computational process

Vladimir Propp’s analysis of Russian folk tales is known to have produced a semi-formal description of the structure of these tales that has acted as inspiration for several story generation systems, both sequential and interactive. Its exhaustive description of the constituent elements of tales of this kind, and the enumeration of the patterns they follow provided a very useful starting point for researchers looking for computational implementations of story generators. However, it is less generally known that in his book Propp also proposed a procedure for the generation of new tales based on his analytical framework. Although this generative procedure is much less formal than its analytical counterpart, it is one of the first existing instances of a creative process described procedurally. Of particular interest for the field of creative story generation is the number of issues that are declared relevant but not explored in detail.The present paper revisits Propp’s description and focuses on the task of generating the sequence of character functions that determine the plot of the tale. For this task, a number of possible computational implementations are explored, in search for those that produce better results in terms of a number of simple evaluation metrics inspired by Propp’s formalism.

Unlike digital poetry, which has pursued process-intensive directions throughout its history, the dominant directions of digital fiction make relatively light use of computational processes. Whether one looks at the traditions of hypertext fiction, interactive fiction, or video games, the primary model is a set of connections (traveled in different manners) between largely static chunks of language. This panel explores a set of alternatives to this model. The suggested potential panelists include the author of the first book on this topic, published in 2009 (Wardrip-Fruin); one of the authors of Facade, the first fully realized interactive drama (Mateas); the creator of Curveship, a new interactive fiction tool that introduces discourse-level variation as a first-class parameter (Montfort); a prominent author, commentator, and tool builder (Short); the author of Blue Lacunae, a vast, highly variable interactive fiction (Reed); the creator of new algorithms for literary variability based on conceptual blending (Harrell); and the author of the mainstream game industry's most ambitious project in this space, Far Cry 2 (Redding). Topics will include the variety of approaches to wide variability, including those focused on character, event, and language. Panelists will also discuss the issue of what is explicitly represented in the system, or even what can be reasoned about or produced by the system, as compared with what is implicitly present via authoring of data (or selection of data). Promising directions for this emerging set of approaches, especially new works and tools on the horizon, will round out the topics.