icosilune

Archive: April 20th, 2009

Historical origins and AI in fictional worlds

[Research] (04.20.09, 5:56 pm)

The following extrapolates some notes taken earlier today.

I have been looking at a lot of AI papers, and these have tended to fall within two camps: modeling characters, and modeling stories. There is also a number of works that fall between these ideas, so they can be seen to represent something of a spectrum of AI work that addresses games and narrative. The group that aims on modeling characters seeks to create intelligent agents within worlds, and these agents generally form their actions according to metaphors of planning. Based on their goals (which may be subjective), they plan actions that will be instrumental in achieving those goals. The second group aims on controlling the story, implementing a story as a system that can be executed within a game. A player’s interaction can be directed to go along with one of the branches or possible directions that the story can take, and usually a director or drama manager can enable or disable parts of the world that will facilitate the player’s progression through the story.

I believe that these approaches are flawed. Hybrids will not solve the problem because there is something missing. Fiction takes place in story worlds, and in simulating these worlds, the most significant element should not be character or story, but mechanics. A story works in a particular way, within fiction, the events that occur happen because of a particular set of values and rules. What is created in fiction is a particular story that happens to occur within a space of many potential stories. It should not be the agenda of adaptation to reconstruct the original story. The character based approach also will fail unless it accounts for the mechanics and values of the fictional world. For instance, Pizzi and Cavazza’s Madame Bovary project takes enormous strides forward in representing emotional logic, but the rules of the world might have been better served by mechanics that explore the sensation of bourgeoisie boredom, dissatisfaction, and romantic disenchantment.

My particular domain of focus has been Jane Austen, and she emerged from the tradition of the epistolary and the picaresque (the traditions of Richardson and Defoe, respectively). Epistolary novels frequently described love, loss, and class boundaries, while the picaresque portrays rogueish realism. An important detail about Austen is that she wrote her novels first in epistolary form, but then rewrote them in third person. This in of itself is a form of adaptation, but also frames the story as taking place within a coherent world, supported not only by narration, but also by a “backstory” of letters, essentially in-world communication. This idea follows in the tradition of the Brontë family, who began writing only after having created an entire imaginary world, originally inspired by toy soldiers. In this tradition, the stories are not the original works, but threads that emerged from this rich and deep world. What is most fascinating is how the evolution of the Brontë’s imaginary world resembles, with stunning similarity, the emergence of tableop roleplaying from war games. This tradition of building imaginary story worlds has continued, and can be seen in the cultures surrounding multiplayer games, and also appears within literary works as well, for instance Faulkner’s imaginary Yoknapatawhpa County.