ELECTRONIC PUBLISHING TAKES JOURNALS INTO A NEW REALM
Publications slip off restrictions of print world and carve out a unique identitySophie L. Wilkinson
A leap forward in information delivery technology can have a dramatic impact on society.
Johannes Gutenberg's printing advances in the 15th century loosened the controls on publishing by broadening and democratizing information content and access, and ultimately helped to promote the cultural and scientific upheavals that followed. The Internet- and the World Wide Web in particular- represent another step in this ongoing revolution.
It will take some time to find out just how important this step will be. Authors, publishers, and librarians are only just beginning to take advantage of the potential of electronic media.
As they do so, they are having to reexamine the "givens" that have defined the world of scholarly print publishing for decades. Both the format and content of scholarly information products are being reevaluated. Other issues under discussion concern the role of publishers, the necessity for peer review, the nature of information ownership, and publishing and subscription costs.
At this point, many of these issues remain unresolved-much as they did three years ago when C&EN last took an in-depth look at electronic publishing (C&EN, March 27, 1995, page 42). On the other hand, much of what was then experimental has now been implemented.
But there are nearly as many notions about the correct way of proceeding in electronic publishing as there are stakeholders. Traditions characteristic of particular scientific fields in the print world carry over into the electronic medium, affecting what R. Stephen Berry, a chemistry professor at the University of Chicago, calls the sociology of adaptation to electronic media. Berry is chairing a study called "Transition from Paper," sponsored by the American Academy of Arts& Sciences.
High-energy physicists, for example, were accustomed to "making preprints of every new paper and shipping these out to about 400 people before anything got published," says Berry. As a result, researchers in this field were receptive when Paul H. Ginsparg, a physics research staff member at Los Alamos National Laboratory in New Mexico, started up an "e-print server" in 1991 to distribute the preprints electronically.
But the success of Ginsparg's server has not been replicated in fields such as chemistry where preprints are uncommon, says Andrew M. Odlyzko, head of the mathematics and cryptography research department at AT&T Labs, Florham Park, N.J. Scientists in these sectors are reluctant to participate because journals such as Science, the Journal of the American Chemical Society, and the New England Journal of Medicine won't consider a paper for publication if preprints have been widely distributed, Odlyzko says.
Some researchers and librarians are fighting such restrictions by starting their own publications, but conventional publishers have the upper hand for now. Commercial, professional society, educational, and other publishers produced nearly 7,000 print scholarly science journals in 1995, according to the latest figures available from Carol Tenopir, an information sciences professor at the University of Tennessee, Knoxville, and Donald W. King, an information sciences consultant based in Ann Arbor, Mich. Life sciences accounted for nearly one-third of the journals, and physical sciences, engineering, and environmental sciences together represented one-quarter. The remainder were in disciplines such as social sciences and mathematics.
The number of electronic publications is small but growing quickly. In physical sciences, life sciences, and technology, there currently are about 2,000 electronic titles, according to the Association of Research Libraries. Somewhat less than half are peer reviewed.
Some electronic journals feature peer review
Peer review is a contentious matter. On the one hand, it could become even more important as electronic documents proliferate. In such an environment, peer review could help select highly credible items the reader wants from an overwhelming supply of information.
But some researchers have suggested peer review is no longer needed and is simply an excuse for publishers to control information.
Unfortunately, the mere perception that the electronic medium is abandoning peer review has given people the sense that "the research isn't as good as research appearing in a paper journal"- regardless of the presence or quality of the actual peer review, noted Karla L. Hahn, bibliographer with the University of Maryland libraries, College Park. She spoke at the Socioeconomic Dimensions of Electronic Publishing Workshop, held last month in Santa Barbara, Calif., in conjunction with the 1998 Institute of Electrical & Electronics Engineers (IEEE) International Conference on Advances in Digital Libraries. The workshop was sponsored by the National Science Foundation and the IEEE Foundation.
