The winner is the one who makes the next-to-last mistake.
Dietrich Günther Prinz (1903–1989) was a German-English computer science pioneer, notable for his work on early British computers at Ferranti, and in particular for developing the first chess program in 1951, at least 40 years after El Ajedrecista of Leonardo Torres.
Prinz was born in Berlin, Germany, where he studied physics and mathematics, and went to work on electronic design at Telefunken. Having some Jewish parentage and fearing persecution, he left Germany in 1935 with the rise of Nazism and settled in England. In 1936 he joined the Research Laboratories of the General Electric Co. at Wembley, where he worked in the valve development laboratory. In 1947, Prinz was recruited to the Ferranti factory at Moston, Manchester, for a team to study the potential uses of electronic computers. After Ferranti was awarded a contract to build a production version of the Manchester computer, which would become the Ferranti Mark 1, the world’s first commercially available general-purpose electronic computer, Prinz worked closely with the University of Manchester team.
Prinz had learned programming on the Mark I (It was quite a difficult task because Mark I didn’t have an assembly language that would have allowed simple helpful mnemonics for data. Instead, the input was very much based on base-32 representation, putting strings of base-32 instructions or constants into base-32 positions in the store.) from seminars led by Alan Turing. Influenced by him, and later by other colleagues, Prinz came to see chess programming as “a clue to methods that could be used to deal with structural or logistical problems in other areas, through electronic computers”. Turing had also worked out an algorithm for playing chess, but Prinz’s work was independent of this. As Mark I was inadequate to play a complete game of chess, Prinz concentrated on the endgame. In November 1951, his program on the Ferranti Mark I first solved a Mate-in-Two problem.
In order to have a program working in the shortest possible time, a few restrictions were imposed on the rules of chess as they were “explained” to the machine. For example, castling was not permitted, nor were double moves by pawns, nor taking en passant nor the promotion of a pawn into a piece when it reached the last row; further, no distinction was made between mate and stalemate.
The program contained a routine for the construction of the next possible move, a routine to check this move for legality, and various sequences for recording the moves and the positions obtained. All these separate subroutines were linked together by a master routine which reflected the structure of the problem as a whole and ensured that the subroutines were entered in the proper sequence.
The program, as well as the initial position on the chess board, was supplied to the machine on punched tape and then transferred to the magnetic store of the machine. An initial routine (sub-program) was transferred to the electronic store, and the machine started its computation. The program was so organized that every first move by white was printed out; after the key move had been reached the machine printed: “MATE.”
The main result of the experiment was that the machine is disappointingly slow when playing chess—in contrast to the extreme superiority over human computers where purely mathematical problems are concerned. For the simple example given in the position reproduced here, 15 minutes were needed to print the solution. A detailed analysis shows that the machine tried about 450 possible moves (of which about 100 were illegal) in the course of the game; this means about two seconds per move on average.
A considerable portion of this time had to be used for a test for self-check (i.e. after a player had made a move, to find out whether his own King was left in check). This was done by first examining all squares connected to the King’s square by a Knight’s move, to see (a) whether they were on the board at all, (b) whether they were empty or occupied, (c) if occupied, by a piece of which color and (d) if occupied by a piece of the opposite color, whether or not this piece was a Knight. A similar test had to be carried out for any other piece that might have put the King in check. This test involves several hundreds of operations and, at a machine speed of 1 msec per operation, might take an appreciable fraction of a second.
The next important time-consuming factor was the magnetic transfers, i.e. the transfers of sub-programs and data (relating to positions and moves) between the magnetic and the electronic store. It is here that improved programming techniques may save time by better utilization of the electronic store, thus reducing the number of transfers (nine for every legal move in the present program).
Compared with these two items, the time spent in computing the moves appeared to be of minor importance although the machine not only computed the possible moves but also the impossible, but “thinkable” moves—meaning those which either carry the piece off the board or lead to a collision with a piece of the same color already on the square. These moves, however, were quickly rejected by the machine and did not contribute greatly to the total computation time.
What was the first chess problem solved by a computer? The task set for the Mark I machine was to find a move by white that would lead to a mate in the next move, whatever black might answer (see the upper image). The move is R—R6.
Biography of Dietrich Prinz
Dietrich Günther Prinz was born on 29 March 1903 on Elsasserstraße 1/2, Berlin, Germany, to the German lawyer Georg Prinz and his wife Erna (Bukofzer) Prinz, a Jew. Dietrich had a sister, Ilse Rose (born 1904).
