Methods for detection of emergent gameplay

Games are no fun to play if there is a perfect winning strategy so that a player can always win by playing it. Game designers are aware of this, but it is notoriously difficult to predict when such strategies exist for a given rule set. This project is about studying the effects of changing the rules of a game or adding new rules, especially non-traditional ones like giving away points, in order to predict the existence or likelihood of a perfect winning strategy. We are also interested in studying the effects of modifying traditional interactions between individual games within multiplayer tournaments to determine whether these changes, additions, or modifications lead to undesired emergent behaviour. We use virtual robots called agents to evolve strategies to determine whether a game or tournament is susceptible to perfect winning strategies. The research done in this project will help to design more interesting and robust games and tournaments.

Faculty Supervisor:

Dr. Raymond J. Spiteri

Student:

Saeed Torabi Ziaratgahi

Partner:

Experience First Design Inc.

Discipline:

Computer science

Sector:

Digital media

University:

University of Saskatchewan