Four Editions and Eight Publishers – A History of the Encyclopedia of Computer Science




by Anthony Ralston




This paper appeared in the Annals of the History of Computing, January-March 2004, pp. 42-52.



"No man but a blockhead ever wrote except for money."

Samuel Johnson, 1776

I. The Beginning

It must have been early in 1971 that Isaac (Ike) Auerbach phoned me to suggest that I edit an encyclopedia of computer science that he wanted Auerbach Books to publish. I asked myself: Was the discipline of computer science ripe for an encyclopedia? After all, it was less than 10 years after the establishment of the first university departments of computer science. After considerable thought, I convinced myself that the time was ripe and the need was there.

That the eldest of my four children was only a few years from college played, I think, no part in my decision. Still, I knew from previous experience both writing and editing books that editing books could be as or more lucrative than writing them. (In retrospect I'm not sure that, based on a return per hour spent on the project, this was a valid calculation for the Encyclopedia of Computer Science.) Anyhow on 8 July 1971 I signed a contract with Auerbach Publishers (Publisher #1) to edit an encyclopedia of computer science that committed me to deliver a manuscript "of approximately 2,400 double-spaced typewritten pages … not later than January 1973". (Remember – this was well before the date of the widespread availability of word processors.)

Publishers almost always put overoptimistic delivery dates for manuscripts in their contracts to put some pressure on authors and editors and almost never cancel a contract because of late delivery. And perhaps, having never edited an encyclopedia before, I can be forgiven for not realizing just how much work it was going to be. (Less forgivable is my overoptimism about delivery dates for the next three editions when I should have known better.) Anyhow in the summer of 1971 I got to work. Since I was chair of my department at the State University of New York at Buffalo (SUNY/Buffalo) at the time, I had good secretarial support and also I had good support from an excellent editor at Auerbach, William Fallon.

II. The First Edition

First, I inveigled Chester Meek, Assistant Director of the Computer Center at SUNY/Buffalo, to serve as my Assistant Editor since I at least was wise enough to know that this was going to be a big job. I couldn't have made a better choice. I had hired Chet when I was Director of the Computer Center and, as I knew he would, he proved indefatigable, conscientious and unfailingly helpful.

Then I assembled an Editorial Board (see Table 1) of leading computer scientists to provide quality control and to help me construct a list of articles and to find authors for these. To persuade people to join the Editorial Board I prepared a two-page prospectus that stated the purpose of the encyclopedia as:

1. To provide a comprehensive basic reference to the subject matter of computer science, and
2. To provide a broad picture of computer science through survey and essay articles on a considerable number of subjects.

The prospectus estimated the size would be "about 1000 pages" (bad guess; it turned out to be 1523) with about 650 articles (another bad guess but in the other direction; the first edition had 470 articles).

The first task, pretty obviously, was to draw up a list of articles with the proposed length of each. The list I sent to the Editorial Board for comments listed 481 articles which argues that I didn't believe my own estimate in the prospectus or, more likely, that I expected many additional articles to be suggested, which didn't happen. (Table 2 contains some vital statistics for all four editions.)

The list of articles was organized into nine main categories – History; Programming Languages; Software; Hardware; Software-Hardware (computer systems, really); General Terminology; Mathematics of Computing; Disciplines in Computer Science; and Applications. Most articles were categorized where you would expect them but there are, in retrospect, some surprises. For example, Automata Theory and Turing Machines both fall under Disciplines together with Artificial Intelligence, Information Retrieval and Numerical Analysis.

In the Classification of Articles in the first edition, these nine main categories were transformed into ten top-level categories. This Classification, really a taxonomy of the discipline, has been a feature of all four editions (and became the basis of the Taxonomy published by the American Federation of Information Processing Societies in 19801). The top-level categories in the second edition were somewhat different from those in the first but then were unchanged in the third and fourth editions (see Table 3). The Classification is, of course, a tree with each branch having at most four nodes (including the top-level node).

In addition to the articles, the first edition had two appendices (see Table 4) and, of course, it had an index. Since I have always believed that any technical book needs as extensive an index as possible, the Index in the first edition was 46 pages long with about 5,000 entries. Each entry was entered onto an index card (remember, this was 1975) and I can still recall alphabetizing them with my wife and four children (aged 8 to 16) on the floor of my den at home.

As you would expect, the first edition proved the hardest of the four to assemble. Much of the work of getting a more or less final list of articles, getting names of potential authors and soliciting them to write articles was done while I was on sabbatical in London in 1971-72 with much of the necessary secretarial support provided by Auerbach Publishers so that, when I returned to Buffalo in the summer of 1972, the project was well under way.