Prinz was educated at the Humboldt University of Berlin in the early 1920s, where his teachers included Max Planck and Albert Einstein. He gained his doctorate (Ph.D. in Physics) under Einstein around 1930 (one of his class reports was signed by Einstein). In his student years, Prinz got his first patents—DE528902 from 1926 for Circuit with continuous change of a voltage applied to the arrangement, DE568860 from 1928 for Arrangement for changing (modulating) the amplitude or frequency of electrically generated musical tones in the rhythm of an alternating voltage, and DE506689 from 1928 for Electric musical instrument.
From 1930 to 1935, Dr. Prinz was working for Telefunken, a German radio and television company, and filed several patents for: Means of synchronizing the horizontal and vertical scanning frequencies in a television set; Single tune radio receiver; Loudspeaker that could be used for the electric rendition of phonograph records and also for broadcast reception; Vacuum tube for short waves.
In 1935, Prinz, knowing what he can expect from Nazies as a half-Jew, left Germany and settled in England, where in January 1936, he joined the General Electric Company (GEC) Research Laboratories in Wembley, on the outskirts of London. He worked on the development of vacuum tubes and was assigned to the Valve Development Laboratory. During World War II, Prinz was interned in Canada, just like Josef Kates. While in the New World, in 1942, Prinz wrote an article called “Class C Telegraphy,” for Wireless Engineer magazine. In April 1944, Prinz authored a paper called “Contributions to the Theory of Automatic Controllers and Followers,” published in the Journal of Scientific Instruments.
In 1945, Prinz returned to work in Leeds for Bowen Instrument Company. By 1945, he had several British patents such as improvements in mercury arc electric converters and improvements in electric voltage measuring instruments. In 1947, Prinz was hired by the instrument department at Ferranti Ltd, a major UK electrical engineering and equipment firm known primarily for defense electronics (radar sets, avionics, and other military electronics), but also for television sets, radios, and electric clocks.
In October 1948, the British government commissioned Ferranti Ltd to manufacture a commercial machine to Frederic Williams’ specifications. In September 1948, Prinz visited the United States to assess computer developments on behalf of Ferranti. He also visited Cambridge University to examine the Electronic Delay Storage Automatic Calculator (EDSAC), the first practical stored-program electronic computer.
In 1949, Prinz helped develop the Digital Automated Tracking and Resolving (DATAR) system. This was a computerized information system that combined radar and sonar information to provide commanders with an “overall view” of a battlefield. Prinz was also studying methods of high-speed data transmission.
By the late summer of 1949, the detailed design of the Manchester Mark 1 was already being transferred to Ferranti. In 1950, Prinz wrote an article for Nature magazine called, “A Relay Machine for the Demonstration of Symbolic Logic.” He and Wolfe Mays built the first electrical relay logic machines in the United Kingdom.
In 1950, Donald Davies published an article called “A Theory of Chess and Noughts and Crosses” in Penguin Science News. This article influenced Prinz to look at computer chess programming. In the same year, he was attending seminars by Alan Turing and Cecily Popplewell and learned about computer programming, which led him to the creation of the first practical chess program. In 1951, Prinz took part in the design of the NIMROD computer of John Bennett, to play the game of NIM. The computer was first displayed during the 1951 Festival of Britain. Prinz and Ian Howard, an undergraduate at Manchester University, played the game of Nim on the machine during the festival. In October 1951 Prinz demonstrated NIMROD in the British pavilion of the German Industrial Exhibition in Berlin (see the nearby image).
From 1951 to 1953, Lord Bertram Vivian Bowden (1910-1989) was responsible for selling early computers manufactured by Ferranti. In his 1953 book, Faster than Thought, he dedicated a chapter to Prinz’s chess program and described the inventor as talented and extremely modest man.
In 1951, Prinz wrote up the first instruction manual for the Mark I, called “Introduction to Programming on the Manchester Electronic Digital Computer.” He was also becoming an expert at linear programming. In 1952, Prinz published an article called “Robot Chess” for Research magazine. By 1952, he was Head of Programming at Ferranti. In 1955, Prinz and David Caplin programmed a Ferannti Mark I computer to play Mozart’s Musikalisches Wurfelspiel. In 1966, Prinz wrote a paper on the possibility of computing odds for place betting.
Prinz married late in his life, in early 1951, to Käthe Anna Hedwig (Schünemann) Prinz (born 27 August 1909 in Berlin). They had a son—Jonathan Franklin Prinz (born in 1953) and a daughter, Daniela Prinz Derbyshire (born in 1963). Dietrich Günther Prinz died on 15 December 1989 in Manchester, England.