Peer review doesn't have to be an all-or-nothing proposition, however. Review could simply evolve into a new configuration. It might take the form of a list of recommended "essential reads" for a given subject from any given time period, Ginsparg says. "There could also be retroactively added descriptive information, saying, 'This paper was important since it drew upon a, b, c (hyperlinks to sources) and led to new developments x, y, z (more hyperlinks).'"
There may ultimately be a "continuum of publications running from preprints up through some kind of a certified result," suggests Odlyzko. The author might first post results on their web page and perhaps a preprint server, and then some scholars could attach comments saying, "'This looks great; if it's true, it is a breakthrough.' And then maybe somebody else later could do a thorough review and say, 'No, the author made this mistake.'"
Some of this continuum is already in place. Based on feedback from readers, authors tend to revise preprints they have posted on Ginsparg's physics server at Los Alamos, for example.
In essence, documents are shifting from being fixed entities to works in progress. Old versions may still be valuable and suitable for archiving; a preprint, for example, may differ considerably from a final paper, but it could have generated significant discussion in the field.
This may necessitate the establishment of "version management" akin to that found in software development. Preprints posted on Ginsparg's physics server already can be viewed in both current and previous versions, which he says can "aid adjudication of priority disputes." And the resulting documents may have a natural life span, ending at an assigned "death date," at which point they are no longer current, some participants at the electronic publishing workshop suggested.Added value
However peer review and other issues are resolved, the central question remains: Will the scholarly world shift to purely electronic journals? And if so, Berry asks, "will we treat them as second-class journals or will they become equivalent to the Journal of the American Chemical Society, Physical Review, and Science?"
The answer depends in part on the concept that "print is a useful medium for ease of reading and portability, and authors continue to favor it," according to Eric A. Swanson, senior vice president of scientific, technical, and medical publishing at John Wiley & Sons, New York City.
Demographics play a role in this." When I went to graduate school, you went to the library and did a literature search using printed Chemical Abstracts," says Mary E. Scanlan, director of the American Chemical Society's journal publishing operations.
But an ACS study of a group of University of California graduate chemistry students last year showed that "if they can't get information via their computers-which, unlike the library, are available 24 hours a day-they may not [bother to] get it," Scanlan says. "It's almost like they would need directions to get to the library."
The answer also depends on whether the journals are simply electronic mimics of paper journals or whether they provide added value in their on-line forms.
Certainly, electronic editions can speed information dissemination. At its publications web site, ACS posts individual journal articles just as soon as they make it through the review, editing, and author-proofing process, a format ACS calls As Soon As Publishable (ASAP). The production process has been modified "so that instead of thinking of journals as batches of articles that get bundled up and put into specific issues that come out once a week or once a month, we now treat each article as it is finished," says Anthony Durniak, director of ACS's special publishing operations. The ASAP system gives these articles a head start-as much as 11 weeks-over their appearance in print. Other organizations face much larger backlogs of as much as a year or longer, and rapid electronic delivery can have a huge impact in skirting this roadblock.
In many cases, however, the most time-consuming phase of the prepublication process is in the review and revision stage. This won't be substantially improved by electronic processing, according to R. Keith Raney, assistant supervisor for the ocean remote sensing group at Johns Hopkins University's Applied Physics Laboratory, Laurel, Md., who spoke at the electronic publishing workshop.
But increased speed isn't enough to satisfy everyone. Many who offer on-line journal articles are simply providing "rapid delivery over the Internet of their standard paper articles," says Peter B. Boyce, a senior associate at the American Astronomical Society (AAS) in Washington, D.C. "You might as well just send them by Federal Express." What Boyce would like to see are electronic articles that carry value-added features that can't be replicated in the paper medium.
For example, an on-line journal can easily add click access to any of its own articles that are cited in a given paper, Berry says. Potentially it could also offer such access to cited papers from other publishers' journals.
Other possible features-several of which are already being used-include the following:
Publishing is shifting from a focus on issues of a journal as the fundamental unit to a focus on individual articles." There is already significant discussion by major scientific publishers around releasing published material on an article-by-article basis, not bound to any specific weekly or monthly issue," said Aaron H. Bigman, vice president of technology at Cadmus Journal Services, Linthicum, Md., who spoke at the workshop. "The driver here is clearly the early release of scientific content."
