In 1993, the Publications Committee adopted a strategy for developing AGU's electronic publishing program. This brief overview summarizes some of the key elements of that strategy. Most of the strategy has stood the test of time but other parts have been adapted to reflect changes in technology or experience gained since 1993. This overview also describes some of the projects that are underway or in the planning stages and tries to show how these projects move forward AGU's electronic publishing strategy.

A. Key Elements of the Strategy

1. Foremost, AGU intends to exploit the electronic medium in its publishing activities and not simply replicate on the screen what can be done on paper. This principle is seen most fully in the development of the new electronic journal, but it is also at the heart of all we are doing in electronic publishing.
    
   Exploitation of the medium means providing such features as manipulatable math, 3D and moving images, other enhanced visualization techniques, data in electronic formats that can be imported for use with other computer tools, links to external data files.
    
   
2. Through the electronic medium AGU will permit members to customize their own information packages across all AGU journals.
    
   The nature of AGU as an interdisciplinary organization means that much of what is published is of interest to scientists in different fields. In the paper world, it is difficult to provide packages that cut across the range of articles in ways that are most interesting and valuable to different groups of members. In the electronic medium, individuals can elect to receive certain types of articles on a subscription basis or to receive individual articles on an as-needed basis.
    
   Much of what we are doing today is preparing the way for this end state.
    
   
3. Through the electronic medium AGU can develop a publication base that evolves with time and remains a living, creative set of documents. For example, it will be possible to provide links to comments on a paper or corrections to it. It will also be possible to link to future papers that cite the article. The links can be developed in ways that provide the reader with additional information about papers being cited in an article, such as which ones might use the same experimental technique, which are tutorials or reviews, or the like.
    
   
4. Because electronic publications will be evolving documents that may link to databases that are also changing and evolving, it is likely to be impossible to provide the intellectual richness they contain by fixing them in time and serving them from remote LANs. Mirror sites will be the primary means of providing multiple access around the world. AGU needs to begin experimenting with mirror sites now and will need to develop trusted collaborations for the maintenance of mirror sites around the world.
    
   
5. AGU has a responsibility to the scientific community to develop and maintain a usable archive of all the material we publish in electronic form. This obligation means that we must continually refresh files and must be able to migrate information to new platforms and new media as technology changes. AGU's economic models are taking into account this new expense. One of the action items Council will consider at its next meeting is a proposal to establish a trust fund for the maintenance and migration of AGU's electronic archives. In the electronic world, the expense of publishing entails much more than the costs of producing materials in the first place or the costs of ongoing distribution. Monies must be put aside and invested for generating the operational dollars needed to take care of the electronic archives.
    
   
6. Some forms of information in electronic form will be made available without charge to all. This material will primarily be the kind that furthers AGU's public information goals or that helps to explain or promote AGU programs and services. Other material will be made available as part of the membership fee and will be closed to those who are not members. Still other information will be made available only to those who pay an extra charge, either on some type of subscription basis or on a per-use basis. This type of material is likely to have different fees for members and nonmembers.
    
   
7. AGU is constrained in the approaches it can take in creating electronic publications by the fact that more than 70% of what we publish is provided by authors as camera-ready material. Unlike most other scientific societies and perhaps all commercial publishers, AGU does not currently incur the expense for typesetting most of what we publish. Thus, we cannot afford to move to a system whereby we are paying significant amounts to a vendor to translate author-supplied material into electronic formats. Our plans are based on working with author-supplied electronic files in ways that will not materially increase the cost of either creating an electronic version or producing the paper version, which we must continue to deliver for the foreseeable future.
    
   
8. To control the costs of publishing in the electronic environment, AGU must work with standard formats. Initially, only LaTeX files prepared with AGU macros will be accepted for electronic submission. LaTeX is the only standard available at this time that can be used reliably for math. At the recommendation of the Information Technology Committee, we intend to stay with LaTeX for electronic submission of journal articles until such time as there is a well-accepted, international standard that easily and reliably can handle math across different platforms.
    
   
9. To meet its obligations to the scientific community, AGU intends to control the ownership of the material it publishes and to be the sole distributor of its electronic publications. Arrangements whereby AGU would license others to provide material on its behalf will undermine the economic basis of the publishing activity. The expenses of editing, review, and production of the electronic publication cannot be recovered by the amounts of income that can be contemplated from royalties paid by other distributors for use of AGU-published material. The types of scenarios being put forward by distributors do not provide any income for the generation of the material in the first place or the costs of having that material available over time.

B. Defined Projects Underway or In Planning

1. Earth Interactions is a totally electronic new journal being developed with the American Meteorological Society and the Association of American Geographers. It will publish material that cannot be displayed on a flat printed page; it will contain quick time movies, 3D images, links to external databases, manipulatable math using tools like Mathematica notebooks, and narrations of visual elements. The journal is being developed with a grant from NASA and with technical support provided by NASA. It is planned that elements that are tested with EI and found to be effective for other AGU journals will be introduced in the electronic versions of the standards journals. The first article to pass the review process was accepted in early December.
    
