Free Internet access to all journal articles will greatly improve the present publishing system and cost substantially less.
Thomas J. Walker, University of Florida, Gainesville
Summary. For nearly
300 years, most scientific journals were published by societies
and were a source of little or no revenue. In the last 40 years,
commercial publishers have displaced societies as the principal
publishers of scientific journals, and journal subscription prices
have greatly outpaced inflation, causing major declines and distortions
in the serials and book acquisitions of research libraries. Within
the next 5 years, central printing and library archiving of printed
issues will cease in favor of distribution via the Internet and
digital archiving. The total cost of publishing will be much less
and will be recovered through page charges. Copyright will be
retained by authors and access to articles will be free. E-published
articles will at first retain the traditional formats but as central
printing is discontinued they will take full advantage of the
interactivity and connectivity of the Internet. The transition to
all-electronic distribution will be led by scientific societies, because
once their members learn the advantages, they will insist that their
societies make the change.
This scenario may seem far fetched, but anyone who has followed the development of the Internets World Wide Web during the past few years knows that information can be made available worldwide very easily and at very low cost. Since scientists want their research results available as widely and conveniently as possible, distribution via the Internet seems inevitable. Nonetheless, important issues must be resolved before researchers benefit from what is technologically simplefor example, what electronic format will be used and how will costs be allocated. Publishers who profit from the present system will be reluctant to embrace free electronic distribution of primary scientific literature, but authors will be eager and will vote with their manuscripts by sending them to the journals that better and more thriftily serve them and their sponsors. This essay first describes radical changes that have occurred in journal publication and journal prices during the past 40 years and summarizes the present economics of journal publication. Next it describes how electronic means can deliver articles directly to their users for almost nothing. Finally it describes the electronic future of journal literature and predicts how the transition is likely to occur.
Traditional publication of journals
Following a very long period of steady growth, the number of articles submitted to science journals surged, thus creating a variety of new problems and opportunities for scientific societies, commercial publishers, and research libraries. Changes in journal publishing are here described as occurring in three periods. The dates of transition from one period to the next are necessarily arbitrary because the changes were gradual.
1665-1960: Journals proliferate and change slowly.
In 1665 the Royal Society of London published the first issue of the first scientific journal (Schaffner 1994). The journals purpose was to disseminate the results of its members research. Researchers who had earlier exchanged letters describing their research could now reach a wider audience.
Journals soon became a means of establishing priority to new discoveries, were accepted as the permanent record of research, and were archived by libraries. Peer review of all or most articles was installed as a means of screening and improving what was published. Citations to earlier articles provided a way to weave previous research into the fabric of the new. As journals and articles proliferated, organizations were established to abstract and index what was published.
For nearly 300 years, the numbers of journals grew steadily, mostly as researchers founded new societies to promote new or newly important scientific disciplines. These societies helped members publish their research results by sponsoring one or more journals. For example, the predecessors of the present Entomological Society of America (ESA) were founded in 1889 and 1906 and published the first volumes of Journal of Economic Entomology (JEE) and Annals of the Entomological Society of America (Annals) in 1908 (Smith 1989).
Until the 1960s, most societies recovered publication costs largely from members dues, which included a journal subscription. Articles per author were relatively few, and many members did not publish at all. Library subscriptions were not a major source of income (in 1960, JEE and Annals each cost libraries $15, the same as membership dues).
Although scientific societies published most science journals, some were published by other nonprofit institutions such as universities, museums, and governments. Commercial publishers were generally not attracted to the field because there was little potential for profit.
1961-1980: Articles surge and not-for-profit publishers dont accommodate.
Precursors. Growth of science had accelerated during WWII and sciences ability to improve standards of living in postwar America was manifest in products as diverse as nylon stockings, televisions, and antibiotics. USSRs launch of Sputnik in 1957 provoked a fear that the U.S. was losing technical superiority, and federal expenditures on research and development and for science education rapidly increased over the next decade, often at double-digit rates in inflation-adjusted dollars (CCFSRD 1995). The demand for researchers and for science educators skyrocketed. In response, the annual number of science and engineering PhDs awarded in the U. S. tripled between 1958 and 1968 to about 18,000, continued to increase until the early 70s, and then declined slowly but remained above 18,000 until 1977 (CSEPP 1995).
Causes of surge. With many more researchers and with support of research easily available and generous, submissions to journals surged. The surge did not abate when grants and academic jobs eventually became more difficult to acquire. After all, an important criterion of research success is the number of papers published, and researchers seeking jobs, grants, tenure, and promotion wanted to improve their chances in an increasingly competitive environment.