Identifying potential authors for articles was seldom a problem. It's hard to realize this now but in the early 1970s, computer science was still a small enough discipline so that you could know almost everyone engaged in any kind of scholarly work in computer science. And I did know just about all of them since from 1970-72 I was Vice-President of ACM and then, from 1972-74, I was President. Indeed, I think my position in ACM was one of the main reasons Ike Auerbach thought of me as the editor of the encyclopedia.

But knowing (almost) everyone was one thing; persuading them to write articles for you was quite another. Actually I had very good cooperation from my friends and colleagues in spite of the tiny honorarium they were offered (2 cents/word and the opportunity (!) to buy a copy of the encyclopedia at a 50% discount). Many saw writing an article for the encyclopedia as the kind of professional obligation that scientists and scholars have traditionally accepted even if paid no honorarium at all such as when editing journals or refereeing articles for journals. But computer science then had – and probably still has – a larger share of people than other sciences who do not recognize such an obligation. Some declined my invitation on the (no doubt valid) basis that they were just overcommitted; others said without any sense of embarrassment that the honorarium was just too small. All this necessitated in a number of cases having to ask two or more people before I received an agreement to write an article. The phone calls, the wheedling, the empty promises from authors are surely the most distasteful aspects of editing a book or a journal.

By mid-1973, although the January 1 deadline had come and gone without comment, the project seemed well on its way to completion. Silly thought. Only then did the problems really begin. Returning from a trip to visit me in Buffalo, the Auerbach editor I was then dealing with (no longer Bill Fallon) actually lost a large portion of the manuscript (about 100 articles) and the accompanying artwork. It took some months to recover from this disaster.

Then in late 1973 Ike Auerbach decided that he wished to focus entirely on the main portion of his business, providing a loose-leaf data processing advice service to business. So he divested himself of Auerbach Books by selling it to the publisher of Auerbach Books, Orlando Petrocelli, who financed the deal by joining Mason/Charter Publishers as the publisher of their computer books. So Petrocelli/Charter (Publisher #2), the actual publisher of the first edition, was born. But sadly for me, the same person who had lost the manuscript somewhere between Buffalo and Philadelphia was now Olie Petrocelli's right hand woman. And she was a control freak. No, the editor shall not speak to the copy editor. No, he shall not speak to the artist. No, production will not start until 100% of the manuscript is in hand. I railed against these restrictions with some but not much effect until the first edition was published in June 19766. On the day it was published, she was fired. Too late! – at least for me.

The first edition, like all its successors was a critical success with nearly all reviewers. This is not to say there wasn't some valid criticism, much of it related to subjects that might have been covered but weren't (e.g. online databases), as well as some listing of the inevitable typos. (The single most embarrassing error, not, I think, discovered by any reviewer, was a figure that purported to show responses at various sites in the brain to a pharmacological agent but was actually a map of Los Angeles census data.)

One footnote about the first edition: It was published before there were widespread networks but there were local networks with many terminals, which had been pioneered by Project MAC at MIT. Shortly after the publication of the first edition, John McCarthy asked me how I felt about making the Encyclopedia available on the Stanford network. I was enthusiastic and raised the matter with Olie Petrocelli. He was interested but soon became negative on the grounds that his fellow publishers would read him out of the fraternity if he allowed this, so worried at the time were publishers about any online access to books.

III. The Second Edition

Just about any encyclopedia becomes obsolescent in the years after publication and, if not revised, will, before too long, become obsolete. With computer science, however, an encyclopedia is sure to be obsolescent when it is published given the rapid changes in the discipline and the fact that almost all articles will have been written at least a year before publication and two or three years previously for some.

When the first edition was published, my nave hope had been that the pace of change in computing would slow down so that a new edition would not be needed for some years. However, it soon became clear that the first edition was rapidly heading for terminal obsolescence. Did I want to do another edition? Well, yes, not least because my eldest child had entered university in the fall of 1977. So on 4 October 1978 I signed a contract for the second edition of the Encyclopedia. But not with Petrocelli/Charter which by that time was no more. Not long after publication of the first edition, Petrocelli/Charter's list had been sold to Van Nostrand Reinhold (Publisher #3) and so it was with VNR that I signed the second edition contract. Having VNR as publisher was fine with me; more than either of its two predecessors, it had a reputation as a quality publisher, well known throughout the technical book world.