Form and content may fragment even further as publishers move away from simply replicating print conventions on-line. At the conference, Joost G. Kircz, visiting scientist at the University of Amsterdam, the Netherlands, and associate publisher for physics and astronomy at Elsevier Science, Amsterdam, proposed that scientific material be prepared in modular form, with elements that can be understood on a stand-alone basis. "The document ceases to be a linear text, written as if it is to be read from top to bottom," Kircz said.
Some of the modules could potentially be used by more than one author. The description of a common experimental technique, for example, could be written up in a module that subsequent researchers could refer readers to via a link, without having to duplicate the write-up themselves.
Regardless of the eventual-and unpredictable-form that electronic products ultimately take, Durniak expects that the number of readers will increase. In the print world, a potential reader of a somewhat esoteric journal might not bother to dig out a copy from a library in another building. But the instant access provided on-line may lead them to do so, Durniak explains.
At the same time, the total number of print subscriptions will drop. ACS has already seen an erosion in subscription numbers over the past five to seven years, says ACS Publications Division Director Robert D. Bovenschulte. In part, the drop is because subscriptions for individuals are being replaced by sitewide, institutional subscriptions. It is also because readers and libraries will want the bells and whistles that electronic journals provide. As a result, "scientific publishers-and maybe journal publishers in general-are going to have to give up counting print subscriptions as a measure of journal success," Bovenschulte says.
Bovenschulte believes "electronic journals are going to take over in the long run, no sooner than five years, but no longer than 10." AAS's Boyce has a view that is more radical. In just three years, he predicts, "people will not be reading the paper copy anymore because there's not enough stuff in there. It will be missing so much of the enhanced content that electronic makes possible."Archiving
The deficiency in content will affect journal archives as well, so in the future a paper archive will not be sufficient to capture the full value of electronic articles, Boyce warns. This brings up the thorny issue of responsibility for sustaining archives and access to the archives over the long haul. Some observers question whether the new independent electronic journals-and their archives-will still be around years from now, or whether their contents will be irretrievably lost.
And what happens when the subscriber exits the picture? Electronic journal subscribers are "buying not only the current material, but also back files and the whole system that allows them to use the journal in an enhanced way," Bovenschulte says.
If a library cancels a print subscription, it still has access to its old issues. But in the electronic world, that's not the case; the library is cut off from archive access. One possible solution Bovenschulte suggests is to provide the library with the information for the covered years via CD-ROM or tape. "But who knows how long CD-ROMs will be around? That technology could well be outmoded even five years from now," he says.
These issues are even starker for electronic-only journals, "because there is no recourse," Bovenschulte warns. "You can't go back to print."
In the long run, Bovenschulte says, the question is: "Who is going to guarantee 'perpetual' availability of information in electronic form?" The answer may lie in creating a central repository. Another option that Ginsparg suggests is a unified database. "People using the current physics [and other] archives find it extraordinarily useful to know they need consult only one comprehensive source, rather than conduct multiple searches."
The beauty of a unified-or even just a richly linked-archive is that it will boost research productivity "as fewer wasteful duplicative experiments are done," Wiley's Swanson says. The time this saves can be "allotted to an increased number of promising points of departure identified by fast, sophisticated, and comprehensive searching."Change at what cost?
While archiving, peer review, and product format are controversial, what really boosts stakeholders' blood pressure is the issue of money.
Library costs are staggering. Harvard University, the biggest academic spender, poured $70 million into its library budget last year, and the University of California, Berkeley, invested $34 million in its own system, according to the Association of Research Libraries (ARL).
Despite these large budgets, rising prices for serial publications are forcing libraries into Solomonic decisions. Even Massachusetts Institute of Technology is canceling journal subscriptions because of high costs, said Brendan J. Wyly, public services librarian at Cornell University's Johnson Graduate School of Management Library, who spoke at the electronic publishing workshop. "And if it's not held at MIT, where am I going to get the darned thing?"
