Representing dynamic difficulty in turn-based role playing games using Monte Carlo tree search

One of the challenges during game development is to find a way on how to make the players actually enjoy the game itself while being quite hooked by its gameplay. In almost in every game, player must play the game through challenges to complete the game's main objectives. Enjoyment is highest when the game's challenges, which are either hard-coded or adaptively put forth by AI, are of appropriate difficulty with respect to the player's skill. In order to balance out between these two aspects, a difficulty adjustment is needed. In this paper, we study an application of Monte Carlo Tree Search (MTCS) for creating such a balancing using role playing games as the case study. The key idea is the intuition that a game's difficulty is balanced if any of the player or the AI can win or lose the game by only a small margin. We conduct experiment to see if the method is appropriate for this problem.