Modern video games have highly developed computational models of physical space, which allow sophisticated play in the physical realm. However, computational models of social interaction are rare, offer limited social play, and require a large amount of authoring to create. We believe that a computational model of social interaction inspired by appropriate humanities and social science concepts could help alleviate these problems and open up new areas of social play. In this paper, we describe a playable model called Comme il Faut that uses a social artificial intelligence system particularly inspired by Goffman’s dramaturgical analysis and Berne’s psychological games, constructed for authoring power rather than fidelity with the everyday world. Our theoretical basis, the system’s relation to other digital media and games, and its implementation are presented to explain Comme il Faut and our approach to enabling social play.