Issue No. 02 - April-June (2012 vol. 34)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/MAHC.2012.30
Mark Priestley, A Science of Operations: Machines, Logic and the Invention of Programming, Springer, 2011, 350 pp.
From the perspective of a contemporary computer scientist or programmer, the connections between programming languages and logic seem to be strong. However, how strong were they really during early developments? How did individual decisions influence this connection? Taking the reader back to 19th century England, Mark Priestley's A Science of Operations narrates the story of communicating the problem to the machine from Babbage's time until the late 20th century—just before object oriented programming overtook structured programming.
Priestley begins by putting the mechanization of various processes into a broad cultural context, starting with Charles Babbage's efforts. He draws Menabrea's and Ada Augusta's mathematical (the user's) view of the engine to the readers' attention as opposed to Babbage's mechanical (the constructor's) view by painstakingly analyzing the original writings of Babbage, Lovelace, and Menabrea.
Connections between automation and logic could be developed from either side. However, as the analysis of the programming of early relay machines such as ENIAC and the Bell Labs relay machine shows, by and large, the engineers were unaware of the developments in logic. Rather, the construction of the machines copied the practices of manual computation. Logicians on the other hand, in particular von Neumann and Wilkes, were interested in the practical aspects of computation.
From the 1930s through 1950s, the concepts employed in programming were being developed alongside machine design. These varied much less than the design of the early machines (ASCC, Bell Labs relay machines, ENIAC, EDVAC, EDSAC, and ACE). Following this period, in the 1950s, programming went through a crucial change from automatic coding to programming languages. Developments in automatic coding intertwined with developments in machine translation and notation. The concept of universal languages also belong to this era, when Fortran and LISP and the first draft of Algol were developed.
Analysis of the Algol research program forms the core of the book. Priestley attempts to answer the question that the Russian computer scientist Andrei Petrovich Ershov posed to Robert Bemer: Why did Algol, despite its seemingly disappointing history, change the face of programming? 1 Priestley views Algol 60 as an "obvious landmark"—a term taken from the first ACM History of Programming Language (HOPL) Conference—and as a paradigmatic change in the Kuhnian sense. Priestley skillfully explains how the story of Algol 60 fits the idea of a research program developed by the philosopher of science Imre Lakatos. Beyond the technicalities of the programming language, Priestley notes the invention of institutions as communication vehicles. Such institutions included the committee style of work on Algol 60, which had already been adopted for the Algol 58 report, and the Algol Bulletin, founded and edited by Peter Naur. However, Priestley does not go beyond the technical level in exploring the relationship between the individual actors.
After the publication of the Algol 60 report, new issues arose in programming. For example, computer programmers dreamed of replacing the debugging of a program with a proof that showed the program met particular specifications. They also developed structured programming and, over time, found that their use of programming languages also influenced their treatment of data structures.
Although Priestley's book is mostly based on English-language published and archival material (as an appendix, the complete table for the universal Turing machine is included), it presents a variety of viewpoints. In fact, the effort to embrace the whole story becomes detrimental. For example, in Chapter 3, Priestley juxtaposes Hollerith machines and the work of Leslie J. Comrie as two examples of semiautomatic computation, and in Chapter 4, he deals with what is traditionally taken as the "other" source of the development of modern computers and their programming: logic and formal systems. These two chapters do not even inspire the author to write his own conclusions, which he does in other parts of the book.
In general, Priestley does an excellent job describing the technical work of the individual characters in his book. Beyond the account of technicalities, he draws general conclusions and relates his work to the classical works in the history and philosophy of science. He provides a sound account of the interconnections between logic and computing and places his findings into a broader methodological framework. Anyone interested in history of programming will certainly find this book both helpful and inspiring.
Helena Durnovà is an assistant professor in the Department of Mathematics at Masaryk University, Czech Republic. Contact her at email@example.com.
Simon Lavington, Moving Targets: Elliott-Automation and the Dawn of the Computer Age in Britain, 1947–67, Springer, 2011, 732 pp.
Elliott Brothers, later Elliott-Automation, was a defining presence in the early British computing sector, responsible in 1961 for half the installations supplied to UK customers. Whereas several of its competitors—Ferranti, Leo, and the data-processing firms that amalgamated as ICT—have received detailed corporate histories, Elliott's story has remained largely unexplored. Simon Lavington's new volume therefore represents a valuable contribution to the national literature.
