstracts, keywords, authors and so forth. Each paper's reference list provides links to other papers in the database. Both the Los Alamos and SLAC systems provide delivery of source documents.
The executive office of the American Astronomical Society recognized the possibilities of electronic dissemination almost three years ago. It embarked upon a plan to use the emerging Internet technology and software to distribute information to the astronomical community.[6] In developing the AAS plan, we focused on providing high-quality publications at low cost to both individual readers and libraries. At the same time we experimented with the form and function of the electronic journal. Recognizing that the technology is developing rapidly, we decided early on that we had to develop expertise within our organization so we could keep abreast of new developments. Therefore we have taken small steps wherever possible, using each one to enhance our capabilities, to gain experience and to obtain feedback from the users.
At the AAS we have chosen to use open standards and freely available software. That allows us a large measure of independence from commercial products and large proprietary systems. Using this strategy we have created a comprehensive program of electronic publications and services. Since 1992, for example, the AAS has been providing electronic receipt, manipulation and delivery of meeting abstracts.[7] This system has all the functions of a full-fledged electronic publishing system, including the collection of abstract fees.
Our experience with the meeting abstracts has been invaluable in helping us design our system for bringing the refereed journals on-line. Since 1993 the AAS has published an electronic version of its monthly register of jobs in astronomy, followed a year later by a monthly electronic newsletter for members. Since last September the AAS has been publishing an electronic version
of Astrophysical Journal Letters on the World Wide Web.[8]The AAS continues to regard its efforts as an exploration. We remain keenly aware of the experimental nature of what we are trying to accomplish. Only through continual modification in response to user reactions will we have a system that is effective, useful and affordable.
While many of the principles espoused by the AAS are shared by the other member societies of the American Institute of Physics, both AIP and the American Physical Society have taken a slightly different tack by going to an outside vendor for their delivery systems. That has allowed them to provide electronic versions of their letters journals sooner than if they had gone through the requisite development in-house. But there are negative aspects to using an outside vendor: Among other things, one has to conform to the standards of the vendor's proprietary software and one worries about the speed with which the vendor can change the system in response to user needs. Another drawback is the possibility of being locked into the use of only one vendor. Both APS and AIP have taken steps to ensure that the cost to the users will be minimized and that they can discard the vendor's proprietary system if new developments warrant such a step. They recognize the importance of maintaining flexibility in anticipation of new developments.
The APS and AIP electronic journals differ from ours in their look and feel on the screen. The Astrophysical Journal has evolved to provide special tools and navigation buttons for making use of an electronic journal. One feature, for instance, is the ability to go directly to a page of figures and if one of them looks interesting, jump directly to the text where that figure is described. The reader can also start with the references and then be catapulted back into the place in the text where a particular reference was cited. This popular feature was created in response to feedback from readers. The ability to make such modifications rapidly has been characteristic of the Astronomical Society's development effort.
The question of ownership of material posted on the Internet is becoming more and more important. Users of the original computer networks developed a strong tradition of openly sharing material on the network. Much high-quality software is still being made freely available on the Internet. In fact several companies are making versions of their World Wide Web browsing tools available free of charge. Many users are taking advantage of all this free software. Most software carries a statement that describes the author's intention as to how freely a program may be distributed. The same cannot be said for most text documents.
In the US the copyright law of 1976 makes it clear that all documents are automatically copyrighted as soon as they are written, with the rights belonging to the author (or, if they were written for hire, the author's employer). Most of us understand that publishing an article on paper implies a copyright, whether or not a copyright statement is printed on the page. But such general understanding does not extend to fledgling electronic publications. Most documents on the Internet and the Web provide no clear indication as to how freely they may be copied and distributed. Authors who want to grant permission for free redistribution should add clear statements of intent to the documents.
Many scientists are not aware that posting a document on the Internet, where it becomes available to everyone, is equivalent to publication of the document, and that posting someone else's document without express permission violates the author's copyright. Anyone who develops a World Wide Web home page is, in effect, a publisher and is thus legally responsible for obtaining the permission of the copyright owner of everything published on that page. This is true not only for manuscripts but also for e-mail messages, which also cannot legally be redistributed without the author's permission.
One of the least discussed consequences of the move to electronic information dissemination is the major effect that it will have on organizations. To incorporate the new technology effectively, organizations will have to adopt a completely new approach to working. This has become clear from our experience in changing from paper-based to computer-based services at the AAS.
In the electronic era, all aspects of a job have to mesh well. The different departments of the AAS used to work rather independently; but that will no longer do. Billing forms, for example, have to be processed automatically. We no longer have a person who transcribes the billing form and fits it into the system. Therefore the design of billing forms and meeting registration forms has to be carefully coordinated with the capabilities of the financial software. An electronic mailing list has to be obtained from the membership department. After the electronic mailout, bad addresses have to be noted and returned to the membership department. Those who create the meeting announcement have to consider how it will look, not only on paper but also on the screen. Forms must be electronically read when they are returned. And so it goes.
We at the AAS have adopted a "matrix management" approach. For every task, we now assign teams that include people with relevant experience from a cross section of internal departments, often teamed with outside consultants. Only in this way can we make effective use of the new electronic tools. It has been a difficult transition for the AAS executive office, and we see similar transitions taking place as our publishers and editorial offices gear up to handle electronic manuscripts. Our publishing effort can no longer be thought of as a number of separate stations that process a manuscript serially. Each stage must now be closely integrated with the others. Compatible software, data-interchange standards and full integration of processes into the overall publishing stream have become indispensible. The difficulty of undertaking such a paradigm shift within an organization must not be underestimated. We see a similar paradigm shift in the way science is being done. Remote computing and the needs of g eographically dispersed research teams will require stricter standards of communication than we have been accustomed to.
