Is the current approach to computing education wrong? The author of this book claims it should focus more strongly on its roots in the liberal arts. The book traces these roots and examines several central tenets of computing in this light.

The introductory chapter sets the stage by quoting recent opinions and controversies related to classifying computing as art, science, and engineering, followed by a summary of the historical evolution of the liberal (as distinct from the mechanical) arts. Sack’s paradigm in this book is an extension of actor-network theory (ANT), whose most important concept is “translation,” with unavoidable loss and addition of information. Using this methodology, Sack traces a number of computer science (CS) texts and their histories and consequences. Of particular interest are equivalences claimed by CS, such as the Church-Turing thesis, which equates effectively computable functions with general recursive functions.

Each of the following chapters focuses on a specific concept and provides historical and philosophical foundations for it before turning to specific examples.

The chapter on translation traces the path from Hilbert’s decision problem to the Church-Turing thesis to the “contemporary common sense” idea that “a computer has the potential to perform an unlimited range of tasks,” seen as a consequence of the universal Turing machine argument.

The chapter on language contrasts the mathematical language of the physical sciences with the language of computation, which focuses on operations rather than functions. As such, it harks back to the work of the French encyclopedists, who described the work of artisans such as “pinmaker” using “the language of the arts.”

The next chapter traces the roots of the algorithm from arithmetic through the arithmetization program in the foundations of mathematics, as well as in writings about the mechanical arts, including Hellenistic engineering, Roman military texts, and guild regulations. It ends with a critique of the misuse of terms in machine “learning.”

Turning now to logic, Sack debunks “the myth ... that the computer is an outgrowth of logic,” instead placing computers as “machines of rhetoric, grammar, logic, and dialectic.” In addition, Sack points out the essential differences between logic circuits and mathematical logic (timing issues are crucial in the former, but have no parallel in the latter) and between logic programming and logic and, more generally, human argumentation.

Demonstration, and “demo” in particular, are the subject of the chapter on rhetoric, which extends logic with nonargumentative forms of persuasion. Specifically, pictures and videos are used in ways that obscure the way they were computed from technical data, making them difficult to refute. Sack attacks Wolfram’s A new kind of science [1] (which is hardly mainstream) on these grounds.

Chomsky’s hierarchy is connected to the arithmetization program, in the chapter on grammar, and critiqued for removing meaning from grammar and for insisting on infinite languages as an excuse to include recursion.

The concluding chapter criticizes CS (as in the ACM/IEEE curricula) for its emphasis on algorithms in the sense of the Church-Turing thesis, pointing out that these don’t cover topics such as concurrency and interactive systems.

The book offers an interesting point of view, but many ACM members will be put off by significant misrepresentations and misreadings of some classic texts in the field. For example, Sack states: “translation--as it is known in computer science and information theory--is ideally lossless.” This is false, even for information theory, where Shannon’s sampling theorem provides the sampling rate for the frequencies that need to be transmitted (such as human speech); it is doubly false for computing, where compilation, for example, omits syntactic information and adds other information about register use.

Sack misunderstands Turing’s imitation game to mean that a computer should “play the role of a man attempting to imitate a woman.” While Turing’s description can be read this way, the paper later makes it clear that the usual reading, where the computer is trying to imitate a human analogously to the man imitating a woman, is the correct one.

Sack fails to distinguish between Turing’s formal proof of the equivalence (in terms of functions computed) between Turing machines and the lambda calculus and the informal argument that these formalisms capture exactly the effectively computable functions. This argument can never be proved mathematically; however, the fact that attempts by Church and Turing (and others) to formalize this notion, although very different, led to exactly the same class of functions provides strong corroboration for this thesis.

All practitioners know that software development requires a lot of communication skills, which to me is the most important argument for why computing education needs to have a strong liberal arts component. To my surprise, Sack doesn’t make this point at all.