The 1951 Conference on Automatic Computing Machines held in Sydney was one of the world's earliest computer conferences. The Proceedings of the Conference provide a detailed insight into the state of Anglo-Australian computing only three years after the running of the world's first stored-program computer (nicknamed the 'Baby') at the University of Manchester on 21st June 1948. A number of the topics discussed at the Conference were issues that remained relevant for decades.
Aim of the Conference
The Conference on Automatic Computing Machines, held in the Department of Electrical Engineering at the University of Sydney on 7th - 9th August 1951, aimed to provide a chance for Australian computer scientists and technologists to review progress in the field of computing machines, in light of developments that had occurred around the world during the previous decade.
Emeritus Professor Sir John Madsen, in his 'Address of Welcome' at the opening session of the conference, commented on recent developments in computing methods and techniques, noting that progress in the field 'can only be described as spectacular; so much so that its protagonists are at considerable pains to keep abreast of it' (Proceedings, p. 5). There was a consensus that developments in computing, stimulated by the Second World War, had been extremely rapid and that there was a pressing need to keep up-to-date with new developments or risk being left behind.
Organisation of the Conference
The conference was convened by CSIRO and the Department of Electrical Engineering, University of Sydney. Professor John Madsen in his opening address paid tribute to the central role that David Myers, Professor of Electrical Engineering at the University of Sydney, had played in organising the Conference. [It is interesting to note that in his reminiscences on the conference Trevor Pearcey stated specifically that the conference was 'organised jointly by Radiophysics and the Section of Mathematical Instruments' (see Bennett, et al, p.28), whereas Myers in his reminiscences (Bennet, et al p.14) failed to mention the role that Radiophysics played in the organisation of the conference. This difference in emphasis on who organised the conference is reflected in the different technologies (analogue versus digital) which these participants were researching, developing and promoting at that time]. The organizing secretary Ross Blunden was from the CSIRO Section of Mathematical Instruments (SMI) and was a close colleague of Myers.
Myers had been responsible for establishing the SIM within the CSIR Division of Electrochemistry in 1948. At the end of 1948 the SMI became a section within the Department of Electrical Engineering of the University of Sydney, where Myers succeeded Madsen as head of department. The main function of the SMI had been to provide advice on the solution of mathematical problems and on the introduction of various mathematical computing aids that would assist in that process.
Analogue versus digital
Under David Myers the SMI concentrated on the design and construction of analogue devices, for example, the CSIRO differential analyser, which could be built quickly and at low cost, in contrast to the larger, more expensive digital machines, then in existence, or on the drawing board. Myers believed, at that stage, that the availability of convenient electronic digital computers at reasonable cost was still many years in the future. The record of the 1951 conference is an historically instructive document because the papers and associated discussions reflect a time where there were two main competing computing technologies, each with their respective possibilities and apparent advantages. Although Professor Madsen in his introduction attempted to present a balanced discussion of the two technologies he appeared to place slightly more emphasis on the more familiar, reliable and convenient analogue technology - while also acknowledging that "steady progress" had taken place "in the field of digital machines (Proceedings, p. 6).
It is only in retrospect that it can be seen that the proceedings of the 1951 conference provide a window into a transition period from analogue computing devices to digital computing devices. Myers was the key initiator of the conference, and at the time the conference was proposed he probably would have expected that the greater emphasis would be on the role and future of analogue computing. However by the time the conference began it was already evident that digital technology was becoming a major force in the field of mathematical computing, and was possibly beginning to eclipse analogue methods. It is interesting to note the comment made during the discussion at the end of Session I by Ross Blunden, i.e., that he 'felt that as some of the delegates to this conference would not be attending further sessions some comment on the relative scope and usefulness of digital and analogue machines should be made at this stage. The digital machine appeared to be superseding the analogue computing instrument mainly because of its greater accuracy and versatility' (Proceedings, p. 54).
