Evaluating the enjoyability of the ghosts in Ms Pac-Man

The video games industry is one of the fastest-growing industries in the world, bolstered by sophisticated technology in gaming consoles and modern trends such as mobile and social gaming. The goal of most video games is to entertain the gamer and in most games this stems from the interaction between the gamer and the non-player characters (NPCs): it is no longer sufficient for a game to be visually appealing but instead, the gamer must be challenged at the right level of difficulty to be engaged by the game. It is thus necessary to develop suitable NPCs that are fun to play against and the realm of computational intelligence offers a variety of techniques to do so. However, the perception of fun is clearly subjective and indeed, difficult to quantify. In this paper we make use of the Ms Pac-Man vs Ghosts gaming competition to gather and analyse data from human gamers regarding their preference of opponent: each gamer plays two games against different ghost teams, indicating their preference at the end. We subsequently use this data to establish which ghost teams are generally preferred and demonstrate that there are measurable differences between these ghost teams. These differences are sufficient to group the ghosts into different categories with a good degree of accuracy. This work is a first step in better understanding the attributes required by NPCs for players to be engaged.