Discovering Unique Game Variants

We present a method for computationally discovering diverse playable game variants, using a fine-tuned exploration of game space to explore a game’s design. Using a parameterized implementation of the popular mobile game Flappy Bird, we vary its parameters to create unique and interesting game variants. An evolutionary algorithm is used to find game variants in the playable space which are as far apart from each other as possible, helping find the most unique versions; we also use a clustering algorithm to find representative variants. Manual playtesting confirms that the discovered game variants are playable and significantly different to the original game in challenge, game feel, and theme.