Iffy index no help to fuzzy logic

Since 1945, computers have permeated our culture, economy and physical environment. For a book with the words encyclopedia, computer and history in its title, the problem is not what to put in but what to leave out.

Raul Rojas explains in his preface that this is only an introduction to "the fascinating world of computer history". It focuses on the history of systems and networks, with a limited amount of material on software and the theory of computation. Some material on contemporary digital media and cyberculture has been included, as if to show where, astonishingly, it all led.

By claiming to cover the entire history of computing, the encyclopedia is perhaps easier to sell than one that ends at, say, 1990. But the rather patchy recent material adds little to its use as a reference source.

There are some good factual articles, such as Chris Woodford's pieces on Wired magazine and Netscape Communications. There are snippets on ASCII art, chat rooms, virtual communities, spam, Yahoo!, netiquette and emoticons. The smiley face :-) was apparently first used by Scott Fahlman, a researcher at the Massachusetts Institute of Technology. Now there's a guy who has made the world a happier place. But the book often lacks the historical dimension: colour management is discussed without reference to Edwin Land's revolutionary work on colour perception, and the article on fonts fails to mention that typographers were creating founts (spelt thus) long before computers appeared.

Much more secure is the material on early hardware. Leibniz and Pascal both made calculating machines in the latter part of the 17th century. Less well known is Gottfried Schickard, a friend of the astronomer Kepler, who invented a calculating machine in 1623. Schickard's letters were lost for 300 years.

Vigorous debate breaks out when we reach the 20th-century pioneers. For each of the major figures, Rojas has been able to find a passionate but scholarly advocate, for example Andrew Hodges on Alan Turing. Rojas himself speaks up for Konrad Zuse, who began building computing machines in Berlin in 1936. But the computer is not a single invention. Even such key ideas as the stored program evolved gradually from theoretical abstraction to engineering reality.

At the back of volume two, there is a list of all known computers as of 1955. At that time, 16 countries had built computers. From 1955 onwards, the book concentrates on machines that are seen as commercially or technically important. For a while, the mainframe business was dominated by a group of companies known as "IBM and the seven dwarfs". Then Digital Equipment led the minicomputer revolution, and Apple made the breakthrough in personal computing. Do not look here for minor brands of PC clone or Unix box. Even ICL, the British company that nipped at IBM's heels, is barely mentioned.

The book does not set out to be an encyclopedia of software, but programs and programming are an integral part of computer history. There are substantial articles on historically significant operating systems such as Multics, Unix, Linux, VMS, Windows and Windows NT, CP/M and DOS. To get Brian Kernighan to write on Unix is a coup.

A general article on operating systems, by John Deane, tells how Unix inspired CP/M, "which was copied as QDOS, then reworked by Microsoft for IBM's personal computer". The "DOS platform" article tells the tale in more detail. QDOS was no mere copy. It had a CP/M-like interface, but handled files more efficiently. It was written for Seattle Computer Products by a student at the University of Washington, Tim Paterson. A small local company called Microsoft offered to market QDOS, claiming that it already had one prospective customer. "SCP came away with $25,000 in cash, and Microsoft had obtained an operating system for their secret customer, IBM." Of course, Microsoft never looked back. The story is told here by Paterson himself, without obvious bitterness.

Turning to programming languages, the coverage is adequate but hardly encyclopedic. There is material on Ada but not Coral; APL but not Forth; Oberon but not Objective-C. Individual application packages generally do not get their own entries, but there are articles on the major personal application categories such as spreadsheets, word processing and desktop publishing.

Ivan Sutherland's invention of the virtual reality headset is mentioned not in his biography but in the general article on VR. The article on the "Reduced instruction set computer" does not mention an important Risc feature, the elimination of microcode, but in the article on "Microprogramming", however, all is made plain. This article contains a better explanation of the whole Risc concept.

The article on searching and sorting mentions only one sorting algorithm, the insertion sort. The efficient and widely used Quicksort algorithm was invented by Tony Hoare and is duly mentioned in his biography, but with only the barest explanation. You might not guess that elsewhere in the book there is a full, implementable description of Quicksort. It is given as an example in the article on algorithms.

"Cray 1", "Cray Research" and "Cray, Seymour" come one after another, by three different authors and with much repetition. The article on supercomputers goes over the Cray saga yet again, when it could have been telling us more about Japanese supercomputers and explaining how vector processing and pipelining work.

There are large overlaps between "Wiener, Norbert", "Cybernetics" and a third article on Wiener's writings. Peter Asaro's clear cybernetics article covers virtually all leading cyberneticists and makes the necessary link to present-day genetic algorithms and artificial neural networks. Asaro argues that the main ideas of cybernetics were presented by Ross Ashby in 1940, three years before Wiener. It was Wiener who eventually gave the field its name, but devoting two more articles to Wiener seems excessive.

The editors have let through a few inaccuracies and some shaky technical explanations. The book sometimes contradicts itself. Minicomputers were distinguished by their physical size (page 489 - wrong) and by their word length (page 551 - correct). A twisted-pair cable has four wires on page 537 and, more plausibly, two wires on page 781. The 6502 microprocessor was made by Motorola (page 849); by Rockwell (page 520); by MOS Technology (page 409). In fact, the 6502 was designed by MOS Technology, though a number of other firms produced it under licence.

The essay-like articles are richly cross-referenced and fine for surfing. But if you want the answer to a specific question, you are compelled to use the index. And here the book does itself a real injustice because, unless my sampling was very unlucky, the index is distressingly incomplete.

Jef Raskin is mentioned in the Macintosh entry but not in the index. Clive Sinclair is absent from the index, even though Jonathan Bowen very properly mentions the Sinclair ZX80 in his article on microprocessors. The ZX80 showed for the first time how small and cheap a microcomputer can be.

The cult programming language Forth is not indexed. But Woodford gives it its place in history as the forebear of PostScript, the language in which computers talk to printers. The Objective-C language is rightly mentioned as an important part of the NeXT computer system. Not a peep about it in the index, though. Neither Symbolics nor Tandem is in the index. You would have to chance on the articles on Lisp and fault-tolerant computing respectively to discover what the two companies did.

Or suppose you wanted to find out about mercury delay-line memory. There is nothing about it in the index under "mercury", "delay" or "memory". But the book contains at least three mentions and one picture of delay-line memory. Turing at the National Physical Laboratory wanted some for the Ace but could not get them delivered in time. Maurice Wilkes was luckier, or more patient, and his Cambridge team succeeded in building the Electronic Delay Storage Automatic Calculator. Martin Campbell-Kelly's article on Edsac has the details, and there is more in Jon Agar's biography of Wilkes. The bulky piece of plumbing that we see Wilkes admiring is in fact a group of delay lines filled with the toxic liquid metal. A memory leak in those days was not a matter for debugging but for decontamination.

Altogether this encyclopedia is a bit of an Easter egg hunt. It has some excellent material, but a book of this kind must be tighter, more comprehensive, more carefully checked and more brutally edited. Above all, it needs to be better indexed. All these things could be achieved in a second edition. Which reminds me: though recursion is explained, there is no entry for "iteration".

Tony Durham is on the staff of The THES .