Dynamic Adaptation of Difficulty in Chess

Most turn-based single-player games allow the difficulty level to be set by the player and provide multiple levels of difficulty, with options most often ranging from very easy to very hard. This basic way of managing the game's difficulty is very fast and easy, but at the same time is extremely clunky and non-adaptive. Driving the difficulty by hard-coded variables can disrupt player satisfaction-it is difficult for the player to choose one of several game levels that he has not played against and has no way to compare. The paper presents a comparison of various methods for dynamic difficulty adaptation in turn-based games, where control based on rules is chosen as the fastest and least computationally intensive method for tailoring game difficulty. It describes a model for adaptive move selection in a chess game with a certain difficulty applying predefined rules for adjusting the difficulty based on specific player actions. The algorithm for managing the difficulty of playing chess is implemented in Python, and Stockfish is used as the game engine. The results of experimental validation of the algorithm with chess players of different experiences are presented, in which the effectiveness of dynamic adaptation of the game difficulty and its positive influence on the gaming experience was evaluated based on a survey. The achieved results can be used to design a compact and fast adaptive chess game for mobile devices that do not require the usage of large computing resources.