Responses to the shortage of journal space: scientific societies. Societies soon faced the problem of having to reject good manuscripts and to delay publication of accepted manuscripts because their journals and their ability to subsidize members publication were at capacity. Granting agencies faced the dilemma of paying for research that could not be published in a timely fashion or at all. Thus in 1961, to alleviate the financial strains on journal publishing, the federal government approved the payment of page charges by federal agencies and from federal grants to nonprofit publishers (Spilhaus 1982). Societies quickly took advantage of this new source of revenue to publish more pages in their established journals and to start new journals. For example, ESA approximately doubled the size of JEE in the four years after it began page charges, and it started a new journal, Environmental Entomology, that by its second year exceeded the pre-page-charge size of JEE. Nonetheless, societies did not come close to satisfying researchers burgeoning needs for publishing outlets.
Responses to the shortage of journal space: for-profit publishers. Because societies failed to increase their publication efforts fast enough to keep up with the demand, commercial publishers seized the opportunity to offer researchers new outlets for their manuscripts. They started new journals in long-established fields; but, of greater impact, they identified new or newly popular research areas (e.g., Microbial Ecology, Behavioral Ecology and Sociobiology, Insect Biochemistry), and established journals in those specialties. In each case, the publisher would invite carefully chosen eminent scientists in the specialty to be members of an editorial board for the journal. Most such scientists were willing to be so recognized and to help establish a journal that would increase status and publishing opportunities for their field. With the endorsement of an international board of distinguished researchers, the publisher attracted the subscriptions and the manuscripts needed to start the journal on its way to becoming indispensable to researchers and to the libraries that served those researchers. Adding to the attraction of these journals to authors were their lack of page charges (only nonprofit publishers could be paid by the page from federal funds).
Consequences to research libraries. As fatter issues and more journals were published, research libraries found it increasingly difficult to pay for all the old and new journals that their clients needed. During the 60s they were largely able to keep pace. For example, between 1960 and 1970, 12 established research universities increased their acquisitions expenditures in constant dollars by150% and the volumes added by 117%. However, in the next decade their expenditures increased only 2% and the volumes added declined by 11% (Cummings et al. 1992). The reason for the discrepancy between constant dollars spent and volumes bought was largely an extraordinary increase in the real prices of science journals. Whereas the costs of publishing, as represented by prices of hardbound books, closely followed the Consumer Price Index, the average prices of science and technical journals diverged upward (Fig. 1). From 1970 to 1980, the cost of hardbound books, in constant dollars, declined 6%, whereas the cost of science and technical journals increased 73 %. (Fig. 1, left half). One means that libraries used to lessen damage to their journal collections was to reallocate some of the funds that would normally have been used to purchase booksat a time when the number of books published was significantly increasing (Cummings et al. 1992).
1981-date: Commercial publishers dominate.
No one seems to have placed a date on the occurrence, but by the beginning of the 80s, more than half of what research libraries paid for science journals was paid to commercial publishers, and since then the discrepancy has grown much greater. This is because commercial publishers continued to start new journals and especially because their subscription prices continued to soar relative to the CPI. By 1990, the average price of science and technology journals, in constant dollars, had increased 230% over what it was in 1970 (Fig. 1)!
It should be obvious that annual prices of journals are influenced by many factors and that gross comparisons may be unfair. For example, if a journal publishes more pages per year, its subscription price should rightfully increase proportionally. But page counts may also be misleading, because journals may increase their page sizes and reduce page margins, thereby giving the buyer more print per page. In 1994, I selected three society-published journals (Florida Entomologist, Canadian Entomologist, and Journal of Economic Entomology) and four commercially published journals (Journal of Insect Physiology, Physiological Entomology, Pesticide Biochemistry and Physiology, and Journal of Applied Entomology) and determined the cost to libraries of each square meter of printed matter in each journal in 1973, 1983, and 1993. Then I adjusted the results for changes in the Consumer Price Index (Fig. 2). The results were clear. The three society journals cost less initially and their prices increased the least over the 21 years (-28% to +166%). The four commercially published journals increased an average of 271%, with JIP having the greatest increase (338%) and PE having the least (228%). [The predecessor journal to PE was published by the Royal Entomological Society. In 1976, the Society reorganized its journals and assigned them to a commercial publisher so that the journals would create income for the Society rather than requiring subsidy (Brady et al. 1978).]
The fact that in 1993 the three society-published journals cost on average only 14% as much as the four commercially published journals can be attributed to page charges and to societies being less motivated to maximize profits. Another factor contributing to the high cost of commercial journals is that many have very small numbers of subscribers, mainly because their prices are so high but also because in many cases the journals serve rather narrow specialties. Making matters worse, when tight budgets cause libraries to cancel subscriptions, prices must be increased further if profits are to be maintained.
