Analyzing Modern Computer Game Narratives

In many computer games, narrative is a core component, with the game centering on an unfolding, interactive storyline which both motivates and is driven by the game-play. For larger games, however, narrative size and complexity makes analysis difficult, and while formal approaches to assessing game narratives and identifying flaws have been proposed, scalability remains a barrier to practical use. In this work we develop an optimized, formal analysis system for interactive fiction narratives. Our approach is based on a relatively high-level game language, borrowing analysis techniques from compiler optimization to improve performance. We demonstrate our system on a variety of non-trivial narratives analyzing a basic reachability problem, the path to win the game. The results are very encouraging, and using our optimized design we are able to analyze narratives orders of magnitude larger than the previous stateof-the-art. This level of performance allows for verification of narrative properties at practical scales.