Book Reviews 263

Chapter 4, “Computers,” is the book’s longest chapter and, except for the brief “Conclusion,” the only one with previously unpublished material. The lengthy discussion of the Turing machine is useful as an introduction to its mode of operation, but the surrounding argument is poorly motivated, and its conclusion, that the Turing machine model is “a mathematical construction, with a specific historical origin” (p. 112) whose value “lies in its inter-subjectivity rather than its apparent objectiv- ity” (p. 115) is, although reasonable, unconvincingly presented. The argument seems to rest on the spread of computational operations “beyond the specialized research communities from which auto- mata theory, formal logic and working computers emerged” (p. 116) or on the fact that these oper- ations “have been partly assimilated into wider social life outside specialized research communities” (p. 126). This spread, and assimilation, of computing reveals the flaws in the picture of computer programs as forms of intelligence, a picture whose force depends upon the control over its interpre- tation by specialized research communities (especially those, as Warner points out, who interpret programs as representations of mental, or even brain states). But the connections between the inter- pretations of computing by specific discursive communities and the truth of claims regarding the intel- ligence of computers are never made. It would seem that Warner wants to argue for a social constructivist view of computing, but if so, his arguments would be stronger if they built upon recent work on social constructivism in general and its application to artificial intelligence in particular (see, e.g., Collins, 1990).

The weaknesses of Chapter 4 reflect the book’s deeper problems. Although there are interest- ing, if introductory, discussions of specific topics, its theoretical apparatus remains obscure and unconvincing. The point of introducing semiotics is to establish that computer programming is a “form of writing which seems never to have been intended as a substitute for speech” (p. 64). The discussion of written language is intended to clear away the objection that programming cannot be a form of writing, since all forms of writing must be oral because they represent speech. Lest this thesis appear crashingly obvious (conversations in C-Plus are unlikely to be overheard in one’s favorite restaurant), Warner explains that he seeks a “‘unifying, intellectual principle’ to connect books and computers” (p. 2). But the reader deserves more, by way of motivating this latter claim, than a brusque reminder that the “significance of such a consideration should not require undue insistence” (p. 2). Even a gesture towards the many other theorists, from McLuhan to Baudrillard and beyond, who have insisted on precisely the significance that Warner takes for granted, would be helpful. A semiotic apparatus is not required to establish that computer programs are written artifacts, that human interpretation is required for their signification, nor, since Searle’s famous 1980 article, that meaning is not inherent in them. The book’s theoretical apparatus appears forced (is it motivated to solve the problem of collecting into book form some previously published material?), a weakness that remains unmitigated by the plausibility of the central, if occluded, thesis that computing must be understood as a form of social and political discourse.

REFERENCES

Collins, H.M. (1990). Artificial experts: Sociul know/edge and intelligent machines. Cambridge, MA, and London: MIT Press.

Searle, J.R. (1980). Minds, brains and programs. The behavioral and bruin sciences. Cambridge, MA, and London: MIT Press.

Graduate School of Library and Information Science University of Western Ontario London, Ontario, Canada

BERND FROHMANN

Claude Elwood Shannon: Collected Papers. N.J.A. SLOANE and A.D. WYNER (EDs.). IEEE Press, New York (1993). xliv + 924 pp., ISBN o-7803-0434-9.

This volume contains the first complete bibliography of Shannon’s works, numbering 127 items; 76 of these are published here (some for the first time) in their original format, including Shannon’s masterwork, “A Mathematical Theory of Communication,” written in 1948. Also included is a 1987 interview with Omni magazine that shows the lighter side of the man- his boyhood interest in Morse Code, for instance, and his lifelong interest in erector sets.

Of particular interest is the inclusion of Shannon’s “lost” M.I.T. Ph.D. thesis on population genetics. The editors of the collection did not find it mentioned in a recent authoritative bibliogra- phy on the subject. They asked two experts in the genetics field to comment on its absence. One of them said Shannon was working in isolation, and this was why it had never been disseminated in the genetics research community. If it had been, he believes it would have made a substantial difference

264 Book Reviews

to the history of genetics. The publication of the Shannon thesis here for the first time is therefore interesting on several levels.

Is communication theory important enough to today’s information science to buy a 900-page volume of Shannon’s collected papers? Can one get by with the original Shannon (plus Weaver), first published in 1949 and still in print? I will try to answer the first question first.

Shannon is best known for his communication theory, or information theory, as it came to be known. In the 1960s and 70s there was a backlash among information scientists against the behaviouralist S-R model that Shannon’s theory, or rather the use of it in information science, seemed to support. The primary criticism then was that the theory seemed to support the notion that infor- mation could be looked at separately from the “effect” it had on the person receiving it-in other words, the “psychological” element of information, or the “meaning” of the information, which crit- ics said was individual and not subject to mathematical laws. But did Shannon’s theory exclude “meaning” from the discussion? That was the debate.

Since that time, information science itself has shifted; information is now seen as a process constructed by the information user; and, at least in the beginning of the process, information and meaning may be separate things. Shannon’s theory once again has possibilities here.

Now to the second question: Is it worth buying Shannon’s Collected Papers, which includes everything important he ever wrote in such diverse fields as cryptography, computer games, and genet- ics, if one already has his Mathematical Theory of Communication in one form or another? The answer to this question can be divided into two parts for two audiences.

The first part is the research library, which aims at an audience interested in what goes into the making of an important theory, a theory still being discussed and debated (part of the reason his theory is still talked about is because it is hard to understand, or open to diverse interpretations). For this audience the present volume gathers into one place articles that are presently dispersed in vari- ous libraries on campus.

The second part of the answer is whether or not the present volume is suitable for an audience not familiar with the mathematical theory of communication, such as a student in communications or information science; for this audience, not only is the present volume carefully and interestingly put together, but it also shows the full breadth of what interested Shannon over the course of his life. The student new to the subject might find one of these other things interesting, and then read the communication theory later.

The present collection’s primary concern, however, is for the researcher. Shannon is not only important because his theory links information science to physics and the central philosophical and scientific issues of our day; because it does this, Shannon continues to influence present-day, state- of-the-art, new-paradigm information science research.

REFERENCE

Shannon, C.E., & Weaver, W. (1949). The mathematical theory ofcommunicafion. Urbana, IL: The University of Illinois Press.

Graduate School of Library and Information Science Elborn College

CHARLES COLE

University of Western Ontario London, Ontario, Canada

Software by Design. HAROLD SALZMAN and STEPHEN ROSENTHAL. Oxford University Press, New York (1994). 348 pp., $39.95, ISBN o-19-508340-7.

Do social values shape technology design? Are design decisions made purely on the basis of efficiency and economy? Must the work organization accept technology “as is” and learn to live with it? These are just a few questions that this book addresses. The authors’ analysis lead them to conclude that technology design is embedded in an interwoven net of people, organizations, and technology. Thus, their argument goes, social values and power are key components in the creation of technology. Orga- nizations often have form rules, policies, and procedures, but operate according to an informal struc- ture that allows for more autonomy and flexibility in procedures. This may result in a more efficient and expedient way of doing things. This book is also about how technology is socially constructed and the underlying values that come into play in that process. The study of software design and the contradictions and conflicts in the workplace and software design may not be confronted to the same degree in other technologies or in different environments. Software may be different from other tech-