EPUBs are an experimental feature, and may not work in all readers.

The Challenge

"The problem with you publishers is you think you add value. Well, you don't. You force me to reduce the information in my scientific papers so that they will fit on a flat printed page." With those words, or ones very like them, hydrologist James Wallis, then of IBM Watson Labs., threw down the gauntlet at a Top Management Roundtable of the Society for Scholarly Publishing. For one area of science that stinging criticism has now been rebuffed by the online journal Earth Interactions, which began publication January 1997.

From the first conceptual discussions, the founders of Earth Interactions shared a common goal: to remove the limitations of the printed page while delivering a high-quality, peer-reviewed scientific journal that permits scientists to present their results in the most useful manner possible. The subject matter of Earth Interactions begs for this approach. The journal serves researchers working in the area called Earth-systems sciences. That area draws together university, government, and industry researchers in the atmospheric, oceanic, hydrologic, solid-Earth, and biological sciences. The problems they study are multi-disciplinary and tend to focus on issues on a global scale. Those researchers routinely use sophisticated computer graphics to analyze their data. For example, potential authors for the journal may create animations of numerical simulations to help visualize the changes in variables like sea surface temperature, cloud cover, or vegetation, over periods of days, years, or decades. Their data are collected by a wide range of instruments from ocean-bottom seismometers to sensors on Earth-orbiting satellites, and they may work with datasets in the terrabyte range. Those researchers use interactive plotting systems to create three-dimensional images that can be rotated to different viewing angles so that the complete structure can be better understood. At scientific meetings they use videos or computer programs to share their findings with colleagues. But when they were ready to publish their results, they were forced to forego the dynamic visualizations and try to convey their information through what the static page permitted. At least that was the difficult choice they had before Earth Interactions.

Today, geophysicists, geochemists, and geographers have an online, peer-reviewed journal that gives them a wide range of possibilities for transcending the printed page. To meet their publication needs requires a journal that is delivered only in electronic form and that is supported by a fully electronic peer-review process.

The Collaboration

Earth Interactions is the joint publication of three learned societies with long traditions of publishing high-quality scientific journals: the American Geophysical Union, the American Meteorological Society, and the Association of American Geographers. Two others, the Oceanography Society and the Ecological Society of America, participated in the planning stages.

Since the subject cuts across the interests of the three copublishers, the journal is stronger scientifically because of that collaborative effort. Together the three cover the interdisciplinary nature of the subject more fully, and thereby provide a broader editorial scope, than any one of the societies could do alone.

The partners bring different perspectives, talent, and resources to the project, and so the publishing experience is also richer. The combination has been particularly valuable when facing the uncharted territory of a multimedia online journal. Sharing the experimentation has also helped share the risk.

Our Online Journal: What Hasn't Changed

The peer review standards of Earth Interactions are as demanding as those of our print journals. We expect to attract high-quality contributions from excellent scientists, and we expect publication in this journal to have a positive impact on promotion and tenure considerations. Despite the eagerness to start publication, there has been no skimping on the review or revision stages for even the earliest contributions, some of which were rejected because they did not meet scientific standards. As one editor said, "Earth Interactions is not a journal for glitzy graphics without substance." In addition to the rigorous peer review, articles undergo full copy and technical editing to ensure a uniform style and a quality consistent with an archive journal.

"Readers can interact with parts of the author's data and observations."

Our Online Journal: What Is Really Different

From the beginning, we were determined to have a journal that would do much more than reproduce electronically what could be printed on the page. Our goal has been to exploit the medium and go beyond the capabilities of the printed page. Earth Interactions authors are encouraged to include sophisticated graphics, data in electronic formats, and even useable computer code — the same tools they use to draw their conclusions. Readers can interact with parts of the author's data and observations and thus more readily verify and expand the results presented.

Moving Beyond the Printed Page

The best way to get the feel of what the journal is offering the research community is to go to the journal's Web site and try the Features button. The article you will encounter was prepared to help potential authors understand the capabilities of this electronic journal.

Because Earth Interactions is disseminated via the Web, the formats currently allowed are limited to those for which Web-capable viewers exist. As the availability of reliable viewers expand, we will extend the allowed journal formats.

