Lewis Cup 2012/13

The Pioneers of Computer Chess Programming

The American mathematician, cryptographer and electrical engineer engineer Claude Shannon is widely acknowledged as the founding father of Information Theory. During World War Two Shannon met and discussed his ideas with a wide circle of eminent scientists and engineers including Einstein, Turing and von Neumann.In 1949 he produced the seminal work 'Programming a Computer for Playing Chess', published in 1950 it is recognised as the first academic treatise on the subject and describes how a machine could be made to play a reasonable game of chess. He is pictured with chess master Edward Lasker demonstrating a very early chess computer that was programmed to his specification. Because of the limitations of 1950's technology this machine was only capable of playing simple end-games and wasn't very strong; it is now in the Museum of the Massachusetts Institute of Technology.The British mathematician and cryptographer Alan Turing produced a paper-based chess playing program in 1950. The following year he tried to implement it on a Ferranti Mark I computer at Manchester University but unfortunately he died before he could complete the task. His colleague, Dr. Prinz, wrote a chess playing computer program for the Ferranti that could solve simple mates-in-two problems and ran it in November 1951. The program would examine every possible move until a solution was found and it took about fifteen minutes to solve a problem. In 1950 the Hungarian/American mathematician John von Neumann designed and built the giant MANIAC I computer for Univac who claimed it was the most powerful computer in the world. In 1956 MANIAC I was programmed by Stan Ulam to play chess using a 6x6 board; no bishops and only six pawns per side. It took twelve minutes to search four moves deep; it is estimated that adding the bishops and additional pawns would have extended the analysis time to three hours per move to search to this depth.

In 1957 Alex Bernstein lead a team at the Massachusetts Institute of Technology that created the first really complete chess program on an IBM 704. It took about eight minutes to make a move. Edward Lasker played the program, easily defeating it, but he commented that it played a ‘passable amateur game.’

Earlier this year (2012) Mathias Feist from ChessBase produced two chess engines based upon the ideas proposed by Shannon and Turing. He identified both engines as being very weak by modern standards but interestingly both proved to be stronger than the early commercially available chess machines that appeared on sale in the late 1970's. For details of Feist's work see http://www.althofer.de/shannon-turing-exhibition-match.pdf