Player-customized puzzle instance generation for Massively Multiplayer Online Games

Abstract

The current generation of MMOGs faces an increasing problem to ensure customized game content for their players. The existing approach, that of growing teams of human designers, does not scale. In contrast, we set to investigate the problem of automated player-customized puzzle instance generation for MMOGs. In this work we have focused on evaluating puzzle difficulty and matching it to player ability, and on generating fresh puzzle instances. Our experimental results show evidence that our approach can be used to generate puzzle instances of commercial quality. We are currently exploring more puzzle games and more solving strategies to expand our framework. Maintaining and analyzing player detailed player information such as solving ability is time- and resource-consuming [12]; we will investigate mapping puzzle difficulty to player skill levels and history.