  • Animation and virtual reality: Our researchers use animated graphics to show changes in both observed and modeled phenomena and are beginning to use virtual reality displays (Wheless et al. 1995). The journal currently supports only MPEG and Quick Time to display image loops and animations because Web viewers are widely available for these formats. It also supports the VRML format for authors who want to publish virtual reality displays. Authors can even narrate those features so that the reader is not forced to jump back and forth between the graphics and the text of the article. Earth Interactions may be the first scientific journal that will need to include pointers on diction in its information to authors.

  • Datasets: Earth Interactions permits small datasets to be incorporated directly into an article, and provides active links to external data-archive facilities that house larger datasets. The data may be in a form that may be directly ingested by analysis packages for further study. To assure that the externally housed data are available for future readers, authors must store their data with facilities that have a mandate for its long-term care.

  • "Live math" and numerical code: "Live math" refers to equations presented in a symbolic form that can be ingested by a mathematical analysis routine (such as Mathematica or Matlab) and through that facility manipulated interactively. For example, rather than displaying only a static graph of an equation with a particular parameter set to a few common values, the equation and parameters could be held as a live-math set so that a reader could produce graphs for values he or she chooses to enter. Earth Interactions currently supports the inclusion of Mathematica Notebooks. Authors may also provide the numerical code for their models and thereby share that kind of analysis tool with the readers.

  • Interactive three-dimensional display: Software already exists that allows true interactive 3-D display (so that a reader can rotate a 3-D object and view it from any desired angle). The standards for Web-capable viewers are yet not firmly established. Thus, while Earth Interactions expects to include those sorts of displays in the near future, it does not at this time.

  • Forward references, linked comments and replies, and corrigenda: Through "forward references" an electronically published article can become a living document. New references to later works can be added to an article continuously; thus, readers can see how an article has influenced later work. To create forward references in an exhaustive fashion is a monumental and exponentially growing task, requiring a database of a scope comparable to that used to create Science Citation Index. To keep our task manageable, initially we will create this type of forward reference link only between articles published in Earth Interactions.

    While the temptation may be strong to adjust text, equations, or figures after formal electronic publication if an error is noticed, it is critical for scientists using published results to know if a correction has been made and when that correction was introduced into the literature. With the exception of updating active links if needed, Earth Interactions will not alter articles after the official date of publication. Corrigenda properly dated will be added to an article and flagged at the appropriate point in the body so that no one reading the article after the corrigenda is placed will miss the fact that a correction has been made. Similarly, comments and replies, which could be critical to proper interpretation of the work, will be linked directly to the article.

  • Other value-added features: Earth Interactions takes advantage of the electronic form to include a variety of other value-added features not possible in print journals. Those include navigational aids such as internal hyperlinks between a mention in text and the corresponding figure, table, or equation, and from bibliographic citations in text to the reference listing at the end of the article. Panorama, the journal's recommended viewer, provides additional navigation aids by displaying an outline of the article in a separate navigation window through which the reader moves easily to different sections. Figures are presented in the text as thumbnails, but can be expanded to full resolution with a mouse click. Journals most frequently cited in Earth Interactions are covered by Meterological and Geoastrophysical Abstracts (MGA), and the MGA abstract is linked to the corresponding citation.

An Electronic Peer-Review System

Because Earth Interactions takes advantage of special electronic formats, we have exploited the device independence of Web browsers and other viewing software to support the review process. The task of matching each author's preferred format for submission with the appropriate reviewer's software capabilities would have been impractical. Authors submit the text and tabular portions of their articles as PostScript or encapsulated PostScript files, and submit the graphical material in a form that can be viewed by any of the standard Web external viewers (MPEG, JPEG, GIF, EPS, etc.). We convert the text to the Adobe Acrobat PDF format, which can be viewed with any Web browser using the freely available Acrobat Reader. We use a simple HTML structure to link the text and graphic or special format files. We also insert links to help the reviewer obtain any additional Web-browser aids needed for that article. The result is a coherent (though not very fancy) document that gives the reviewer access to the substance of the article and provides some feel for how the special features will be incorporated in the final publication. It is important for the editor to have the reviewer's comments on the usefulness of the special features as well as on the scientific merits of the article.

"We believe that ensuring the availability of authenticated copies of peer-reviewed scientific articles for future scientists requires an income stream from the users."

