An algorithmic approach to analyzing combat and stealth games

Combat and stealth games give players the option of engaging or avoiding enemy agents at different points. Level-design in this context is complex, however, requiring a designer to understand how different design choices impact difficulty under multiple play-styles. In this work we describe a unified algorithmic approach that can perform abstract analysis of both combat and stealth behaviours. Our proposed solution builds on an existing stealth-level analysis tool, incorporating combat activities into the abstraction in order to allow exploration of level feasibility and difficulty. We demonstrate our approach on a non-trivial example level, showing how such a tool can be used to evaluate and control the player experience.