A Computational Method of Optimizing Chess Compositions to Enhance Aesthetic Appeal

Chess compositions are typically positions set up with a stipulation, e.g., ‘White to Play and Mate in 3’. They are also usually challenging, educational or possess some aesthetic appeal. In this article, a computational method of optimizing compositions is presented that can help improve them, primarily in terms of aesthetics as described by the conventions of problem composition. This is useful in automatic chess composition, which is fairly new in artificial intelligence (AI) and still a largely open area of research. The method involves seven aspects, namely 1) removing pieces, 2) swapping white pieces for weaker ones, 3) swapping black pieces for stronger ones, 4) moving the white king to avoid check, 5) looking for a shorter mate, 6) moving the white king closer to the other pieces, and 7) preventing major black pieces from being en prise (capturable). All of these are applied in order and together whilst preserving the original solution as much as possible. The effectiveness of the method is demonstrated using examples (before and after optimization) illustrating the positive changes to the positions. Human expert assessment is also shown to be comparable in outcome. This method can therefore be used to automatically improve, in general, the aesthetic appeal of compositions to a significant or noticeable degree.