Throughout the review process, the article is held in a password-protected directory accessible only to the reviewers, the editor, and the editor's assistant. A few authors have argued that reviewers could simply be directed to the author's Web site to study the article. Under that scenario the journal would have no control over an author's natural tendency to make minor improvements, and therefore the editor may not know which version the reviewers had seen or which version the editor had accepted. Even the anonymity of the review process could be compromised, since the author could identify the addresses of those who accessed the article during the review period.

SGML for Delivery

Like many publishers, we chose SGML as our archival format, believing it will give us the best option for transporting the journal forward as technology changes. We also chose SGML rather than HTML or PDF as our format for delivery because SGML provides much greater control over the presentation, ensures integrity of the displayed information, and takes advantage of the available technology to display mathematics and special characters efficiently as coded text rather than as a series of embedded graphics files. A comparison of some of the characteristics of HTML, PDF, and SGML is given in Table 1 (Sears, 1997).

Table 1: Attributes of HTML, PDF, and SGML for electronic publishing
Attribute HTML PDF SGML
Markup ASCII-based, easy to code, good editing tools, e.g. HoTMetaL, HotDog, or any wordprocessor Easy to create from PostScript or typeset files or scanned images of printed pages Complex: requires expertise and expense to set up; expensive tools; steep learning curve
Cost Inexpensive for simple applications; investment increases with complexity and functionality Economical Expensive
Portability Portable: Standard for the Web on all platforms; free browsers Easily distributed on Web: free viewer available Portable but Web viewer not yet available for all platforms
File size Small: ASCII markup + content Smaller files than Postscript, but still quite large Small: ASCII markup + content
Web features Supports hypertext with tables, lists, interactive forms, graphics, animation, etc. Adheres exactly to page layout User-defined tags and architecture (DTD) allow rich document structure; better suited to longer complex documents with long life-cycle
Hyperlinking Yes Yes Yes
Format User has limited control over format (fonts, background..); printed document is usually large with lots of white space Excellent print quality Screen/print display controlled through (multiple) style sheets; user can modify; no two-column styles as yet
Access Cross-platform, standard-based core syntax, but proprietary extensions lead to lack of stability; not user-defined Proprietary Cross-platform; long-term viability through open standard
Layout Flexible Inflexible; tied to numbered, printed page Flexible
Tagging Limited tag set that defines appearance rather than structure/content No hierarchical structure/document elements Open-ended, user-defined tag set; structure/database- oriented; context-sensitive
Speed Small files load quickly in browser Large files slow to download (but new streaming plug-in speeds viewing somewhat) Small files, but loading takes a little while to interpret "support" files (DTD, style sheet, navigator ...)

Through an agreement with SoftQuad, subscribers to Earth Interactions are provided a commercial copy of the Panorama viewer. Unfortunately, Panorama is currently available only for Windows platforms, though Mac and UNIX versions are due soon. As an interim measure, the articles are also being provided in a less functional PDF form with nonprintable graphics and interactive content linked through an HTML page. That interim presentation is clearly inferior to the full SGML delivery, and will be dropped as soon as Mac and UNIX versions of Panorama (or some other SGML browser) are available.

Some Aspects of the Economics of Earth Interactions

The copublishing societies are determined that Earth Interactions be structured and managed so that it will be financially self-supporting (even though outside funding from NASA's Mission to Planet Earth program is being used during the start-up phase). At least two years ago, most of the electronic-only journals were offered free to subscribers (Okerson 1995) and were produced with some level of subsidized support. The debate on the economics of electronic journals compared to print ones continues to rage (e.g., Okerson and O'Donnell 1995). We believe that ensuring the availability of authenticated copies of peer-reviewed scientific articles for future scientists requires an income stream from the users. Furthermore, the marketplace will help to sort out the useful from the chaff. Our print journals cannot subsidize the production of an electronic-only journal on a continuing basis, especially as we move the traditional journals to some form of electronic delivery.

Our evidence to date proves that a high-quality scientific journal like Earth Interactions will not be less expensive to produce than a comparable print one. The review process still requires staff support. Preparing the electronic files for the reviewer more than offsets the savings in postage or express packages. The expenses associated with structuring and formatting the text and special files and the time required to introduce intelligent hypertext links adds to the cost of copy editing. The journal will also incur new but continuing expenses associated with producing forward references, maintaining active links, upgrading files and operating software to state-of-the-art technology for access, storage and delivery of previously published material.