By the early 90s, with access to the primary scientific literature increasingly costly and with many libraries no longer able to pay for many of the journals their researchers considered essential, a switch to electronic publication of research results was becoming viable.
Table 1. Simplified history of primary scientific publication and a prediction of the near future.
|Means of distribution||Issues and reprints||Issues and reprints||Issues and reprints||Internet|
|Dominant publishers||Scientific societies||Scientific societies||Commercial publishers||???|
|Principal items paying for publication||Dues, member subscriptions||Subscriptions, page charges||Library subscriptions||Page charges|
|Principal payers||Members||Authors grants or institutions||Authors institutions (through their libraries)||Authors grants or institutions|
Economics of traditional publishing
An understanding of the costs and sources of income for traditional journal publication is needed to clarify some of the opportunities and problems in switching to electronic publication.
Costs. Most costs of publishing a journal can be assigned to one of three categories: editing, composing, and runoff (i.e.,production and distribution) (Fig. 3, top). Editing includes receiving manuscripts, arranging reviews, corresponding with authors and referees, judging the quality and suitability of submissions, negotiating revisions, and making the manuscript conform to the standards and style of the journal. These are labor intensive tasks. Journals depend on mixtures of volunteers and part- and full-time paid professionals to accomplish them. Composing consists of typesetting and page makeup and is now done with computers. Its end products are digital files used to make the plates for press runs. Production and distribution involves printing, binding, and mailing. How costs are allocated among the three categories varies greatly among journals, chiefly because of diverse levels of expenditures for editing and numbers of subscribers. The first two categories are pre-run costs. The last cost category (runoff) is largely eliminated by e-publication and amounts to 30 to 70% of total costs (King et al. 1981, CBE 1982).
Income. Subscriptions are the chief source of income for journal publishers. For society-published journals, member subscriptions boost subscription income (Fig. 3, middle), but commercial publishers rely almost entirely on library subscriptions (Fig. 4). Furthermore, societies generally have page charges, whereas commercial publishers do not. Both types of publishers generate significant revenues from selling reprints. Society publications have negligible royalty income [less than 50¢ per page for ESA journals]. Because commercial journals are less widely available (no member subscriptions) and have higher photocopying fees (e.g., $2.25 per article for ESA publications vs. $17 for Journal of Insect Physiology), their royalty income is undoubtedly greater, but probably still amounts to no more than a few percent of the total.
Profits. Journals are often quite profitable (Fig. 3, bottom). Exactly how profitable cannot be determined for most commercial publishers, but Reed Elsevier, which publishes more than 1200 science journals with sales of ca. $880 million (Frank 1997), has an estimated pretax margin of nearly 40% (Hayes 1995). That may explain how it has amassed $1.7 billion with which to acquire other companies (Frank 1997). (Its biweekly Brain Research costs $14,919 annually and its monthly Journal of Insect Physiology costs $1,241.) In 1996, ESA made 53 % of its $2.3 million income from its journals, with libraries paying no more than $190 for an annual subscription to any of its bimonthly journals.
The electronic future of journals
The digital revolution
At the same time that prices for science journals were increasing faster than the ability of research libraries to pay them, changes were occurring in information technology that would eventually be the key to a better, more affordable system of primary publication. Most of us have so directly experienced the digital revolution that it needs no detailed description. For example, instead of writing manuscripts by hand and giving them to secretaries to type, we now create them digitally with word processing software on the computer on our desk. Many of us can easily use the same software to produce versions of submitted manuscripts that look just like articles from the journals we hope will eventually publish them. Finally, most of us can, without charge, put these manuscripts (in plain or desktop-published form) on our departments or institutions WWW server and thus make them directly available to most of the worlds researchers. We can also put materials relevant to these manuscripts on the same serverfor example, spreadsheets filled with data, graphs, and statistical analyses; color pictures; and audio and video clips.
Although we can Web-publish our manuscripts, submitting to the inconvenience and expense of traditional publishing in a recognized journal adds value to manuscripts by (1) certifying their worth (good for pay raises, tenure, and promotion), (2) soliciting peer review (which may reveal serious omissions or misinterpretations), (3) dating and preserving a fixed version (good for establishing priority and insuring continued access to the article), (4) improving spelling, grammar, and syntax (should it be needed), and (5) applying a standard style (some would say this is immaterial).
How then can traditional journals exploit the new digital technology of Web publication?