Here's the problem: Between 1986 and 1996, North American research library expenditures for all serials grew nearly 10% per year, while the average cost of an individual serial subscription increased 147% to nearly $220, Mary M. Case, director of ARL's office of scholarly communication, told workshop participants. Those increases compare to a rise in the Consumer Price Index of just 44%.
A study of more than 5,400 arts and humanities, social sciences, and science titles indicated that "the journals with the highest average prices in 1997 were all in the sciences," Case said. This year, physics titles top the charts at about $1,600 per title on average, followed by chemistry at $1,577. Although engineering journals cost a "mere" $867 on average, they are one of the leaders in terms of price increases, jumping 76% in the past five years.
Admittedly, publishers face significant expense in producing such journals. A typical scientific scholarly journal's fixed" first-copy" costs, which cover activities such as peer-review administration, editing, and layout, run to $400,000 annually, while production and distribution comprise a mere $40 per subscriber, according to Tenopir and King. "This means that the publisher must charge $840 per subscription for 500 subscribers and $80 for 10,000 subscribers, just to recover costs," they estimate.
Because the production and distribution cost is relatively small for most journals, and "most activities that are performed are common to both [electronic and paper] media," Tenopir and King conclude the costs of electronic journal publishing aren't much less than traditional paper.
While these fixed costs are substantial, librarians and scholars believe that publishers-particularly commercial ones- are getting greedy. And they are beginning to band together to encourage publishers to price electronic journals with restraint. Librarians "became angry" with Reed Elsevier because the publisher initially planned to "charge quite a lot for access to the on-line versions of its journals," Berry says. "Librarians at a number of universities got their backs up and said,' No, we won't pay.'"
Commercial publishers offer the defense that they put out journals "for niche markets where individual subscribers are minimal in number and institutional subscriptions must carry the load. The publishers also argue that they are pressured by their editors to expand the number of issues and pages published each year to accommodate the increasing numbers of excellent manuscripts," driving up costs, Case noted.
Peter T. Shepherd, managing director of Elsevier Science Switzerland, Lausanne, adds that professional society publishers have recourse to" more revenue streams, notably page charges, member subscriptions, and advertising."
Reed Elsevier's scientific publishing division's operating margin (before exceptional items) is about 40%, according to Case. And Wiley's Swanson says that while his firm "is a healthy growing company with steadily improving financials, the company's overall operating margins are not in the 40% range." Professional society margins are even lower. ACS's journal surplus (revenues less expenses), for example, falls in the single-digit range as a percentage of revenues, according to Bovenschulte.
These differences result in the following kinds of figures: In 1995, commercial publishers of scholarly science journals on average were charging institutional subscribers $487 per journal, which translates into $4.10 per article and 27 cents per page, according to Tenopir and King's latest figures. That compares with professional society publishers' average charges of $229 per subscription, $1.10 per article, and 10 cents per page.
Publishers are hunting for the right combination of price and content for their electronic and print products. Many are still trying to gauge customer response, in some cases by putting a price option in place and then waiting for feedback. At the electronic publishing workshop, Frederick T. Andrews of IEEE's strategic planning and institutional research department, Piscataway, N.J., characterized this as the "Do the dogs eat the dog food?" approach. "Unfortunately, the magic recipe is very difficult to find."
All 26 ACS journals became available on the web last September. Articles are provided in both HyperText Markup Language (HTML) and Portable Document Format (PDF) versions, and links to other databases such as Chemical Abstracts, GenBank, and Medline are available. Users can search the journals via keywords in titles, abstracts, or full text, and can browse the author index or tables of contents. Customers also can access more than 200 on-line journals from ACS, the American Association for the Advancement of Science, the Royal Society of Chemistry, Springer-Verlag, and others via a link to the ChemPort collaborative chemical information database. Customers pay according to the amount of searching they do.
ACS has caught a lot of flak over its experiments with pricing schemes. "Pioneers are the ones that always have the arrows in their chest," Durniak says ruefully.
