Finding Comfortable Settings of Video Games - Using Game Refinement Measure in Pokemon Battle AI

This paper explores an innovative way to find a comfortable setting of video games. Pokemon is chosen as a benchmark and game refinement theory is employed for the assessment. Pokemon has many entertaining factors to be considered like other video games. Interestingly, there are some factors which are still remained after the first episode of Pokemon was released in 1996. Among them, the number of Pokemon that one trainer can carry (n = 6) has never been changed. It is assumed that the setting may be optimal, which has been realized by players as well as composers. In this paper, Pokemon battle game is simulated by simplifying many factors and different types of computer players (AIs) are developed for the experiments. The experimental results show that the setting (n = 6) is the best in the sense that the setting covers most players of various levels and that Pokemon is mostly played by children. Moreover, there are other reasonable settings for some specific levels. For example, n = 5 and n = 7 may be a comfortable setting for experts and beginners, respectively.