Since Chet Meek had left Buffalo before the publication of the first edition, I felt I needed a new collaborator. I persuaded Edwin D. Reilly to become Associate Editor of the second edition. Ed had been a friend and colleague since 1965 when he became Director of the Computer Center at SUNY/Albany at the same time that I became director at Buffalo. I knew him not just as someone very knowledgeable about computer science broadly but also as a hard-working, conscientious colleague who would add much strength to the editorial team for the second edition. I also recruited a new Editorial Board, six of whose nine members had also been on the first edition board (see Table 1).

The contract for the second edition called for a book that was "not to exceed 1600 pages" and whose manuscript would be delivered "on or before the 29th day of February 1980". I had learned something from the first edition (the second edition was actually 1664 pages) but not enough. Once again the delivery date was wildly optimistic since the book was not published until October 19826 about a year after the final manuscript was delivered.

At least this time I knew the drill: list of articles, find authors, get manuscripts in, etc. Many of the articles in the second edition were, naturally, updated versions of first edition articles although some 40 were rewritten and others were extensively modified. One problem was finding the authors of first edition articles so that they could be invited to update their articles. Given the peripatetic nature of computer scientists, this was not always trivial. In the end, the second edition had 550 articles of which 90 were new. Only 18 articles in the first edition were dropped in the second, including one idiosyncratic one, entitled Dangling ELSE, that I wrote myself and which had the following sexist, ageist example:

if SEX = FEMALE then if AGE = 30 then RECONSIDER else PURSUE.

The top-level categories in the Classification of Articles added two new categories, Information and Data, and Methodologies, the latter mainly a list of subdisciplines of computer science such as artificial intelligence and image processing. In addition, four of the categories in the first edition taxonomy were combined into a single Computing Milieux category (see Table 3). In this edition there were six appendices (see Table 4). The Index was 50 pages long and had about 5500 entries.

Perhaps because I was dealing with a larger and more established publisher than with the first edition, the production of the second edition was much smoother (and less stressful) than with the first edition. Also I was blessed with an excellent editor at VNR, Althea Gordon, who was accurate as well as being unfailingly polite and helpful. Still nothing is easy with publishers. During the time the Encyclopedia was housed at VNR, the company went through three presidents and, in an attempt to save money, bought some paper stock cheaply for the second edition that was too narrow (from the top to the bottom of the page). But it was used anyhow. The result was that, in many copies of the first printing, there were pages that had the same length text columns as in the first edition but that were a full inch shorter than the first edition pages. Thus, upper and lower margins were far too small. Indeed, I have a copy where some pages have bottom margins of only 3/16 inch. This was embarrassing to all but, if the truth be told, more to the publisher than to me. I still don't understand how this could have been allowed to happen.

A reviewer of the first edition2 remarked that a better title would have been "Encyclopedia of Computer Science and Technology". In effect, we agreed and so the title of the second edition was "Encyclopedia of Computer Science and Engineering", with "engineering" felt to be as accurate as "technology" and more appropriate given the emerging discipline of Computer Engineering (a term that is, however, not in the index of the second edition).

The second edition was the most successful of the four editions as measured by total sales. In 1985 it was number 18 on a list of the top 100 books held by member libraries of the Online Computer Library Center. The first three editions all had comparable sales through normal channels (see Table 2). But only the second edition was published at the height of the popularity of scientific book clubs which, given the availability of so much online now, have declined considerably in importance. The second edition remained for some time as one of the almost free books you could get by becoming a subscriber to the Library of Computer and Information Science. Under its auspices some 100,000 copies were distributed, a more precise number being impossible to ascertain. Particularly because of the book club sales, the second edition became, I was told, only the seventh book in Van Nostrand's history to achieve one million dollars in sales.

The second edition was also generally well-reviewed but it received its single worst review from Bob Patrick in Datamation5 who complained that it was unsuitable both for the executive unfamiliar with computing or for the young and uneducated. Since I had always wished to produce an encyclopedia, most of which would be accessible to the bright high school student, this criticism hurt but not as much as the reason Patrick ascribed it to: poor editing.

IV. The Third Edition

It looked as if Van Nostrand would be the publisher of the Encyclopedia for a considerable time, particularly so when Ed Reilly, this time as co- editor, and I signed a contract to compile the third edition on 16 March 1987. This was four and a half years after publication of the second edition, surely later than it should have been since major obsolescence of the second edition was looming, given the continuing breakneck pace of change in computing. This contract called for the delivery of the manuscript by "the first day of September 1989". I had learned something, since this contract had, at least, 2 years from signing to manuscript due date compared to 1 years for the second edition, but not enough. The third edition was not published until November 19926, ten years – and far too long – after the second edition. The contract called for a book of between 1700 and 1800 pages, predicated on the same page size as the first two editions. The actual book had only 1558 pages but since there were about an eighth more words per page (on a larger page size) in the third edition, we were in fact pretty much on target. The third edition for the first time had color pages, regrettably as a 12-page insert rather than including the color illustrations with the articles they illustrated. The third edition General Index covered 52 pages with about 6500 entries. In addition, we added a Name Index of 11 pages with over 1500 names. The Editorial Board for the third edition had nine members (see Table 1), two of whom had been members of the previous two editorial boards, and seven new members.