Electronic publication of journals
For articles in the primary scientific literature, the initial stages of traditional and electronic publication are the same. With the help of referees the editor decides which manuscripts will be accepted, the authors and editor agree on the final content of the articles, and the editor sees that the articles are copyedited and converted to digital files. Proofs are output from the digital files, and corrections made. In traditional publication, the final digital files make photographic plates used to print pages of the article on a large mechanical press (Fig. 5). Separate press runs, using slightly different plates, produce issues, which are mailed to individual and institutional subscribers, and reprints, which are purchased by the authors of articles. In electronic publication, digital files of the articles are posted on a server on the Internet. Articles then may be read and printed from any computer on the Internet, or access may be restricted to those with usernames and passwords or to those with certain Internet addresses.
Articles can be posted on the Internet in many formats, but only two are relevant: PDF and HTML (Walker 1997b). PDF (Portable Document Format) retains the exact appearance of the original article no matter what platform (PC, Mac, or Unix) it is viewed or printed from. It puts the equivalent of a photocopy onto the computer screen or into the hands of those who access it. HTML (Hypertext Markup Language), on the other hand, cannot do this, because its markup tags are rendered differently by different browsers, browser settings, and computer setups. HTML is designed to take advantage of the special features of the Webin particular, on-screen viewing and interactivity. With HTML the user may jump from an in-text citation, to the entry in the Literature Cited section, and sometimes (but increasingly) from there to the abstract or even to the full article. With HTML, the article can link to the authors raw data and to audio and video clips. With HTML, the article can include forms on which users can comment, add data, or suggest linkswhich will then be screened for possible posting with the original article. The power of HTML is illustrated by the interactive version of this article, which is at http://tjwalker.ifas.ufl.edu/aedraft.htm.
In spite of the advantages of HTML, PDF will be the dominant means of e-publication of articles as long as the articles are also centrally printed. The chief reason is cost. For articles digitally composed for paper, conversion to PDF is almost free. Making PDF files requires software costing a few hundred dollars and is as simple as sending files to a printer. On the other hand, composing in HTML is a large additional expense for articles originally formatted for paper. Figures must be converted to separate files and links to them inserted manually. Installing the hyperlinks that make HTML so appealing is labor intensive. Some types of internal links can be automated (e.g., links to entries in References Cited and to higher resolution versions of in-text graphics), but external links require knowledge, judgment, and manual editing. As an example of the high cost of HTML, the American Society for Biochemistry and Molecular Biology (ASBMB) paid HighWire Press at least $35,000 to develop an HTML version of Journal of Biological Chemistry (Fig. 6, top) and pays several thousand dollars per month in maintenance costs (Young 1997). In contrast, the Florida Entomologist went on the Internet for less than $500, and an issue of 150 pages now costs less than $400 to post (FE 1997a).
Another reason that PDF will dominate e-publication of centrally printed journal articles is that it delivers a format that researchers know, trust, and find convenient to use. Although HTML may be superior to PDF on screen, researchers often want hardcopy that they can annotate, read almost anywhere, and show or loan to colleagues and students. When an HTML version of an article is printed, its hyperlinks are dead, including those to the full-sized pictures of the in-text thumbnails. The large-print, one-column format that was appropriate on screen may require twice as many pages to print as the PDF version. For these reasons, ASBMB posts the PDF version of Journal of Biological Chemistry along with the HTML (Fig. 6, bottom).
The fact that PDF documents are not currently indexed by the robots of Web search services (e.g., AltaVista) can be overcome at low cost by posting simple text files in HTML. For example, for $30 an issue the Florida Entomologist posts minimal HTML files of all articles. These files lack figures and coherent tables but have all words and phrases intact for robot indexing (FE 1997b). Persons finding such a file are instructed to take the hyperlink to the PDF file of the article.
Advantages of e-publication
Traditionally published articles are inconvenient to access even for the technologically advantaged. To acquire a copy of a traditionally published article, a researcher must solicit a reprint from the author or make (or buy) a photocopy from an issue. Responses to reprint requests are slow and uncertain. Even the best-stocked research library may lack a needed journal or issue, and many researchers are served by minimal libraries or none. When the issue is in hand, photocopies may be of disappointing qualitye.g., if the librarys copy machine needs maintenance or if the issues of a journal were too tightly bound into a volume. In contrast, copies of an electronically published article can be printed directly off the Internet and the digital files can be saved for viewing or printing elsewhere or later. Those not on the Internet can ask anyone who is on it (not just the author of the article) to print them a copy or send them a file.
