From time to time the Communications of the ACM publishes papers that are primarily historical. There is nothing wrong with that; indeed, a historical perspective is often just what is needed to supply context to the technical papers that typically appear in this journal. It need not be said that the editors should hold historical works to the same standards of scholarship as they do any other papers.

This paper falls short of these standards, I believe, although in the main it addresses an important topic that deserves to be carefully examined. The paper’s thesis is that the news media of the late 1940s and 1950s, primarily newspapers and newsreel films, were primarily responsible for introducing and establishing a view of computers that is fundamentally misleading, one that the author implies has adversely affected the course of computing technology since then. The author supports this thesis by first discussing the recognition that journalists writing about science often feel constrained to humanize or otherwise add an element of drama to a story to keep the reader’s attention; this leads to stories about computers as “awesome thinking machines.” She then shows how many computer pioneers themselves, including Mauchly and Turing, tried to correct this view, in vain. She finally shows that as ordinary people began to use computers, the myth faded away, to be replaced at first by a disillusionment with computers and then by a public perception of the computer as a tool.

My criticism of this paper is twofold. First, the evidence presented to support the case is flawed. The author relied on files of newspaper clippings at two repositories, but she apparently did not adequately research the topic on which those newspaper clippings were based. A keystone of her thesis is that the early computer projects were “well-funded, top-secret projects…not announced to the public until 1946” (p. 124). That statement is simply wrong: Vannevar Bush’s Differential Analyzer was well-publicized in the mid-1930s, and these written descriptions were used as a basis for as many as eight copies made elsewhere around the world before World War II. Howard Aiken’s Mark I, while built in secrecy, was publicly dedicated in the summer of 1944, two months after the D-Day invasions of Normandy. All the major newspapers and science magazines carried a brief but fairly detailed description of it. Also, it is not true that Atanasoff’s work on an electronic computer was held back because of “concerns about patent issues” (p. 124); the patent issue arose only much later. The author implies that Atanosoff did not receive sufficient publicity for his machine at the time, but she fails to mention that the publicity (for an incomplete machine) was enough to attract the attention of the National Defense Research Committee, which sent representatives to Iowa in 1940 to have a look at it. Errors like these severely undermine the author’s thesis, whatever its merits.

My second objection is to the thesis itself. The author concludes that this “myth” may have retarded a perception that the computer is a “beneficial tool for everyone” (p. 132). Does she wish that this had been more widely believed? Is that what the computer really was (or is)? The implication is that such a belief is far less of a myth. One could write volumes about that implication. A few paragraphs later, she does admit to the “real problems caused by the computer revolution of the past 25 years,” but these are hardly emphasized. In short, I remain unconvinced that the “myth” of the computer as an “awesome thinking machine” is a myth after all. It is an awesome machine--its stored-program architecture transforms it into something far greater than a super-fast calculator (an image the author would have preferred). We do understand now that computers do not really think, but we have this understanding only after two decades of AI research, in which the very definition of “thinking” has been examined. It is unfair to castigate writers of the 1940s and 1950s for failing to have that sophistication. It is relevant that in 1949 Edmund Berkeley wrote a book called Giant brains, or machines that think [1] and that Konrad Zuse, who built a series of electromechanical calculators in Germany during World War II, as early as 1937 talked about the feasibility of building a “mechanical brain.” (Although Berkeley was not only a computer pioneer but also one of the founders of the ACM, Martin does not cite his book.)

Despite the flaws in this paper, I hope CACM continues to explore the historical and social dimension of computing in its papers. I suggest only that the journal consider a more thorough refereeing and review process for papers that are not in its usual subject area.