Since I was no longer the chair of my department, I did not have the secretarial facilities available that I had had for the first two editions. However, I was just completing a five-year stint as editor of Abacus during which Springer (the publisher of Abacus) had funded an almost full-time assistant for me. So, for the third edition, Van Nostrand provided Ed Reilly and me with a budget that I used to hire that assistant, Caryl Dahlin, as Managing Editor of the third edition. She took almost all the burden of contacting – and dunning – authors off our shoulders as well as handling all the administrative tasks necessary to the project. (Dunning authors is always a difficult and sometimes unpleasant task; it was particularly difficult for one of the third edition authors who went to jail some time before the book was published.) The result was a relatively uneventful production cycle. The third edition contained 605 articles of which 174 were new and 70 were rewritten articles from the second edition. That these numbers were considerably higher than for the second edition was a clear result of the 10-year gap between the two editions. Another reflection of this gap was the 120 second-edition articles that did not appear in the third edition. Some of these were just dropped as no longer being sufficiently important to be included but other deletions resulted from a need to save space in order to try to accommodate an entire discipline in one volume. Thus, for example, seven articles on specific user groups like SHARE were all amalgamated into a single Users Groups article. One thing that didn't change from the second edition was the top-level categories in the Classification. One new appendix, a Timeline of Significant Computing Milestones, entirely the work of Ed Reilly, was added.

The title of the third edition reverted to the first edition title, Encyclopedia of Computer Science, although "and Engineering" would have been as accurate for the third edition as it was for the second. This was mainly a result of publisher pressure because of a belief, which we thought not unreasonable, that the second edition title was too unwieldy. In the Preface to the third edition, we assured readers that the reason for the title change wasn't an algorithm by which even-numbered editions would have the "and Engineering" and odd-numbered editions would not. No reviewer – or anyone else, so far as we know – remarked on this attempt at jocularity.

In 1989 there was as yet no Internet but Arpanet had been around for 20 years and there were various newer networks such as BITnet and NSFnet. While it was still relatively rare for books to be online in these networks, the Encyclopedia was, after all, an encyclopedia of computer science. So we strove mightily to get Van Nostrand to agree to an online version. I arranged a presentation at the Van Nostrand offices for all the top executives about how this might be accomplished and what it would mean for the perception of VNR in the computer science community. But I failed to persuade, the sticking point being the still high cost of converting a book like the Encyclopedia to online use. In retrospect, I should have tried harder. An online version in 1992 would have been good for VNR and for the Encyclopedia.

Shortly before publication Van Nostrand, a division of International Thomson Publishing, decided or, more likely, was ordered by the parent company to divest itself of computer-related books to a newly formed subsidiary, International Thomson Computer Press (Publisher #4) located in Boston. Thus, although the first printing of the third edition had the VNR imprint, the subsequent printing had the ITCP imprint.

V. The Fourth Edition

By early 1995 it was only just over two years since the publication of the third edition. I certainly wasn't anxious to consider a fourth edition. Equally certainly I knew that I was no longer competent to be an editor of an encyclopedia of computer science since my interests had strayed from computer science to discrete mathematics and mathematics education in the late 80s. I probably wasn't even competent for the third edition but, since Ed Reilly certainly was, no matter. But by 1994, Ed was about to retire from SUNY/Albany and, while surely more knowledgeable about computer science than I was, he, too, was not anxious to embark on the editing of a fourth edition. But the publisher, ITCP, was quite anxious to get the next edition into production. Ed and I were willing to do this only if we could find a third person, highly competent in computer science and as an editor, who would do most of the work in assembling the fourth edition.

Our search for such a person was surprisingly short and very successful. Through contacts at Union College, we were urged to consider David Hemmendinger whom one of us knew slightly and the other not at all. Although (rightly) wary at first, David succumbed to our charms after a couple of meetings and on 26 June 1995 the three of us, as co-editors, signed a contract for the fourth edition. We couldn't have made a better choice. David, despite mistakenly having begun his career as a philosopher, is broadly and deeply knowledgeable about computer science and is a superb editor. The lion's share of the credit for the fourth edition belongs to him.