   
2. With GRL Online we are developing an integrated electronic publishing stream from submission through review to distribution in electronic form. The techniques developed and tested for GRL Online will be used for the other journals. The submission scheme prepared for Earth Interactions has been modified to meet the requirements for GRL. Articles prepared in LaTeX using AGU macros are deposited to a password-protected directory assigned to the author. A "robot" developed by AGU staff performs a virus check of the files, translates the LaTeX to the four formats that are made available to the reviewers, stores these files in another directory that is closed to all except the editor, reviewers, editorial support staff, and the headquarters production staff, and assigns a password for the directory. Approved articles that are submitted in electronic form will be available online for subscribers at the same time the camera-ready copy is sent to the printer for the paper version. This means that journal articles will be available at least three weeks and up to six weeks earlier in electronic form than the paper versions, depending on the distance between the subscriber and Richmond, Virginia, where the printed journal mails. Two formats are used for electronic delivery: html for online viewing and two-column pdf for local printing.
    
   
GRL Online was announced on the Web in late November. After a little experience from this low-key announcement, a more aggressive campaign to attract electronic submissions will appear in Eos.
 
Before the fully integrated electronic process is ready for other journals, we need to include the copy editing steps. Some staff members are learning to use the editing software that has been selected for Earth Interactions. The next steps are to determine how this editing software can be integrated into the work flow that begins with the author's LaTeX files.
 

3. WRR legacy files are being translated to SGML format and will be made available via the Web. We are just in the beginning phases of this project. Since WRR is the only research journal that is commercially typeset, it is the only journal for which we have complete volumes in electronic form. With this project we will begin developing the techniques for searching across multiple years. As part of this project we will be developing and testing techniques that will let users obtain single articles on some per-item fee basis. These techniques are necessary for meeting the expectation that members will be able to customize their information packages. We also will be developing the forward-referencing capabilities for these articles so that it is possible to link to the articles that cite a particular paper as well as to link to comments on or corrections to a given paper. Back files that can be reliably translated into electronic form exist back to the 1990 issues. The first articles will appear online about March 1997. At that time, the 1997 issues will also be published electronically concurrent with the print publication.
    
   
4. A significant part of customizing information packages is the availability of good alerting tools. A few simple steps have been taken to date: (1) providing on diskette a bibliographic database for all AGU journals, books, translations, and Eos articles from 1988-1995; (2) providing this same information on a monthly subscription basis for 1996 and forward; (3) putting the tables of contents on the Web for the three most recent issues of each periodical; (4) having authors provide the GAP abstract of journal articles in electronic format rather than paper and linking these to the contents pages on the Web. Several different types of alerting services can be developed from this information base, including the ability for members to be alerted to articles that match an interest profile they would create. The first such service will be to deliver selected tables of contents to individual members via e-mail on some type of subscription basis. This service will be further developed to provide members an alert of articles that match a user-defined profile.
    
   
5. Readers may now access supplementary information in electronic form. Small data sets for journal articles are available through AGU's ftp site. About 70 such datasets are now available. This electronic service can also be used to publish more color figures and thereby reduce the costs to authors for the color. Certain software-related articles are now published in an electronic supplement to Eos. It is possible to provide more of these articles, attach updated information, and include more detail than was possible through the limited pages of Eos. Both types of supplementary material can extend the information content of the research journals and help to meet the goals set for exploiting the new technology.
    
   
6. Getting time-sensitive material published by AGU into the hands of members more quickly is also a goal for AGU's electronic publishing activities. Meeting abstracts that are submitted electronically are now distributed through the Web well in advance of the printed abstract volumes; starting with the 1996 Fall Meeting they are linked to the detailed program. The full program for the meeting is online soon after the program committee completes its work. We are in the process of creating a fully integrated system for handling the electronic submitted meeting abstracts; this new system will replace the combination of miscellaneous programs and very labor-intensive steps that are now required to manage and publish the meeting abstracts. Job announcements, open to members only, are now online; this is the first of the "members only" services to be provided. The membership directory will be an online service for members in early 1997. Also in 1997 members will be able to access and correct the demographic information in their AGU membership record. Ways to provide information to all authors about the status of their papers is also being planned as part of a new manuscript tracking system.

C. Planning for Electronic Publishing Activities

1. To improve the timeliness, economy and usefulness of traditional print publications by employing electronic techniques judiciously throughout the publishing and distribution process.
    
   
2. To make AGU's publications a more attractive place for authors to submit outstanding work, particularly by improving the ease of submission and publication, the timeliness of publication, and the utility of AGU's vehicles for communicating information.
    
   
3. To make the information within individual articles more accessible and valuable by providing the means for readers to interact with and extend what was contained in someone else's paper in ways that are impossible or impractical in other media.
    
   
4. To improve an individual researcher's ability to find and retrieve relevant items from the scientific literature.
    
   
5. To enhance an individual's ability to have personal copies of publications or personal access to individual articles and full publications particularly in light of shrinking shelf space for journals.