Using Monte Carlo tree search and google maps to improve game balancing in location-based games
Luís Fernando Maia Santos Silva, Windson Viana, Fernando Antonio Mota Trinta · 2017
Location-Based games (LBGs) are a subtype of digital games that uses the location of players as a key component for playability, including changes to the game state. However, a significant challenge that threatens the development and popularization of LBGs is the game balancing. Since LBGs rely on players' location, it is hard to manually design interactions, challenges, and game scenarios for each part of the world. Thus, the same LBG is likely to present varying difficulty levels depending on the player's location due to differences in terrain, distance, and transport availability. As a result, even modern LBGs show huge balancing differences between regions and they do not explore competition between players like other game genres. In this paper, we present measurements to estimate game balancing in modern LBGs and introduce a method that uses Monte Carlo Tree Search (MCTS) to automatically edit instances of these games to minimize differences in game balancing. Additionally, we present a study detailing the improvements in game balancing when using the proposed method in today's two most popular LBGs (Ingress and Pokémon Go).