Traditional publication produces numerous paper copies that have a low probability of being used and yet must be stored and managed. Each library that subscribes to a journal and archives it must bind its issues into volumes (>$6 per volume) and provide space for the additional volumes, at a cost per volume of as little as $2 (for a book warehouse [Canelas 1997]) or as much as $30 (for a library annex underground in earthquake country [Lesk 1997]). Then, to maintain full-service access to the volumes, the library must pay more than $3 per volume per year for staff and operating costs (Canelas 1997). Authors who buy reprints must store them and contend with filling reprint requests. Each researcher that subscribes to one or more journals must contend with issues that take up much shelf space but contain few articles of direct interest. Electronic publication produces no unwanted paper copies and allows access to the articles as long as the files are posted on at least one Internet server. Paper copies can be made when needed and in the number required by any researcher, at any time, anywhere.
Traditional publication is expensive. Publishing a journal electronically costs less than traditional publication because printing and distributing paper issues is much more expensive than posting files on the Internet. Keeping an article mounted on one server is an infinitesimally small expense compared to hundreds of libraries keeping bound volumes secure and accessible. For example, the hard-disk space required to post on a Web server the 6,000 journal pages published annually by ESA costs about $100.
Traditional publication is restrictive. Color figures are so expensive that they are little used. Space is so dear that articles must be made brief by deleting details and abbreviating discussion. Even so, publication may be long delayed by lack of space in the current or next issue. On the other hand, electronic distribution makes publishing all forms and amounts of information economically feasible. Color pictures, sound and video clips, and full data sets are no problem, and there need be no publication queues.
Traditionally published articles are expensive to index. University of Florida pays $71,000 annually for seven major indexes to the agriculture and life science literature (e.g., Science Citation Index, $16,808). Electronic publication permits free, thorough indexing of articles by automated indexing systems. This is demonstrated by AltaVista and HotBot, free search services that index every word and phrase in tens of millions of Web documents.
A view of the electronic future
The move to electronic publication of journals has begun (Abate 1997, Hitchcock 1996, Walker 1997e). I expect these major changes to occur within five years:
Posting articles on the Web will replace central printing of issues and reprints.
Accessing an article on the Internet and making a copy is much easier than photocopying from an issue or volume at a library or obtaining a reprint from the author. It also costs much less. When researchers are willing to abandon less convenient, more expensive access, publishers will abandon central printing and save 20 to 60 % of the cost of publishing.
Journals will be digitally archived.
Libraries will continue to maintain permanent copies of what is published in scientific journals, but they will preserve digital files rather than ink on paper. As software and digital media change, digital technology will facilitate errorless transfer.
HTML will replace PDF.
When central printing is abandoned, the cost advantage of PDF will vanish and HTML will dominate. Problems with the present versions of HTML that must be addressed include how to help the user print hardcopy that will be as useful as a traditional reprint and how places within articles will be denoted (HTML documents are not paginated). The all-electronic, refereed journals that societies have started should help publishers learn how to exploit the strengths of HTML, how to compensate for its weaknesses, and what changes to lobby for. Conservation Ecology, recently begun by Ecological Society of America (1997), illustrates some of the innovations made possible by an all-electronic format.
Articles will be freely accessible.
Authors want to remove all impediments to others learning about their refereed, published research. After investing many $1000s in each completed research project and its write-up, authors and their sponsors are willing to pay a fair price for unimpeded access. The fact that most authors buy reprints (~95% in the case of ESA) and bear the burdens of fulfilling reprint requests is evidence that authors will pay to give others more convenient access to their articles. Another reason that electronically published articles will be freely accessible is that controlling their distribution is impractical. Once an author, or anyone else, has downloaded the file for an article, he/she can attach it to any e-mail message to any colleague. Since publishers will be unable to stop this practice, they will choose to charge authors enough to allow it. A final reason that access will be free is that research libraries will permanently post articles on their WWW servers at no cost to the publisher or for a small initial fee (FCLA 1997), but they will not support charging per use or restricting access to a specified number of downloads.
Libraries will no longer subscribe to journals.
Since access to journal articles will be free, subscriptions will be unnecessary. Researchers who presently must go to a research library to access the current and recent literature will be able to do so in their offices, laboratories, and homes (and libraries, should they be there for some other purpose).
Authors will retain copyright.
Having nothing to gain, publishers will no longer require researchers to sign over control of reproducing and distributing their articles. (Current copyright law provides that copyright resides with the author of a work or the entity that paid for the work unless copyright is transferred [LC 1997]. Prior to 1978, ESA did not ask authors to transfer their copyrights.)
Costs will be recovered through page charges.
