Thursday, May 26, 2011

Modern Trends in Machine Computation

The paper "Modern Trends in Machine Computation" was published in Supplement to the Australian Journal of Science, 21 February 1948 (Volume X Number 4). This is by Trevor Pearcey, then an officer at the Radio-physics Laboratory of the Council for Scientific and Industrial Research (predecessor of CSIRO). This describes the state of the art in what would now be called computing. The paper starts by discussing the difference between analogue and digital machines. The accuracy and flexibility of digital machines is pointed out. Charles Babbage's difference and analytical engines are mentioned as precursors to model digital machines. The work of Dr. Vannevar Bush at MIT is noted. The difference between simple repetitive computations needed by commerce and complex scientific calculations is discussed.

Pearcey provides a diagram of the components of a digital computer, which is essentially the same as would be used today in teaching, except for the terms used. The diagram (Figure 2) is captioned "The Organic Structure of a Computer", and has a "Control Organ" at the centre, surrounded by and connected to, the "Memory Organ", "Output Organ", "Arithmetic Organ", and "Input Organ". The difference between decimal and binary number representation are then discussed and the advantage of binary for processing by simple telephone type relays. Valve based electronic computer circuits are then discussed, with a circuit of a binary counter showed.

Memory devices are then discussed: punched tape, punched cards, magnetic tape or wire, photographic film, relay and valve registers, acoustic tube delay lines, and electrostatic storage (in a device similar to a Cathode Ray Tube). The timing needed for sequential computation is then discussed.

The Automatic Sequential Control Calculator (ASCC) at Cruft Laboratory, Harvard is then described in detail (it was an electromechanical relay machine). ENIAC from the Moore School of Engineering is then compared with its predecessor, ASSC from Harvard. The MIT Differential Analyser is described last.

Pearcey then speculates on future developments, with more memory capacity and faster speeds being desirable. The the USA's EDVAC project and UK's ACE are mentioned.

Pearcey ends with an often quoted, and from my reading of a facsimile of a copy of the actual journal article, misquoted sentence:
"In the non-mathematical field there is a wide scope for the use of the techniques in such things as filing systems. It is not inconceivable that an automatic encyclopaedic service operated through the national teleprinter, or telephone system, will one day exist."
When asked about this in his talk in Canberra, 23 May 2011, James Gleick, author of "The Information: A History, a Theory, a Flood", said he doubted that such a paper would mention computers in 1948. Having read the paper I can confirm it does discuss "Computors" (note spelling), "computing" and specifically electronic digital computers. The body of the paper does not discuss the use of computers for handling non-numeric data, but the mention "filing systems", and an "encyclopedic service" operating over a national teleprinter, or telephone system, indicates that this was in Trevor Pearcey thoughts.

What also makes this paper significant is that unlike many dreamers who speculated about computers, Trevor Pearcey was leader of the team (along with Maston Beard) who built CSIRAC (CSIR Mk 1), the fourth stored program digital electronic computer (1949).

This paper deserves to be more widely known. The poor quality facsimile copy I obtained through the Australian National University, appears to be from a USA based document service. This paper should be digitised at high resolution and transcribed and placed online, so its place in the history of computing can be properly recognised.

No comments: