MIT OpenCourseWarehttp://ocw.mit.edu CMS.608 / CMS.864 Game DesignSpring 2008 For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.CMS.608 – 1 April 2008 Notes by Clara Rhee Game TheoryPrisoner’s Dilemma - what is “rational” play? - what is optimal play? - you have to assume that the other player the same decision that you will – that’s one line of argument – so stick to your story - the other line of reasoning says you should minimize your losses – so defect! - “game theory has mostly stuck to economics – it’s marketable! - “it predicts people’s decisions better than probability” - game theory for games works best for games designed for game theory - let’s talk about some cheesy, annoying winning strategies - a lot of discussion about Super Smash Brothers… - “degenerate strategies” - any game that allows “infinite” cycles of damage - optimal strategies don’t allow for decision making, maybe skill - so if you have a comprehensive possibility space (which applies to computer games too!) but there’s a narrow slice that guarantees a win, it’s a problem - important when designing AIs - you can’t just take the optimal strategy - “what would a player do?” - if it’s really an optimal strategy, the game is impossibly hard - otherwise, it can be just too predictable - but players don’t like it when you “let” them win (as in trivially easy) - it can be fun to lose! - optimal
View Full Document