A five-year plan for automatic chess

Young animals play games in order to prepare themselves for the business of serious living, without getting hurt in the training period. Game-playing on computers serves a similar function. It can teach us something about the structure of thought processes and the theory of struggle and has the advantage over economic modelling that the rules and objectives are clear-cut. If the machine wins tournaments it must be a good player. The complexity and originality of a master chess player is perhaps greater than that of a professional economist. The chess player continually pits his wits against other players and the precision of the rules makes feasible a depth of thinking comparable to that in mathematics. No program has yet been written that plays chess of even good amateur standard. A really good chess program would be a breakthrough in work on machine intelligence, and would be a great encouragement to workers in other parts of this field and to those who sponsor such work. In criticism of the writing of a chess program, Macdonald (1950) quoted a remark to the effect that a machine for smoking tobacco could be built, but would serve no useful purpose. The irony is that smoking machines have since been built in order to help research on the medical effects of smoking. This does not prove that a chess program should be written, but suggests that the arguments against it might be shallow. Many branches of science, and of pure and applied mathematics, have started with a study of apparently frivolous things such as puzzles and games. It is pertinent to ask in what way a good chess program would take us beyond the draughts program of A. The answer is related to the much greater complication of chess, the much larger number of variations and possible positions. In fact, the number of possible chess positions is about the cube or fourth power of the number of possible draughts positions (see Appendix E). Samuel was able to make considerable use of the storage of thousands of positions that had occurred in the previous experience of the machine, and this led to a very useful increase in the depth of analysis of individual positions. The value of this device depends on the probability that, at any moment in the analysis, we run into a position that has already been analysed and stored. This applies more generally to the goals and subgoals that occur to the chess player. Thus there should be'specificallydirected' as well as'routinely-directed' analysis. Another important aspect of chess thinking, also required in most other problem-solving, is what de Groot (1946, 1965) calls'progressive deepening' of an analysis. Typically an analysis of a position by a human player does not simply follow a tree formation, but contains cycles in which a piece of analysis is retraced and improved.