Reflecting the momentous changes in computer science in the ten years since the publication of the second edition of this “basic reference work for [technically sophisticated] non-specialists,” the current edition contains 40 percent new or significantly revised material. The overall size remains about the same, however: 1,558 double-columned pages in one bulky but nicely produced volume. It contains about 10 percent more articles, arranged alphabetically from “Abstract Data Type” to “Zuse, Konrad,” followed by six appendices and two indices. Most articles are preceded by cross-references and followed by short reference lists.

Those who wish to use this volume for self-study can escape from the alphabetical sequence by consulting the editors’ “Classification of Articles.” This taxonomy divides computer science into nine fields: hardware, computer systems, information and data, software, mathematics of computing, theory of computing, methodologies, applications, and computing milieux. Each field is then divided into 2 to 16 categories, and some of these subcategories are further divided. All articles are included in the taxonomy, and some articles appear more than once. The order of the primary divisions is not alphabetical and suggests a progression from the most basic to the peripheral, but articles listed at lower taxonomic levels are listed alphabetically. The taxonomy would have been more useful had the editors also ordered articles at the lower levels, perhaps from the more general to the more specific or, where applicable, chronologically. Nevertheless, I found the taxonomy most valuable for both organizing my reading and stimulating my thinking about the discipline.

For me, the most interesting question is whether a one- or two-volume encyclopedia fulfills a reference need. If all one needs is a comprehensive definition, several adequate dictionaries are available [1,2]. For more extensive information, one can consult a more specialized encyclopedia (such as Shapiro [3]) or a book-length treatment. But does an intermediate niche exist for a volume of short articles? Is there a level of curiosity that cannot be satisfied by dictionary definitions but would be sated by a book or a long article? The subjective answer is yes, but in this volume the mean article length is about 2.4 pages, which is often too short. That I found many articles competently presented but intellectually unsatisfying reflects, generally, on neither the editors nor the authors. It is a comment on the current state of the discipline, the many and complex elements of which simply cannot, in many cases, be adequately explicated in articles of such modest size.

Many articles have been extensively rewritten, and virtually all articles have been at least minimally revised, according to the editors, but a few articles in need of major revision were not updated. The piece on C, for example, is almost the same as its counterpart in the second edition published in 1983 [4]. Consequently, the ANSI C standard adopted in 1989 is mentioned but not discussed; the major program example uses the obsolete (though still legal) form of function declaration; a syntax error in that example is retained from the second edition; and the widespread use of C in microcomputer program development in the last decade is ignored.

This volume appears to have a disproportionate number of minor errors, omissions, and questionable editorial decisions; none of them is fatal, but they are nonetheless irritating to the reader. The errors include (but are not limited to) page and figure references, spelling, English and programming language syntax, and diction. In addition, the article bibliographies are too brief (sometimes nonexistent), the objectivity of company articles written by that company’s employees is questionable, the use of a major article that has appeared elsewhere is redundant, the appearance of not otherwise identified last names in the text and in the name index is unhelpful, and the biographies are insufficiently critical. Also, two articles written by the same author disagree on a matter of fact. Finally, I cite one technical quibble: the discussion of the Waring problem states that “every positive integer is the sum of…nine cubes,” whereas a more precise statement would include “no more than” before “nine,” since it has been shown that natural numbers greater than 239 require fewer than nine cubes.

A work by 370 authors will inevitably be uneven in style and substance. Repeatedly, having enjoyed a well-written and highly informative article, I found my expectations accordingly raised, only to be dashed when the next selection turned out to be a clinker. Unfortunately for the editors, the clinkers tend to be remembered more than the routinely competent pieces, perhaps even more than the tours de force. Reflecting on the work as a whole, however, I find the time I invested in it well spent. I recommend this book, warts and all, for self-study. For occasional consultation, its use is more problematic, since the article one needs may happen to be a loser. There is a workaround for this problem, however, since a particular query will likely require one to read more than one article, and there is some redundancy among articles.