Around 1957, a team of IBM programmers led by Alex Bernstein - a mathematician and chess player - wrote what is generally regarded as the first computer program able to play a complete game of chess. It ran on an IBM 704, one of the last vacuum-tube mainframes, and it played a recognizable, if weak, game from the opening to checkmate.
Bernstein’s program followed Claude Shannon’s “Type B” idea rather than brute force. Instead of examining every legal move, a set of “plausible move generators” picked roughly the seven most reasonable moves in a position, ranked by goals such as king safety, development, and attack or defense. The program then searched four plies (two moves for each side) ahead and scored positions on material, mobility, area control, and king defense. A single move took about eight minutes on the hardware of the day.
The Computer History Museum holds film of Bernstein demonstrating the program, drawn from a 1958 Educational Testing Service broadcast called “Horizons of Science: Thinking Machines.” The program was no match for a strong human, but it established that a machine could carry a full game of chess by itself - a stepping stone from Shannon’s 1950 paper toward the tournament-strength programs that followed.