Procedurally generating a digital math game’s levels: Does it impact players’ in-game behavior?

Abstract

Developing games can be improved with Procedural Content Generation (PCG), the automatic creation of game contents, such as levels, music, and vegetation. However, few researches have addressed the impacts of PCG on players, especially in terms of games with educational ends. Hence, this understanding is a research area that demands further studies. To expand on this gap, this article presents a study concerning whether PCG for level creation impacts players’ in-game behavior, based on interactions with a digital math game. This game features two versions that contain a unique difference: while one features human-designed levels, these are procedurally generated on the other version. To compare them, PCG’s impact on players’ in-game behavior was measured through the total played time, number of played levels, and number of retained players. In this context, the findings demonstrated a significant difference in the number of retained players, which was higher for the PCG version in comparison to the other. In contrast, the other two metrics were insignificantly different between versions. Therefore, game designers and developers can exploit these findings to employ PCG in games, taking advantage of its impacts on development and players, knowing how it is expected to affect players’ in-game behavior.