Moving Targets is a weighty tome in several senses. Comprising 500 pages of body text, with numerous appendices specifying system architectures and instruction sets, cheerfully assuming familiarity with eigenvalues (p. 184) and priced for institutional purchase, it is surely unlikely to "appeal to the general reader curious about the emergence of digital computing in Britain," as the cover blurb gallantly asserts. Such readers are far better served by Lavington's own Early British Computers (Manchester Univ. Press, 1980), which is now available online at http://ed-thelen.org/comp-hist/EarlyBritish.html.
Rather, its major accomplishment is in matching the depth of other corporate histories while starting from a much less promising archival base. The physical uprooting, confusion, and disillusionment that marked Elliott's disappearance, in a flurry of ignominious mergers around 1967, involved the loss of many valuable records. Lavington has painstakingly plugged the gaps through interviews and email exchanges with former staff—many now deceased—and from the firm's surviving research reports, itemized in a useful appendix.
The explanation for the book's size is simple: Lavington leaves out nothing he encountered. Most of the secondary literature mentioned is summarized and primary sources are quoted verbatim at length, with two of the 14 chapters based directly on personal histories drafted by ex-Elliott staff. Coverage runs far beyond the book's stated timeframe, moreover, in places into the 1980s. Perhaps inevitably, the welter of detail frustrates long-range narrative coherence. Lavington helpfully acknowledges that few will read the volume from cover to cover and provides a chapter map locating coverage of the three main themes: general-purpose computing, industrial automation, and defense (see p. x).
Despite its subtitle, the study is not a contribution to the wider social or business history literature. There is no reference, for instance, to Marie Hicks's recent work on the status of women in the nascent British industry, although it becomes gradually clear that the Elliott environment both fostered and constrained a number of notable female developers such as Dina Vaughan/St Johnston, who left the firm to found the UK's first independent software development agency. Similarly, although Lavington refers briefly to John Hendry's work on the National Research Development Corporation (p. 147), there is no systematic engagement with the literature on national computing policy.
The defense theme is more pronounced. Elliott's main computer research laboratory at Borehamwood grew from a Royal Navy "shadow factory," part of a wartime initiative to disperse production, and most of its early projects were classified. Although the firm was strongly committed to civilian commercial computers, it was chiefly defense patronage that brought it to international prominence in the 1950s, and Lavington argues, the cancellation of cutting-edge military projects was the principal factor in the collapse of the Elliott identity.
This focus sits in interesting tension with the treatment of Leon Bagrit, Elliott's long-term managing director. The trading name "Elliott-Automation," formally the consequence of a merger in 1957, in practice reflected Bagrit's personal vision of industrial reform grounded in efficient, systematic process control and monitoring. Bagrit's self-positioning as "Mr. Automation," articulated publicly through a series of BBC Reith Lectures, strongly differentiated his company from its competitors. Bagrit's corporate strategy, meanwhile, was to license crucial American technologies on a piecemeal basis, quietly acquiring the means to dominate niche areas internationally. On the face of it, this approach was closer to Prime Minister Harold Wilson's "toolroom of the world" manifesto than to the grandiose aerospace visions that preceded it. Nevertheless, it was under Wilson that the Elliott name, and the automation ethos, disappeared. Further research is needed to unravel the real extent of Bagrit's relevance to national policy.
To question Springer's commitment to this volume might seem churlish; the British retail price of £41.99 represents exceptionally good value for a specialist hardback of such length and will hopefully inspire some non-library sales. Nevertheless, the text as published resembles an author's typescript to the degree that it is difficult to see what, beyond distribution and binding, the publication process has added. A sharp editorial eye could have drawn out wider resonances, excised repetitions, and restrained idiosyncrasies such as the fairytale exegesis of Elliott-Ferranti relations (p. 170). On a more technical level, it should not be beyond an academic press to regularize formatting or ensure that terminology such as "similar to a pseudo DDA" (p. 417) is not left entirely unexplained. The index, however, is usable.
Overall, Moving Targets is an invaluable reference work and an essential sourcebook for first-hand accounts of an otherwise neglected episode.
James Sumner is a lecturer in history of technology at the University of Manchester, UK. Contact him at firstname.lastname@example.org.
Contact Reviews Department Editor Andrew Russell at email@example.com.