How can game mechanics be automatically generated which can be understood and used by human players?
Procedural generation has become a fundamental element in many games. It focuses on content such as levels or aesthetic content, but research into the generation of game mechanics has been neglected. Previous approaches have often generated mechanics that enable the player to reach a predefined goal. But this functional approach normally leads to unusable results for human players.
In this Master thesis, two existing approaches have been implemented anew and carefully studied. Additionally, a novel game mechanic generation pipeline is proposed, which is directly implemented inside a game engine. It focuses on integrating the generation of game mechanics as a game element itself. The resulting game mechanics do therefore not necessarily lead to a goal but are learnable for human players through direct interaction with the game.