The emergence of the new field of Computer Science
Professor John Madsen's initial introductory comments (Proceedings, p. 4-5) are also interesting in that he noted that computing per se had been traditionally viewed as a sub-discipline of mathematics, and mathematicians as both "producers" and "users" of mathematical theory had recently combined to stimulate the progress of computing methods and techniques. One can interpret in his comments a dawning awareness of the emergence of computer science as a discipline and profession in its own right - not just a tool of mathematicians (see also, Myers in Bennett, et al. p.14).
Madsen acknowledged the influence of business machine manufacturers on the development of computers with their long history of the development of business machines (mainly digital machines) for commerce and, in particular, their quest for higher capacity machines and higher speeds. He noted how these developments had extended to meet the demands of industry and agriculture and more recently scientific research. He also mentioned the incentive provided by the Second World War for the development of computers for both military purposes and more general applications (he noted here, in particular, the use of analogue machines for this). Computers were still essentially seen as mathematical devices. There was no mention yet of computers for broader applications outside of commerce and science, for example, as word processing machines, or audio or video machines and 'higher-level' applications.
Douglas Hartree's introduction to automatic calculating machines
Madsen's introductory remarks were reinforced by the special guest to the conference Professor Douglas Hartree, a professor of Mathematical Physics at Cambridge University, and an expert in numerical analysis. Hartree was one of the world's leading figures in computing, and had worked on both analogue and digital computers. After having studied the Bush Differential Analyser at MIT he built the first differential analyser in United Kingdom in 1935. He had also taken part in the development of automatic digital machines which involved extensive travelling in the USA and Europe. He was well placed to give an informed overview of the latest developments in entire field of computing. In fact, out of 13 papers presented at the conference, four of the papers were given by Hartree covering the entire field of computing including analogue computing, digital computing, numerical analysis and programming. In addition Hartree was involved in many of the discussions following the presentation of other conference papers. As indicated on the 'Acknowledgments' page of the Proceedings, Hartree 'contributed unsparingly to the discussions, and whose presence assured the success of the conference' (p.210).
In his first paper 'Introduction to Automatic Calculating Machines' (Proceedings, p. 10) Hartree firstly distinguished between what he terms instruments (or "analogue machines") and calculating machines (or "digital machines") and drew attention to their respective limitations - much more evident in his discussion of analogue devices than digital devices. He then went on to describe the operation of the differential analyser of which he was a major developer. A differential analyser is essentially an analogue mathematical instrument used 'for obtaining solutions of differential equations by mechanical means' (Proceedings, p. 12) and is applicable for a wide range of scientific and technical problems, for example, chemical kinetics, fluid dynamics, meteorology and atomic structure. Nevertheless, despite its utility it was not an automatic device and was limited in its accuracy and applicability.
Hartree went on to discuss 'automatic general-purpose digital machines' (Proceedings, p. 13-15) in which he stressed that "the main developments of calculating machines in the last ten or twelve years have been in digital machines with two features, which are expressed by calling them automatic and general purpose or universal". He gave as examples the Harvard Mark I Machine and the ENIAC, which however were not yet "stored-program" machines. Harteee continued on to give a brief outline of the standard basic units of any general purpose or automatic digital machine, and their interrelations, e.g. the control unit, arithmetic unit, main store and auxiliary store.
Finally Hartree (Proceedings, p. 15-17) talked about 'recent developments in digital machines' by contrasting the above digital machines with the EDSAC, the ACE and the CSIR Mark I (CSIRAC) which were all stored-program machines. Hartree regarded the capacity for automatic stored-programming as 'such a profound influence on the process of organising calculations for the machine that I regard machines which provide it as forming a new stage in the development of automatic calculating machines, essentially different from the first stage, now past, formed by the machines which do not provide it. It is incidentally, the main point of principle which was not foreseen a hundred years ago by Babbage' (Proceedings, p. 17). Hartree concluded his presentation with a discussion on various forms of electronic information storage such as mercury delay lines, cathode ray tubes, and magnetic drums.
Trevor Pearcey and the "Radiophysics Mk. I Automatic Computer" (CSIRAC)
In addition to the four papers presented by Douglas Hartree, four papers were given by Trevor Pearcey from the CSIRO Department of Radiophysics - between them, these two participants presented over half of the papers at the conference.