Objections spring in part from other publishers, such as the Royal Society of Chemistry, the American Physical Society, and the Institute of Physics, offering free electronic access to their print subscribers. ACS responds that its electronic journals offer considerable added value compared with the print versions, such as linked references, supporting information, and ASAP articles, but some readers say they would rather pay less and do without all the bells and whistles.
One pricing option ACS developed allows subscribers to cancel current print subscriptions and replace them with electronic ones, or to combine the two formats. (But subscribers are not required to get electronic access to all the journals for which they have print subscriptions.) Depending on the customer's usage levels, such subscriptions range in price from 5% more than the print subscription for web-only access for a single department to 90% more than print for both print and electronic access combined for an entire work site.
A second option offers libraries a site license for the ACS web journals at a 25% premium as a supplement to print subscriptions. In this case, the institution must keep all the print subscriptions it already has in place for that year.
Another plan permits libraries at different universities to band together in a consortium to buy web access to journals for a 25% premium, providing each library with electronic access to all the titles any one consortium member subscribes to in print.
Nonsubscribers can purchase individual ACS articles over the web by credit card at $25 each or through other document suppliers.
Wiley, which says it is the number one commercial chemistry publisher worldwide, launched Wiley InterScience in October 1997 to offer Internet access to 50 of its journals. Currently, 150 Wiley journals are on-line in PDF format, and by early next year the company expects nearly all of its 400 scientific, technical, medical, and professional journals will be on-line. Publication is roughly simultaneous with print, and the company's goal is to post articles on-line several weeks prior to print.
Wiley's electronic journals currently are free to anyone; the company is gathering user feedback and will announce a pricing plan by early summer. Wiley offers searchable contents listings and abstracts, as well as personal home pages where users can activate automatic customized searching and store search results and personal annotations.
Elsevier Science describes itself as the world's leading scientific publisher, with more than 1,200 science journals. All of the firm's journals are available electronically in PDF format and most are in HTML. Subscribers can take these either in lieu of or in addition to paper editions. Most of the on-line journals appear simultaneously with the print editions, Shepherd says. Tables of contents can be supplied prepublication via e-mail to customers.
Shepherd says Elsevier Science has experimented with a range of formats for its electronic journals in recent years. The company found that "stand-alone on-line journals that are the same as the hard copy without added features such as links to supplementary data have not had a great impact," he says. On the other hand, "electronic-only journals, such as New Astronomy, which offer entirely new ways of presenting information and do not mimic hard copy, have had a huge impact and have attracted many authors and readers."
One of the most intriguing is the firm's Combinatorial Chemistry-An Online Journal, which will draw articles from Elsevier Science's other journals together with commissioned reviews and summaries of articles from other publishers. "Instead of stimulating further fragmentation of the literature in this field-which the three new traditional combinatorial chemistry journals recently announced by other publishers will do-this new service will be integrative and a good model for dealing with emerging fields in an electronic environment," Shepherd says.
Elsevier also found that use of electronic journals is enhanced by making them available "in the context of a range of other 'community' features, such as those in our ChemWeb system, where there is a virtual conference facility, shopping mall, database section, etc.," Shepherd says.
The company is experimenting with different pricing models, he says. For example, one subscription package provides access to abstracts of articles from, say, 300 Elsevier Science journals in a given field, along with downloads of a specified number of full articles from these journals.
Institutions subscribing to AAS's print journals are charged nothing extra for electronic access; the libraries simply sign a sitewide institutional license. By redoing its production process, and taking advantage of automation that has been built into it, AAS can now produce its paper and electronic journals at less than the cost to produce just the paper versions three years ago, Boyce says." The people who say, 'It's going to cost us an extra 20% to make our stuff electronic' are going about it wrong. It doesn't have to be that way."
The American Physical Society (APS) also makes electronic versions of its journals available at no additional charge to hard-copy subscribers. Editor-in-Chief Martin Blume, Ridge, N.Y., says the costs have been covered by working them into the price for paper subscriptions. The journals are in PDF and Adobe PostScript format, with references in HTML so they can be hotlinked. Subscribers and nonsubscribers alike can browse tables of contents, published abstracts, and an advance listing of accepted papers scheduled for upcoming issues. Nonsubscribers can purchase individual articles.