The established journal publishers are widely perceived as not proceeding fast enough in adopting the new technologies. Why is this so? Obstacles are raised not only by the sheer size of the job but also by the very nature of how organizations adapt to new technologies. Established organizations, even ones that became preeminent by introducing new technology in the first place, have difficulty continuing to do so.[9] This seems to be the situation with many publishers. The experiments in electronic publishing are being promoted from outside the organizations responsible for the daily production tasks. History shows that, as technology advances, good companies lose their preeminence by failing to sustain vigorous innovation. For the established publishers the warning is clear: Move rapidly into this new arena or run the risk of becoming supplanted by new methods.
Another factor that retards the adoption of electronic publishing is sheer size. The scientific literature is the final, vital link in our research enterprise. We must maintain the production of the present journals, even as we move toward electronic publishing. Journal production is an enormous task. Even as small a society as the AAS publishes more than 20 000 pages a year. That's about 100 pages every working day. Such an undertaking requires a smooth, well-automated system. To experiment on that scale is to invite disaster. Individual human intervention on a manuscript is prohibitively expensive. No electronic system that requires a lot of additional handwork on each manuscript will serve the needs of a major publisher.
As any copyeditor will tell you, authors are notoriously poor at adhering to publication standards. As we move into the period when manuscripts will become part of a searchable database and the references are expected to become links to other electronic documents, it becomes absolutely essential for authors to adhere precisely to the standards for markup, references and even the specification of characters. If you want a minus sign, a hyphen will not do. If you want a hyphen, a dash will not do. If you expect that your references will actually be able to link the reader to another document, you must be "computer precise" in specifying the reference. We foresee publishers being forced to charge authors extra for making special corrections.
One of the major challenges facing producers of electronic information is how to charge the users. They must collect enough revenue to support the infrastructure needed for a robust information delivery system. On the other hand, revenue collection should not inhibit the use of the electronic tools.
As publishers gain experience in the day-to-day production of electronic products, the issue of costs versus savings is becoming more focused. At first it was widely speculated that producing a refereed journal in an electronic form would cost about half of what it costs to produce the same journal as a traditional paper product. Because the authors would be doing the typing and formatting, publishers assumed they would no longer have to hire their own keyboarders. The predicted savings on printing and mailing costs alone were enough to impel most of the publishers to investigate the possibility of producing an electronic journal.
We now know that the savings realized by the elimination of in-house keyboarding are just about offset by the additional cost of copyediting a manuscript on-line. Moreover, there is strong demand that journals continue to be produced in the old paper form as well as in the new electronic form. Preparing the electronic material for both paper and electronic delivery is a significant additional cost that makes electronic production a break-even proposition at best. As publishers gain experience, and more effective production tools are developed, some savings should be realized, perhaps as much as 25 percent. Steven Harnad's provocative estimates of 70 percent savings[10] are, we believe, unrealistic if the research community expects the present high standards of scientific publication to be maintained.
In the astronomical community, production costs for the main journals are supported largely by page charges assessed to authors. Subscription fees finance the remainder of the operation. As electronic production becomes more efficient, we may see a lowering of page charges. Subscription revenues are likely to be replaced by a system of site licensing for access to electronic journals, with large institutions paying a higher charge than smaller institutions or individual subscribers. The charging system for a journal will inevitably be tied to its distribution method.
As we move into the future, the research community must encourage small experiments that will help to define just what it is that the users need. There should be, for example, experiments on the value of open commentary for assessing the importance of a paper. There might also be experimental development systems that incorporate video and audio into research papers, and personal alerting systems to help researchers stay abreast of the literature. The possibilities are almost limitless. We invite the membership of the AIP societies to help us develop new solutions to old problems and find new ways of communicating in our quest to understand the universe.
We thank the National Science Foundation for supporting the development of an electronic version of the Astrophysical Journal Letters. We also thank the editors, Helmut Abt, Alexander Dalgarno and Gene Avrett, for their suggestions and support. Evan Owens of the University of Chicago Press has carried the burden of translating our ideas into an efficient production process. Finally, none of the progress toward electronic publishing at the American Astronomical Society would have been possible without the contributions of our electronic publishing team: Chris Biemesderfer, Archie Warnock, Jim Fullton, Jeanette Barnes and Lee Brotzman.
1.
B. Hayes, American Scientist 82, 416 (1994).2. A. Heck, http://cdsarc.u-strasbg.fr:80/~heck/p-eih.htm.
3. Available at http://chara.gsu.edu/BeNews/intro.html.
4. Available at http://xxx.lanl.gov.
5. Available at http://www-spires.slac.stanford.edu:80/FIND/ spires.html.
6. P. Boyce et al., Electronic Publishing Insert, AAS Newsletter, no. 62 (1992) or http://www.aas.org/ and click on "electronic publishing projects."
7. To see the AAS meeting abstracts, visit http://www.aas.org/ and go to the link for "meeting information."
8. Astrophysical Journal Letters is available free of charge until the end of 1996 from the AAS home page at http://www.aas.org/. Follow the "Astrophysical Journal on-line" link.
9. J. Bower, C. Christensen, Harvard Business Review 73, 43 (1995).
10. S. Harnad, various papers at http://www.princeton.edu/~harnad/.