Of the four papers given by Pearcey, three were specifically related to CSIRAC. There is no doubt that CSIRAC (still known then as the "Radiophysics Mk. I Automatic Computer", see Proceedings, p.42) was a major focus of the conference and that Pearcey was a conspicuous presence as he carefully, clinically and clearly outlined the development, design, logic and operation of a brand new technology with its extraordinary versatility, speed of operation and accuracy. The high profile of CSIRAC was aided by the fact that along with other commercial calculating devices and computing equipment it was available and on display at the conference. Its sheer size alone - far larger than any other item on display - guaranteed that it was a focus of attention for the conference attendees. Although it is not mentioned in the Proceedings, the Radiophysics Mk. I even played some elementary music at the time of the conference (acknowledged as being the first computer music ever played, as subsequent historical research has shown).
Some enduring issues discussed at the Conference
A number of the papers and discussions necessarily concentrated on the hardware and logical design of the respective computing devices. Programming was also a topic discussed at some length. In addition, some topics which remained of enduring interest for many years were raised and examined.
Douglas Hartree noted that 'the subject of automatic digital calculating machines is a new one [and] the terminology has not become standardised' (Proceedings, p. 29). New words were being coined and old words redefined by various research teams. To facilitate communication and avoid confusion there was a need to monitor and be systematic about the use of novel terms and concepts.
One of the issues raised that remained relevant for decades was whether to put more emphasis on building a simple computer which would require more processing steps but would be faster, or building a more complicated computer which would require less processing steps but would be slower. The issue was one of a trade-off between hardware and software, or between machine and human efficiency. A decision had to be made on whether to attempt to write more sophisticated programs to make the calculations faster and allow the computer to do less work, or conversely, to write less sophisticated programs which would make the calculations slower, letting the computer do more work. On this point Hartree judged that 'Here there seems to be a difference of policy between the group at the Mathematical Laboratory at Cambridge and the group at CSIRO Radiophysics in Sydney. The policy at Cambridge is to simplify the engineering at the cost of lengthening the programming...The policy at Sydney is to shorten the program by incorporating more components in the machine' (Proceedings, p. 73).
Other issues of long-term currency that were discussed in a similar vein included the reliability, accuracy, speed and applicability of various calculating machines and the desirability for error correcting codes.
Another paper by Trevor Pearcey addressed the topic of Programming for Punched Card Machines which was one of the promising post-War technologies which was ultimately largely superseded by the electronic automatic digital computer. A similarly interesting paper from an historical point of view was titled 'Digital-Analogue Conversions' by Ross Blunden who reasoned that 'there appears to be a very promising new field for machines combining certain features of each' (Proceedings, p. 185). Again, like the punched card machines, this promising new direction in research was rendered relatively inconsequential following the digital "revolution".
Finally, various supplementary computing components were described and discussed at the conference, both digital and analogue. These included techniques and devices such as magnetic switching methods, acoustic delay lines, vacuum tubes, and notably, the magnetic drum store built for CSIRAC by Brian Cooper.
The Proceedings of the 'Conference on Automatic Computing Machines' is a valuable and irreplaceable record of a seminal conference that marks the beginning of computing science as a discipline and a profession in Australia.
It would be difficult to disagree with Professor David Myers, one of the organisers of the conference, who later wrote that 'Many would agree that the 1951 conference was a turning point in Australia and that since that time computing became increasingly a study in its own right rather than a plaything of specialists. Anyone wishing to consider the state of the art at the end of the first half of this century would be well advised to read the report of that conference' (Myers in Bennet et al, 1994, p.14). The other main organizer of the conference, and a principal participant, Trevor Pearcey, concurred unequivocally that the conference 'represented a significant moment for Australian computing' (Pearcey in Bennet et al, 1994, p. 28).
Conference on Automatic Computing Machines (1951). Proceedings of Conference on Automatic Computing Machines held in the Department of Electrical Engineering, University of Sydney, August 1951. Sydney: University of Sydney