The contract called for the delivery of a manuscript of a book not to exceed 1700 printed pages (the published book has 2034 pages, about the limit for a one-volume encyclopedia) no later than 15 May 1997. This was clearly too short a time to get the fourth edition done but, for reasons that will appear shortly, the publisher was very anxious to see the fourth edition published. In fact, the final manuscript was not delivered until 1999 and the fourth edition was published in July 20006.

The eagerness of ITCP to see the fourth edition published was not due to love of the book or the service ITCP might be doing to the computer science community. Rather it was a question of survival. ITCP had not been doing well since it was spun off from VNR and International Thomson, with a beady eye on the bottom line, was threatening to wind it up if the finances did not improve rapidly. The fourth edition was, it seemed, the only chance for the ITCP executives to retain their jobs. But not only could we not meet the contract date or anything close to it but also we did not see it as our obligation to make superhuman efforts to save the ITCP bacon. So in late 1997, ITCP duly folded and International Thomson transferred the Encyclopedia to International Thomson Business Press (Publisher #5) in London. A business press did not seem a wonderful fit to us but ITBP had published a well-received business encyclopedia and it was enthusiastic about publishing the fourth edition. Moreover, since I had by that time moved from Buffalo to London, having a local publisher was a plus.

No matter. We received good support from ITBP but it soon became clear to them (although not, at the time, to me) that they had insufficient understanding of the U.S. market or contacts in the U.S. for them to be able to market the book successfully. (Although the Encyclopedia has sold well all over the world, the U.S. has been by far the biggest market.) Or perhaps, for arcane corporate reasons, Thomson had just decided that it didn't want to publish the Encyclopedia. In any case, in mid-1998 ITBP put the Encyclopedia out to tender. Four publishers expressed interest, three well-known American publishers and the eventual "winner", Macmillan Reference (Publisher #6) in the U.K. This was good news for us because Macmillan Reference is a division of The Macmillan Group. Macmillan is an old (founded in 1843) and famous publisher (of, for example, Boole's Laws of Thought in 1854 and the first edition of Alice's Adventures in Wonderland in 1865). Moreover, although not known particularly for computer science publishing, Macmillan Reference has wide experience in publishing encyclopedias and dictionaries (e.g. The Dictionary of Art in 34 volumes and a four-volume Encyclopedia of Cognitive Science published in December 2002). (Macmillan in the U.K. was once the parent company of Macmillan Publishing in the U.S. but they are now quite separate companies; the main arm of Macmillan Reference in the U.S. is Grove's Dictionaries.) Indeed, in terms of getting the fourth edition assembled and published, Macmillan was a good choice. There were no unusual problems for the editors in compiling the fourth edition unless you wish to count such things as trying to locate an author of a third edition article who had changed both institution and sex since the publication of the third edition. The Web, by the way, was a great help in locating third edition authors.

The fourth edition has 623 articles including 103 new articles (among them: Internet and World Wide Web) and 29 rewritten third edition articles. 76 third edition articles were dropped or amalgamated (e.g. 11 of 17 third edition articles on particular computer societies were covered in a single Computer Societies article). The number of authors was 450, more than double the number in the first edition (210) although the number of articles had increased by only 33% since the first edition. The primary explanation of this is that updates of articles from a prior edition were often done by authors other than those who wrote the original articles. Of the 450 authors, 50 (11%) were women compared to, respectively, 10 (5%), 21 (7%) and 38 (10%) in the first three editions. The color insert was expanded to 16 pages. The top level of the Classification remained as it had been in the previous two editions. There are two new appendices. One has lists of presidents of major computer societies, lists that had been included with articles on these societies in the previous editions. The second lists all the articles that had appeared in a previous edition but are not in the fourth edition. A perusal of this appendix gives an interesting picture of how the discipline of computer science has changed over the past three decades. Table 5 gives another perspective; it indicates how the number of pages for three articles has changed over the four editions. The fourth edition Index has 43 pages, fewer than in the third edition, but because of a new three-column format and smaller type, it has more entries than before, over 8,000. The Name Index covers 14 pages and includes about 3,000 names. There was no Editorial Board for the fourth edition because Ed Reilly and I decided that we no longer needed this support. We did, however, consult widely with experts about the material to be added to (and subtracted from) the third edition.