The income model for Earth Interactions is quite similar to the model used for our print journals. Revenue will be derived from both author charges and subscription fees. That combination provides for the broadest possible dissemination. Funding from NASA is allowing us to waive author charges and subsidize subscription fees as we are getting the journal started. We are gathering cost information to use in establishing fair charges and fees to reach our economic goals.

"We learned that we could not truly test the system in an alpha or beta mode."

Subscriptions are an important component of the economics of print journals that allows the author charges to be kept reasonable. Subscription revenue also allows societies to publish work even when the author's institution does not have funds available to cover the author charges, and the collaborating societies intend to continue that policy with this journal.

The subscription model being planned seeks to emulate the best features of the print model and address some of the concerns of both individual and institutional subscribers. The subscription will be a flat fee for a volume year of the journal, with a fee structure that provides for the lowest cost to members of the five societies in the collaboration, somewhat higher fees for individuals who are not members, and still higher (and graduated) fees for institutional subscribers that provide the journal to patrons on an institutional network. Since the subscription provides access to an electronic source, the subscriptions take on more of the character of individual and site (institutional) licenses under which the subscribers agree to adhere to the terms of their particular subscription type (including copyright restrictions).

An important component of the proposed subscription model is finding a way to permit subscribers for a given year to have access to that material on an ongoing basis. That addresses the concerns of both librarians and individuals who do not want to be in the position of having to download the contents of an electronic journal during the subscription year in order to have access to it at a later time. Subscribers who drop their subscriptions will, of course, miss out on the upgraded features and the opportunity of linking to added information.

In Conclusion: Some Other Things We Did Right and Some Surprises We Had

In designing this bold adventure in scientific publication we concentrated on establishing the principles. We considered what was most valued in the print journals and determined what the counterparts would be in the electronic-only mode. We agreed that we would choose the specific technologies to be employed only when we could no longer put off that decision. That approach would let us use the most current electronic tools and save on software development.

Most importantly, we engaged the potential authors and readers throughout the planning phases. This is, of course, the common approach in society activities. But in this case we were able to do some wide-ranging exploration of what the scientists wanted to see and how they expected to interact with the material. Some of these ideas are still awaiting the right technological solutions.

We learned that we could not truly test the system in an alpha or beta mode. We needed live articles because existing published material did not contain the graphics and interactive features key to the journal's mission. Luckily, all of the early authors have enjoyed the experimentation and been willing to work with us in shaking down the systems.

Some of the younger authors were more conservative than we expected. Perhaps because they are still establishing their credentials, it is the younger set who has questioned the decision not to have a print byproduct of Earth Interactions.

We encourage readers to visit the Earth Interactions Web site and see for themselves a journal that goes beyond the printed page.



Judy C. Holoviak, director of publications for the American Geophysical Union, has been engaged in society publishing for more than 30 years, joining the AGU staff almost directly after receiving her B.A. from Westminster College, Salt Lake City, Utah (1963). She was a founder of the Society for Scholarly Publishing and its president 1990-1992; she has also been president of the Association of Earth Science Editors. One of the most unusual aspects of Judy's career is having an Antarctic glacier named for her in recognition of her publishing service.

Keith L. Seitter, associate executive director for the American Meteorological Society, joined the society staff in 1991. After receiving his Ph.D. in geophysical sciences from the University of Chicago (1982), he served on the faculty of the University of Massachusetts at Lowell. At AMS he is responsible for the society's publications and is primarily in charge at its historic Boston headquarters. He has been a frequent speaker at the Council of Biology Editors, the Society for Scholarly Publishing, and various other groups concerned with electronic publication. Keith is married to Julie, whose voice welcomes visitors with audio capability to the Web site of Earth Interactions.


References

Okerson, A., ed., 1995. Directory of Electronic Journals, Newsletters and Academic Discussion Lists. 5th ed. Association of Research Libraries.

——, and J. J. O'Donnell, 1995. Scholarly Journals at the Crossroads: A Subversive Proposal for Electronic Publishing. Association of Research Libraries.

Sears, J.R.L, 1997. Earth Interactions: An electronic journal. Paper presented at Joint Assembly of the International Association of Meteorology and Atmospheric Sciences and the International Association for the Physical Sciences of the Ocean, Melbourne, Australia.

Wheless, G.H., A. Valle-Levinson, and W. Sherman, 1995. Virtual reality in oceanography. Oceanography 8:52-58.