In the past 20 years, Annals has continued to grow as an independent general journal in a sea of ever-more-subspecialized publications. Now published twice monthly, its readership stands at over 96 000; it costs more than $100 for a year's subscription; and it receives more than 2400 manuscripts a year, one third from outside the United States-all surely signs of our times. Annals continues to publish the highest-quality clinical science balanced with a broad mix of material on the social, economic, and human dimensions of the specialty. The journal maintains a tradition of leadership in the analysis and reporting of medical information: systematic reviews [2, 3], structured abstracts [4], the use of statistics [5-8], peer review [9], and standards for reporting randomized trials [10].

Anniversaries are occasions for looking back but also for looking to the future. For medical publishing, that future has a distinctly electronic shape. Along with most other journals, we have for some time been pondering our place in that unstable but exhilarating electronic world. As early as the 1980s, the full text of Annals was available on-line, but in a system that now seems impossibly archaic. We've learned a lot since those ancient days. Annals content is now available on CD-ROM, and for more than a year, selected full-text Annals articles, complete with graphics, have been available on the Internet at the American College of Physicians Web site. Described as "one of the better sites on the Web" [11], it can be reached at http://www.acponline.org, and we invite readers to look at Annals there (readers can also go directly to the table of contents of the most recent issue at http://www.acponline.org/journals/annals/annaltoc.htm).

Medical readers still get most of their information in print, and we see a central role for print publication for a long time to come. (The introduction of fax and e-mail did not, after all, eliminate the written letter, the telephone, or the face-to-face meeting.) This devotion to print seems to be less a reflection of inherent physician conservatism or Luddism with respect to computers than of print's unique assets: simplicity, readability, portability, stability. Electronic systems, in contrast, provide new and unimagined power of a different kind: speed, searchability, structure, storage capacity, flexibility. Readers are therefore likely to continue using print for browsing text, reading and reflecting on it, broadening and deepening their understanding, creating new and improved mental models-in brief, learning. Electronic systems, in contrast, will be used for searching and manipulating text, filling in the blanks in robust mental models, communicating with others-in brief, transferring chunks of information. And in the ideal world of the future, the two will complement each other to meet individual physicians' information needs.

Although it is important to make journal information available electronically, as we've begun to do, simply producing "shovelware" (that is, reproducing existing printed text and graphics on a screen) does not allow either print or electronic media to realize its full potential. By way of contrast, Index Medicus, MEDLINE, and the structured abstract all had transforming effects on use of medical information; they did so by adding intelligence to the structure of existing text but without changing content. For all their strengths, however, when it comes to locating and appraising medical information, those innovations still leave a great deal to be desired. Therefore, together with Drs. Valerie Florance and Gretchen Purcell, information scientists with respective backgrounds in library science and medicine, respectively, we have been working on new ways to add meaning to the structure of existing text. The result so far has been two varieties of "clinical extract": smarter views, we think, of electronic journal information. Extracts for selected recent Annals articles are now available as a demonstration project on our Web site, at http://www.acponline.org/journals/annals/extracts/. We urge readers to use them and are eager to know whether they are in fact helpful. Other elements of an electronic Annals are on the drawing board and will be rolled out as they develop.

Seventy years from now, we will probably look back on today's seemingly sophisticated electronic medical publishing and realize it was still in its infancy. We may then find ourselves reflecting on the similarity between the technological history of electronic publishing and that of other modern technologies. Air travel, for example, needed a full 30 years and five major technical advances to move from the Wright brothers' invention to the innovation of the first successful commercial aviation [12]. Likewise, it was 40 to 50 years between the invention of the dynamo and the time when electrical power had its full effect on productivity, as measured by the conventional indicators [13]. And 70 years from now, being so much wiser, we may have even rediscovered that it's the content that matters, after all.

Frank Davidoff, MD

Editor