The fourth edition has also been well-reviewed. But its sales will probably fall well short of those of the first three editions, largely, no doubt, due to the $150 list price (compared to $60, $87.50 and $99.95, respectively, for the first three editions; it is doubtful that the 50% increase in price over the third edition had any economic justification but publishers, not editors, determine the price of books). The most critical review was that in Computing Reviews8 which, although generally complimentary, expressed a strong belief that in 2000 this Encyclopedia should have been published "in a CD-ROM or DVD format with appropriate search software", perhaps, the review implied, to the exclusion of a print edition. Our first reaction to this was to consider contacting the author of the review to tell him that, if only he had checked with the publisher, he would have found that, while we were in favor of a full online edition rather than CD-ROM or DVD, an online edition was on the way. In any case, we agreed fully with the reviewer that, "an encyclopedia on this subject [should be] an exemplar of the benefits of computer technology". Lucky we didn't contact the reviewer because, sadly, there still isn't an online fourth edition. I still hope, even expect that there will be one but for too many reasons, some implied below, it hasn't happened yet. To my co-editors and me, it is highly embarrassing that in 2003 there still isn't an online edition of the Encyclopedia of Computer Science.

The Macmillan Group in the U.K. is a subsidiary of the Georg von Holtzbrinck companies, a large German publishing group that includes, among many others, the publishers of the Scientific American and Nature. Shortly before publication of the fourth edition, much of Macmillan Reference was absorbed by the Nature Publishing Group (Publisher #7) which became the imprint of the fourth edition. Although there are few, if any, periodicals in the world as prestigious as Nature, the Nature Publishing Group has very little expertise in computer science. By early 2002, realizing this and knowing also that the marketing of the fourth edition and the organization of an online edition had been going quite badly, Nature Publishing decided to sell the Encyclopedia and did so in April 2003 to John Wiley & Sons (Publisher #8).

VI. Related Matters

1. The Concise Edition

From the very first edition I had thought that the Encyclopedia should be in all high school libraries. Even with the budgets for these libraries steadily shrinking, if a high school library was only going to purchase one reference book in a year, surely this should be the one. However, either because the price was just too high or because the marketing was ineffective, it never happened although, of course, it was purchased by some high school libraries.

In addition, as noted earlier, we have always believed that each edition of the Encyclopedia should be, in large measure, readable by bright high school students and that it might make a suitable high school (or college) graduation gift. This doesn't seem to have happened too much either. With the third edition Ed Reilly and I decided that the best way to overcome sales resistance at the high school level would be to produce a concise, paperback version of the Encyclopedia. The publisher was willing but we waited rather too long to pursue this idea and, although we thought we had found a conciser for the third edition, ultimately the deal fell through.

We were, however, determined to achieve a Concise Edition of the fourth edition and, indeed, that is now in production and should be published by John Wiley in early 2004 with Ed Reilly the conciser. It will have somewhat less than half the words of the fourth edition.

2. History

Including a major component of computer history has been an aim of all four editions. In addition to the articles specifically devoted to computer history, all authors have been urged to assure that their articles do justice to the history of their subjects; in the main, they have complied. The fourth edition has 51 articles specifically devoted to aspects of the history of computing (not counting the biographies – see below). This is 13 more than in the third edition, most of this increase coming from 10 articles on the history of specific programming languages.

3. Biographies

A reviewer3 of the first edition remarked that, "To one who considers computer science to be a young field, it is a trifle disquieting to find biographies in this work. Further, some of the biographees are actually dead." In fact, of the 19 biographees in the first edition, 12 were dead at the time of publication.

Publishing biographies of living persons in an encyclopedia is a tricky business. Who should be included, who not? The simplest solution, guaranteed not to lose any friends or make any enemies, is just to omit any living person. For the first three editions the decision was made to include living people deemed to be so eminent that no one not included should be distressed. (Of course, deciding who is eminent in any branch of learning is a bit like deciding who is the best bridge player in the world; you never ask a prominent player who is the best player but rather who the second best player is.) For the fourth edition, we decided that the inclusion- exclusion problem was just getting too difficult so no new biographies of living persons were added. The fourth edition contains 38 biographies, of which only four were of persons alive at the time of publication and three of those have since died.

4. Language

Much as we might like to have thought that the various editions of the Encyclopedia were having an effect on computer terminology, more often than not, we were playing catch-up . For example, the third edition used "e-mail" but in the fourth edition we adopted the standard "email". One term on which we may have had some effect was "database" which we have used since the second edition whereas the first edition used "data base". We were perhaps a bit ahead of the game with sexist language. Whereas there were plenty of generic male pronouns in the first edition, these were rigorously excised from the later editions.

