John Bennett

In 1950 the British company Ferranti Ltd., which produced the world’s first general-purpose commercial computer, the Ferranti Mark 1, had undertaken to display a computer at the 1951 Festival of Britain (a national exhibition and fair that reached millions of visitors throughout the United Kingdom in the summer of 1951). By late 1950, it became evident that this promise could not be fulfilled. Then John Bennett (1921-2010), an Australian scientist, who worked for Ferranti as a computer specialist, suggested that a machine to play the game of NIM (an ancient game of strategy) against all comers should be constructed with a versatile display to illustrate the algorithm and programming principles involved. The design was implemented by a Ferranti engineer, Raymond Stuart-Williams. The computer was called NIMROD and became the first digital computer designed specifically to play a game.

Bennett got the idea of a Nim-playing computer from the Nimatron, an electro-mechanical machine exhibited at the 1939-1940 World’s Fair in New York. Because of the number of vacuum tubes involved, NIMROD was huge—12 feet wide, 5 feet tall, and 9 feet deep, and its weight was three and a half tons. When the Festival of Britain ended, in October 1951, the computer was displayed at the Berlin Industrial Show, again with great success, demonstrated by Dietrich Prinz, a German-English IT pioneer who would succeed, in November of the same year, in realizing the first computer program playing chess. To help explain the NIMROD computer to the British public Ferranti published a pamphlet entitled Faster than Thought. The Ferranti Nimrod Digital Computer. Discovery magazine published an artist’s watercolor impression of the NIMROD in their March 1951 issue.

Later John Bennett reminisced:
“In its simplest form, two players with several piles of, say, matches play the game of Nim. The players move alternately, each removing one or more of the matches from any one pile. Whoever removes the last match wins.”

“The machine was a great success but not quite in the way intended, as I discovered during my time as spruiker on the Festival stand. Most of the public was quite happy to gawk at the flashing lights and be impressed. A few took an interest in the algorithm and even persisted to the point of beating the machine at the game. Only occasionally did we receive any evidence that our real message about the basics of programming had been understood.”

The Nimrod computer did not use a Cathode-Ray Tube display but used a set of fixed lights that turned on and off for a visual to describe what was happening throughout the gameplay. The interesting thing about NIM is that there is a non-obvious strategy that will ensure a win, once it can be applied. Once one player is able to make a “safe” move, the other cannot and must leave it “unsafe”, so that the first can again make it “safe” next time. Furthermore, the strategy involves viewing the numbers of tokens as binary numbers, which makes it perfect for a computer algorithm. NIMROD was an entirely fixed program. Indeed it didn’t really have a “program” in the sense used today—just logic gates wired together to compute a suitable response to each gameplay.

Biography of John Bennett

John Makepeace Bennett was born on 31 July 1921 in the southern Queensland town of Warwick, Australia, the son of Albert John Bennett and Elsie Winifred (née Bourne) Bennett (1887-1962).

John Bennett was educated at The Southport School, then he went to the University of Queensland to study civil engineering. From 1942 until 1946 he served in the RAAF, where he worked on a radar unit on the Wessel Islands and later worked in airfield construction. He then returned to the University of Queensland to study electrical and mechanical engineering and mathematics.

In 1947 Bennett went to Cambridge University to become Maurice Wilkes’ first research assistant as part of the team working to build EDSAC—this was the world’s first practical stored-program electronic computer and the world’s first computer in regular operation from 1949. He used EDSAC to carry out the first-ever structural engineering calculations on a computer as part of his Ph.D. From 1950 until 1955 he worked for Ferranti in Manchester and London as a computer specialist. Here he designed the instruction set for Ferranti Mark 1.

In 1956, Bennett returned to Australia to become a Numerical Analyst (and later Senior Numerical Analyst) at the Adolph Basser Laboratory of the University of Sydney. Until 1958 he taught associated courses in the use of computers. In 1958 he established a Postgraduate Diploma in Numerical Analysis and Computing which was later changed to the Postgraduate Diploma in Computer Science.

In 1961, Bennet became a Professor of Physics (Electronic Computing). In 1972 the Basser Computing Department was split into the Basser Department of Computer Science (for teaching and research) and the University Computer Centre. John Bennett was appointed head of the new Basser Department of Computer Science, but it was not until 1982 that John Bennett’s title was changed to Professor of Computer Science—a title which he held until his retirement.

On 26 January 1952, Bennett married Rosalind Mary Elkington (1926-) (who was also working at Ferranti). They had four children: Christopher John (1953-), Ann Margaret (1955-), Susan Elizabeth (1957-), and Jane Mary (1960-).

In 1986 Bennett, aged 65, retired with his wife to Sydney’s Northern Beaches. He died at home on 9 December 2010.