Electronic publication nearly eliminates the costs of production and distribution, but the substantial costs of editing and composing must still be met, and both societies and commercial publishers will still want to make a profit. Revenues from publishing will come almost entirely from page charges or their equivalent (HTML documents currently are unpaginated). Societies, unlike commercial publishers, have the option of continuing the practice of using a portion of their members dues to support their journals. However, society members will have no more convenient access to the journals than do others. Of the other current sources of publishing revenues, only advertising remains, and that could become more lucrative.
Costs will be less.
Because the electronic substitutes for central printing and mailing are almost free, the cost of publication should be reduced by 20 to 60% when central printing ends. Additional cost reductions e.g., from better use of digital technology in editing and composingwill be encouraged by the ease with which authors can shop for the best publishing value. Both society and commercial journals will rely on page charges. No longer will costs of reprints and costs of library subscriptions confuse an authors choices. And no longer will the number of subscribers to a journal influence the accessibility of the articles it publishes.
Research institutions will continue to pay these costs.
Institutions that support researchers have always been the source of publication revenues. In the current system they pay through library subscriptions and through page charges. With e-publication, they will no longer buy subscriptions but page charges will apply to all journal articles and may be higher. (But remember, the total bill for journal publication should be much less.)
Back issues will be e-published and e-indexed.
Digitizing older journal literature improves access and reduces the need for libraries to use costly space and other resources to maintain volumes that are rarely consulted. In fact, a major project with this goal has been underway since 1995. JSTOR (1997), a nonprofit organization begun with a $1.25 million grant from the Andrew W. Mellon Foundation, plans initially to put on WWW complete back runs of at least 100 journals in 10-15 fields. Among the nearly 30 already posted are three journals of the Ecological Society of America (e.g., Ecology, vols. 1-75, 1920-1994). For about $2 per page (Guthrie 1997), journals are scanned, the bit-mapped image optically character-read, the OCR corrected, a full-text index created, and the bit-mapped articles put on line. The JSTOR database can be boolean searched for any word or phrase in the fulltext of the articles. (When I entered the name of one species among ca.150 such names in an Ecology article published in 1957, JSTOR found the article among all articles in its ecology journals in ca.12 seconds!). JSTOR will continue its conversions by charging libraries to use its database, with larger libraries paying proportionally larger fees.
As described above, e-publication of journals will serve the research community better and more economically. How is the transition from the present system likely to occur?
Publishers will profit from selling electronic reprints of traditionally published articles.
Publishers will substantially increase their profits by selling a new type of reprint. Instead of selling paper copies of articles to authors, publishers will, for a price, make a PDF file of the article and post it on the Web, where it can be viewed and printed by anyone that has Internet access. If such e-reprints are produced and posted concurrent with central printing, their total cost is at most a few dollars per page, whereas providing traditional reprints costs 4 to 5 times as much. Thus if authors buy e-reprints rather that traditional ones and pay the publisher the same amount, reprint profits will nearly double. If authors buy both types of reprints, profits will nearly triple (Walker 1997a). [Of course, a publisher could choose to make no more on e-reprints than on traditional ones, or it could charge more for e-reprints than for traditional ones, because they are of greater value to the author.]
Authors will buy e-reprints because they are more conveniently and more continuously accessible to potential users and because they dont have to be stored and dispensed. As originally conceived (Walker 1996), electronic reprints were to be available without limit and for all time. While it is possible to block access after a paid-for number of downloads, that is not what authors want and can easily be thwarted by e-mailing the PDF files as attachments or even by unauthorized posting on another server. Posting e-reprints indefinitely is no problem: libraries have WWW servers and are anxious to make the journal literature easily accessible to their clients (FCLA 1997).
Publishers could also make money by selling retrospective e-reprintsi.e., e-reprints of older articles. However, the costs would be higher and the demand much lower than for concurrent e-reprints. Therefore they will probably permit authors to make and post their own e-reprints (which authors can do by scanning pages, combining the images in a computer document, and converting to PDF ). Some publishers may impose an embargo period of up to 5 years, but federal employees will not have to wait, because their publications are, by current law, in the public domain (e.g., Mankin 1997). The Florida Entomological Society has begun posting retrospective e-reprints of articles in its journal as a service to its authors and to entomology. The costs (65¢ per page or about $100 per recent issue) are borne by members sponsoring issues or portions of issues. Societies could use some of the increase in reprint profits to retrospectively put all articles on line. For example, for no more than 20% of the expected increase in its reprint profits, ESA could make PDF files of all published articles and put them on line after whatever embargo period was needed to protect present subscription income (Walker 1997c).
Publishers will profit from selling hyperlinks to authors.
Another way that publishers will garner added income from e-publication during the transition is by selling links in the tables of contents where e-reprints are accessed. Each link will be inserted beneath an authors name and can be to a file that is submitted by the author and posted on the same WWW server as the article, or it can be to any URL that the author submits (Walker 1997d). In either case the author can use the link to publish as much or as little additional material relative to the article as desirede.g., additional data, color photographs, audio clips. If the link is used to jump to an HTML file that the author controls, the author can update the supplemental material at will. Because the author is entirely responsible for any additional material and because no more than one file will be posted with the e-reprint, the cost to the publisher is very low. The Florida Entomological Society has set the price of its AuthorLinks at the price of publishing one page in the Florida Entomologist (presently $45). If six or more authors buy AuthorLinks per issue, the entire cost of putting Florida Entomologist on WWW will be recovered.
Publishers will change the terms of e-reprints as their context changes.
E-reprints will be popular with authors because they will no longer have to store and mail reprints and, in most cases, the supply will never be exhausted. [ESA currently plans to end Internet access to an article after it has been downloaded the number of times an author has paid for.] When all or most authors elect to pay for e-reprints, subscription revenues will be threatened. Should this occur before the publisher wishes to abandon central printing, the threat can be met by various meansfor example, by adding a surcharge to e-reprints to compensate for declining subscription revenues, or by adding the surcharge only to those e-reprints that are posted concurrently rather than one year after publication.
E-reprint charges will become page charges as central printing ceases.
When subscription income is threatenedi.e., when libraries are willing to drop subscriptions to long-acquired journals because their clients no longer insist on photocopying articles from issuesthe time will have come for publishers to cease central publishing and gain a 20 to 60% reduction in publishing costs. At this time they will convert e-reprint charges to page charges. Should this not maintain profits (Fig. 3), they may increase page charges, seek more advertising revenues, and reduce the costs of editing and composing. They must not reduce the quality of editing and composing or else authors will have no reason to spend time and money dealing with them.
Articles will be digitally archived.
With e-publication, researchers will be concerned that their articles are no longer secured for posterity in bound volumes on the shelves of research libraries. Publishers will assuage this concern by consulting with research librarians and following their recommendations for digital archiving. Presently that might be to copy all files to an annual master CD-ROM and distribute copies to libraries throughout the world. For example, the annual journal output of ESA could be digitized and put on one CD-ROM. An annual master CD-ROM could be made for less than $100 on a device costing less than $1000. Copies could be made and distributed for less than $10 each.
Scientific societies will lead the way to free access.
Societies are positioned to lead the way to free access. Firstly, they are ultimately controlled by members, who will see the promise of e-publication and well expect their societies to fulfill it. Secondly, societies are accustomed to much lower profits from their journals than are commercial publishers. Therefore, on the level playing field of all cost-recovery being from page charges, scientific societies should be champions. Thirdly, society members are accustomed to page charges and supporting publishing with their dues.
Other views of the transition
There is a consensus that journals will go all electronic. After all, the current paper medium is difficult to produce, difficult to distribute, difficult to archive, and difficult to duplicate, thus requiring numerous redistribution points in the form of research libraries (Ginsparg 1996). That said, there is no consensus on how it will happen. Below are three views that broaden the possibilities over what Ive described above.
A task force of librarians from three universities (TRLN 1994) proposed that authors insure that their articles can be freely copied for noncommercial purposes by refusing to transfer their copyright to the publisher. This would favor publishers that treated scholarly publications as knowledge to be shared rather than a commodity to be sold at maximum profit and would be an impetus for freely accessible electronic publication. The Association of Research Libraries has discussed but not endorsed this view. One of the articles posted on their Copyright Web site (ARL 1997) discusses whether the employer (rather than the author) owns the copyright to scholarly articles, since they could be considered works for hire, (Shores 1996).
Stevan Harnad (1994), a cognitive psychologist at University of Southampton, and Andrew Odlyzko (1994), an AT&T mathematician, believe that posting preprints is the key to changing the current system. They note that physicists have adopted the practice of posting their submitted manuscripts on a preprint server thereby making the preliminary version of their papers immediately accessible to all who are interested (Taubes 1993, Ginsparg 1996, LANL 1997). Such posting also can establish priority to ideas more firmly than the present system, in which two or three referees have access long before the paper is published (if it is not rejected). The success of the physics e-print archive is illustrated by it serving 35,000 users from over 70 countries, by it processing 70,000 electronic transactions per day, and by the American Physical Society allowing submission of manuscripts to its journals directly from the archive (APS 1997). Harnad and Odlyzko also note that scientists are increasingly posting their preprints on their home pages, and should this become a universal practice, free access to the preliminary versions of all journal articles would be accomplished. Harnad points out that authors could subvert the system further by subsequently substituting the refereed text for the unrefereed preprint text and suggests that publishers might change their policies rather than suing subversive authors. He has recently initiated a CogPrints Archive modeled after the physics archive with the goal of stimulating those in psychology and other disciplines to make their articles as accessible as those in physics already are (CogPrints 1997).