In July, the society will put its Physical Review archives on-line, with issues going back to 1985. Access will be free for six months as the kinks are worked out of the system, and the cost will later be bundled in with the price levels for institutions that carry all of APS's journals. Those who take just one or two journals will be able to obtain archive access for $300 per year, Blume says.
There are, of course, many other ways to charge for access to electronic products. IEEE's Computer Society is considering a plan to offer package subscriptions, including several on-line journals, at a price that is greater than the cost of one journal but much less than the cost of all of them, Andrews told the electronic publishing conference.
Another alternative is to bill authors for submissions, and perhaps charge another fee upon acceptance of their papers, with the total coming to about $300 to $400 per article, says AT&T's Odlyzko. "With these fees paid by the author, the journal can make its articles freely available to everybody."
In fact, Odlyzko, who is on the boards of several free electronic journals- including the independent publication Electronic Journal of Combinatorics- believes that doing away with subscription charges altogether is feasible.
Blume, on the other hand, says that scientists are rebelling against page charges, and APS is phasing them out for manuscripts that are submitted electronically. Such fees can represent a significant part of scientists' relatively small research budgets, particularly for theorists, while these expenses could be more easily absorbed by their institutions. "The small fish can make a significant meal out of the scraps that fall out of the shark's mouth," Blume says.
APS is relying on such institutional funding for its new electronic journal, Accelerators & Beams, so that it can be distributed free. But Blume notes the journal serves "a special community where there is interest on the part of large, well-endowed institutions to support a high-quality, peer-reviewed journal." He realizes, however, this experience may not translate into other fields.Publisher value
Part of the resentment about journal prices can be traced to publishers' charging subscribers a lot of money for access to research papers and refereeing services that they receive for free. Some question whether publishers themselves bring much to the table.
Bovenschulte believes publishers such as ACS add considerable value to the articles they disseminate. "We still do a great deal of manuscript editing, whereas many publishers have very substantially curtailed how much they do."
Publishers also are investing heavily in technical infrastructure development. ACS has spent "many millions of dollars over the last five years getting to the point where we can offer not just a page image thrown up on the web, but HTML format as well as PDF, which allows for all the linking and fancy graphics," Bovenschulte says. "It is costing us approximately $2 million per year to support web publishing," what with investments in new staff, training in computers and networks, more servers, and backup systems. That expense "is likely to increase-and we still have all the costs associated with doing print," notes Bovenschulte.
Nevertheless, authors and libraries believe that journal prices are too high, and they are casting about for alternative means of broadcasting research results." As a matter of public policy, it is time for the research and educational community to act to ensure long-term access to the research conducted with public funds and by the faculty of our universities," ARL's Case said at the electronic publishing conference.
ARL, which consists of 121 of the largest North American research libraries, has launched the Scholarly Publishing & Academic Resources Coalition to" seek publishing partners interested in entering the serials market in areas in which prices are highest and there is the greatest need for alternative models of research communication," Case said. These sectors include expensive commercially owned journals in inorganic chemistry and nuclear physics, she added. The group has three projects under way and has begun negotiations with scholarly publishers to launch competing titles.
While these efforts mature, Case said, other strategies need to be put in place, including options for "faculty and the university to retain and better manage intellectual property rights and the decoupling of the academic credentialing process from formal publication."
In the meantime, the shift to electronic publishing will continue. And it will no doubt flower into something we can't even conceive of at this point. Robert A. Kelly, APS's director of journal information systems, draws an analogy with the history of filmmaking, which in its first decade simply recorded stage plays with a stationary camera. Once filmmakers realized they could move the camera, and began experimenting with other innovations, cinematography exploded into an entirely different medium, freed from the strictures of the stage.
In publishing, too, we have successfully duplicated print conventions in the electronic world, and we are just beginning to realize that we can "move the camera" as we explore the medium's unique capabilities. Eventually, Kelly says, "you won't be able to replicate an electronic journal on paper, just as you couldn't put 'Star Wars' in a stage play."