5. International Contributions

The contribution to the Encyclopedia from authors not in the United States has been considerable. For the fourth edition, for example, while 80% of the contributors are American, the other 20% were located in 18 other countries when they wrote their articles. With the number of authors in each country in parentheses, these 18 countries are: Australia (3), Austria (3), Belgium (3), Canada (11), Czech Republic (1), Denmark (2), Finland (1), France (6), Germany (6), Israel (2), Italy (1), Netherlands (1), New Zealand (2), Norway (4), Sweden (1), Switzerland (5), Thailand (1), UK (40).

6. Humor

An encyclopedia should be scholarly but it needn't take itself too seriously. We have tried to inject humor in appropriate places in all editions. I was particularly pleased with a reviewer of the first edition2 who wrote that, "An attractive feature of ECS is a sense of humor", referring to articles on Kludge, Glitch and Bug. The fourth edition retains Bug and Kludge but Glitch was dropped after the second edition. The "tradition" of including humor was beautifully continued by Ed Reilly in the Timeline of Significant Computing Milestones (e.g. the entry for 2678 (sic) reads, "100th edition of the Encyclopedia of Computer Science, the first not edited by Anthony Ralston").

7. Translation

To the best of my knowledge there has been only one translation of the Encyclopedia, a Japanese translation of the second edition. This lack of translations is partly, no doubt, due to the cost of translating such a large book but it also reflects the fact that English has been the lingua franca of computing since the time of publication of the first edition.

VII. The Future of the Encyclopedia

Any encyclopedia in book form is a snapshot of the discipline it purports to cover, to be sure, a blurry snapshot since the articles will have been written over a one to three year period before publication. As noted earlier, particularly in a discipline changing as rapidly as computer science, that snapshot will already be somewhat outdated when the book is published and will rapidly become more so. Indeed, this might suggest that a competing encyclopedia might well be published in between the editions of the Encyclopedia of Computer Science. But the investment – in time and effort – to compile the first edition of any encyclopedia is so large that no truly competing volume has been published yet. In 1976, just before the first edition was published, the first volume of the Kent, Belzer multi-volume work was published4. This has now grown to 43 volumes. It has clearly been intended for libraries and cannot be considered a real competitor of the Encyclopedia of Computer Science. The Computer Science and Engineering Handbook9 covers a lot of the same ground as the Encyclopedia but a handbook, since it is addressed to professionals, is not really the same as an encyclopedia so probably few people consider the two books to be competitors. The publication in 2001 of the Encyclopedia of Computers and Computer History7 comes closer to being a competitor than any other book. But it is less than half the size of the fourth edition, has mainly quite short articles and, with a price tag of $250, is probably intended for a different market than the Encyclopedia.

Work on a fifth edition should probably be started in 2004 or 2005. But should there be a print fifth edition or only an online one that can be regularly updated to forestall obsolescence? Although, as noted, a reviewer of the fourth edition thought that it should have been published only in electronic form, it is not clear (to me anyhow) that in 2007 or 2008, when a fifth edition would likely be published, the Encyclopedia should be only electronic. The convenience of a one volume reference book that can be accessed immediately on a shelf may well mean that it will still be desirable for some years to have a printed Encyclopedia of Computer Science.

What is clear, however, is that if there is a fifth edition in print form, simultaneously with its publication there should be an online version available. In addition, the planning for the fifth edition should include a mechanism for regular updating of the online version so that, if desired, it should be possible to assemble a sixth and any subsequent printed editions directly from the online encyclopedia.

Acknowledgement:
David Hemmendinger, Ed Reilly and Eric Weiss not only improved the presentation above but also remembered various things I had forgotten and, as well, corrected my sometimes faulty memory.

References
1. American Federation of Information Processing Societies, Taxonomy of Computer Science and Engineering, AFIPS, 1980.
2. American Library Association, "Review of the first edition", Booklist, Dec. 15, 1977, pp. 699-700.
3. Jackson W. Granholm, "From Access Methods to Konrad Zuse", Datamation, Nov. 1976, pp. 122-123.
4. Allen Kent, Jack Belzer eds. (with others), Encyclopedia of Computer Science and Technology, 43 volumes and ongoing, Marcel Dekker, New York, 1976 (for first volume).
5. Robert L. Patrick, "Review of the second edition", Datamation, Jul. 1983, pp. 241-242.
6. Anthony Ralston, ed., Encyclopedia of Computer Science, 1st ed., Petrocelli/Charter, New York, 1976. Anthony Ralston, ed., Encyclopedia of Computer Science and Engineering, 2nd ed., Van Nostrand Reinhold, New York, 1983. Anthony Ralston, Edwin D. Reilly, eds., Encyclopedia of Computer Science, 3rd ed., International Thomson Computer Press, Boston, 1993. Anthony Ralston, Edwin D. Reilly, David Hemmendinger, eds., Encyclopedia of Computer Science, 4th ed., Nature Publishing Group, London, 2000.
7. Raul Rojas, ed., Encyclopedia of Computers and Computer History, Fitzroy Dearborn, Chicago, 2001.
8. Melvin L. Tobias, "Review of the fourth edition", Computing Reviews, vol. 41, no. 10, Oct. 2000, p. 363 (review #0010-0534).
9. Allen Tucker, ed., The Computer Science and Engineering Handbook, CRC Press, Boca Raton, FL, 1997.

