Murder She Programmed: Dynamic Plot Generating Engine for Murder Mystery Games

There is an increased demand for more interactive and dynamic single-player computer games. Most currently proposed solutions usually focus on multi-linear narrations. The problem with those solutions is that they do not actually create a dynamic game, they only increase player choices; the games themselves continue to follow a fairly strict storyline. In this article a new game engine is proposed that dynamically creates new game plots for a murder mystery based game. The game engine uses Bayesian networks to create a new plot based on a probability map of a typical murder mystery novel. To abstract plot elements for the construction of the Bayesian network a unique morphology similar to the morphology of the Russian structuralist Vladimir Propp is created. A directed graph, which is inferred from the resulting Bayesian network, is used as a knowledge base for a murder mystery game. This game sets up a complete and consistent murder mystery that is solvable with logical inference. Each new game generates a unique plot, a murder mystery with all the attributes that are needed for such a plot to be logically consistent, coherent, and complete, and make sense to the human perspective. This game engine is responsive to preset constraints rather than a pre-authored narration, which opens up numerous possibilities for integration in computer games where the constraints can be manipulated by interactions from game events and player interaction. Quite a large portion of the game market focuses on producing games with a high degree of narration. These games, especially Role Play Games (RPG), are highly interactive with very advanced game engines offering all the best in combat, graphics and sounds that is available on the market. These games are specifically trying to meet a demand for non-linear game interaction. Their answer has been to offer increasingly multi-linear narrations. The problem with that is that the games are still linear: The player is still forced to follow a set line or lines of preset prerequisites to advance or complete the game. This increases the danger of the player becoming stuck because he did not follow the set storyline. More importantly, it ruins the replayability of the games. It is also not responsive to player actions or role development in the game. I propose a dynamic plot generating engine as a first step in solving this problem. The engine utilizes a Bayesian network, which is a probability-driven causally connected …