1950

Chess Computer

Alan Turing (1912–1954), Claude Elwood Shannon (1916–2001)

In 1950, American mathematician Claude Elwood Shannon wrote a paper about how to program a computer to play chess. In 1951, British mathematician and computer scientist Alan Turing was the first to produce a program that could complete a full game. Since then, software engineers have improved the software and computer hardware engineers have improved the hardware. In 1997, a custom computer called Deep Blue, developed by IBM, beat the best human player for the first time. Since then, humans have not had a chance because computer chess hardware and software keeps improving year after year.

How do engineers create a computer that can play chess? They do it by employing machine intelligence, which in the case of chess is very different from human intelligence. It is a brute force way to solve the chess problem.

Think of a board with a set of chess pieces on it. Engineers create a way to “score” that arrangement of pieces. The score might include the number of pieces on each side, the positions of the pieces, whether the king is well protected or not, etc. Now imagine a very simple chess program. You are playing black, the computer is playing white, and you have just made a move. The program could try moving every white piece to every possible valid position, scoring the board on each move. Then it would pick the move with the best score. This program would not play very well, but it could play chess.

What if the computer went a step further? It moves every white piece to every possible position. Then on each possible white move, it tries every black move, and scores all of those boards. The number of possible moves that the computer has to score has grown significantly, but now the computer can play better.

What if the computer looks multiple levels ahead? The number of boards the computer has to score explode with each new level. The computer gets better. When Deep Blue won in 1996, it was able to score 200 million boards per second. It had memorized all common openings and gambits. It could prune out vast numbers of moves by realizing certain paths were unproductive. Today the computational power in a laptop smart phone allows it to beat most people at chess using the same techniques.

SEE ALSO Microprocessor (1971), Smart Phone (2007), Watson (2011), Brain Replication (c. 2024).

Pictured: IBM’s supercomputer, Deep Blue.