Table 1: Editorial Boards for the First Three Editions
First Edition Affiliations Second EditionAffiliations Third Edition Affiliations
John M. Bennett U. of Sydney Aaron Finerman U. of Michigan Malcolm P. Atkinson U. of Glasgow
A. S. Douglas London School of Economics Werner Frank Informatics Inc. Peter J. Denning George Mason U.
Nicholas V. Findler SUNY/Buffalo George Glaser Consultant Aaron Finerman U. of Michigan
Aaron Finerman SUNY/Stony Brook C. C. Gotlieb U. of Toronto Bernard A. Galler U. of Michigan
Patrick C. Fischer Penn. State U. J. A. N. Lee Virginia Poly. Martin Campbell-Kelly U. of Warwick (UK)
C. C. Gotlieb U. of Toronto Saul Rosen Purdue U. Laurence Press California State U. at Dominguez Hills
J. A. N. Lee Virginia Poly. Gerard Salton Cornell U. Jean E. Sammet Consultant
Saul Rosen Purdue Univ. Mary Shaw Carnegie-Mellon U. Eric A. Weiss (Chair) Consultant
Gerard Salton Cornell U. Eric A. Weiss Sun Oil Co. Ian H. Witten U. of Calgary
Andries van Dam Brown U.
Eric A. Weiss Sun Oil Co.
Table 2: Vital Statistics for the Four Editions
Edition Publication Date Number of Articles Number of Authors Number of Pagesa Salesb List Price
First June 1976 470 210 1523 + xxviii 10700 $60.00
Second October 1982 550 301 1664 + xxix 12300 $87.50
Third November 1992 605 370 1558 + xxv 11100 $99.95
Fourth July 2000 623 450 2034 + xxix 6500 $150.00
a. Roman numerals are pages of front matter; the third and fourth editions had approximately an eighth more words/page than the first and second editions
b. Approximate non-book club sales; for the fourth edition through 30 June 2002.
Table 3: Highest Level Categories in the Classification of Articles
First Edition Second, Third, Fourth Edition
Hardware Hardware
Computer Systems Computer Systems
Software Information and Data
Basic Terminology Softwarec
Theory Mathematics of Computing
Mathematics for Computer Science Theory of Computing
Applications Methodologies
Management, Societal, Economic and Legal Aspects Applications
Professional and Educational Aspects Computing Milieuxd
History
c. In the fourth edition, all of Software was grouped under three second-level headings: Programming and Programming Languages; Software Engineering; and Systems Software.
d. These milieux are, in the fourth edition: Computer Industry; Computer Science and Technology; Computer Societies; The Computing Profession; History; Internet; Legal Aspects of Computing; Management of Computing; and Society, Computers in. In addition there are several articles with no other articles at a lower level (e.g. Literature of Computing, Personal Computing)
Table 4: Appendices in Each Edition
Name of Appendix Editions in Which It Appeared
Abbreviations and Acronyms 1, 2, 3, 4
Mathematical Notation, Units of Measure 1, 2, 3
Notation and Units 4
Numerical Tables:
Powers of Two 1, 2, 3
Base Conversion Tables 1,2
Numerical Constants 2,3
Computer Science and Engineering Research Journals 2,3
Computer Journals and Magazines 4
Departments of Computer Science 2
Ph. D. Granting Departments of Computer Science 3, 4
Presidents of Major Computing Societies 4
Key High-Level (Programming) Languages 2, 3, 4
Glossary of Major Terms in Five (Natural) Languages 2, 3, 4
Articles Deleted from Previous Editions 4
Timeline of Significant Computing Milestones 3, 4
Table 5: Pages Devoted to (a) Analog Computers; (b) Computing Center; and (c) Personal Computinge
Edition Analog Computers Computing Center Personal Computing
1 27 8 0
2 22 8 4
3 19 5 5
4 6 0 7
e. Aspects of all three subjects were, of course, covered in other articles.