Testing correctness of strategies in game-playing programs
Max Bramer · International Joint Conference on Artificial Intelligence · 1979
This paper considers the distinction between winning strategies in game-playing programs which are either 'optimal' or 'correct'. The relative merits of these two types of strategy are considered and methods are proposed for producing correct algorithms by a process of iterative refinement based on an analysis of 'win-trees'. An example is given of the development of a fully correct strategy for the King and Rook against King chess end game.