Ken Rouse (1997), a librarian at University of Wisconsin-Madison, suggested that libraries give the money for journals to the appropriate departments and let the departments decide whether paying exorbitant prices for certain commercial journals is the best use of the funds. He also conjectured that major universities might aggressively cancel any commercial journal they could live without and use a small percentage of the savings for launching affordable, alternative journals.
The transition to e-publication of journal articles has begun.
It will quickly accelerate because e-publication is better and cheaper than
the present system. Societies will lead the way to freely accessible
e-publication at a fair price because their members will insist.
Dale Canelas, Michele Crump, Sam Gowan, Stevan Harnad, Andrew Odlyzko, Sarah Sully, and Marcia Tuttle were particularly helpful in providing information and references during the preparation of this article. John Capinera, Stevan Harnad, Gary Noonan, and Andrew Odlyzko, helped greatly by offering constructive suggestions on earlier drafts of this article.
Comments from Stevan Harnad, University of Southhampton, made me realize that I should describe the S/SL/PPV view of the futurei.e., subscriptions, site licenses, and pay-per-view. S/SL/PPV is apparently the goal of commercial publishers and the leadership of some societies. In this scenario, paying for journal publication remains predominantly a function of research libraries (via site licenses, which make the journals available to those with Internet addresses ending in specified domains). For researchers not served by a research library and for researchers whose library cannot afford a particular journal, S/SL/PPV envisions individual subscriptions or a payment to the publisher each time a non-subscriber chooses to view all or part of an e-published article. [Collecting small charges per transaction, as required by pay-per-view, is not yet a familiar protocol on the Internet, but likely will become one in the next few years.] S/SL/PPV requires that authors not distribute electronic versions of their articles, although e-publication and the Internet makes it easy for them to do so. S/SL/PPV prevents the journal literature from becoming a seamless web, where anyone reading an article can jump (without charge) to the full text of any cited reference. (8 Sep 1997)
Shealagh Pope, of Ecological Society of Americas all-electronic journal Conservation Ecology, pointed out a problem with HTML articles that I had not considered: External hyperlinks are likely to be to temporary rather than permanent sites, and thus the links will be dead in a few years or decades. Currently, the only certain way to overcome this problem is to archive the linked materials along with the article that links to them. This can be done with pages posted by the author of the article, but it precludes the author easily correcting, updating, or adding to the materials. (8 Sep 1997)
Dale Canelas recently brought to my attention an e-published conference on Scholarly Communication
and Technology, which included a paper by H. R. Varian (1997), a UC Berkeley dean. In it, he details
possible reductions in the costs of editing and estimates that use of readily available technology
could cut costs by 50% through reduction in clerical labor costs, postage, photocopying, etc.
He also describes how academic publication may evolve after it becomes all electronic. He suggests
that publications will be of unrestricted length but will have the general form of a one-paragraph
abstract, a few-page overview, a 20- to 30-page article, and unlimited appendices. He questions
whether peer review is working appropriately in the current environment and suggests that with e-publication,
submissions might be required to have a 5-page summary that on that basis would be quickly scored
for interest (not correctness) by members of an anonymous editorial board. All papers would be
accepted and posted on the Web unless the authors decided to withdraw them after seeing their
interest ratings. Once posted, a paper could not be withdrawn but would accumulate additional
evaluations, both for interest and correctness, by readers who cared to contribute. Papers deemed
highly interesting would attract the most readers and the most evaluations for correctness.
(12 September 1997)
Varian HR. 1997. The future of electronic journals. Andrew W. Mellon Foundation Conference on Scholarly Communication and Technology (24-25 Apr 1997, Emory University). (http://arl.cni.org/scomm/scat/varian.html and http://arl.cni.org/scomm/scat/index.html)
Thomas J. Walker, an entomologist at University of Florida, began teaching a course in information techniques for incoming graduate students in 1965. When computers migrated to desktops in the early 80s, he decided to keep the course current. He is still trying. In 1994, he helped the Florida Entomological Society put its Florida Entomologist online (http://www.fcla.edu/FlaEnt/). His long-term interest in singing Orthoptera is finding an outlet in a Handbook of Katydids and Crickets.