Using an algorithm (or ml model) to find out the best game design

[bhauptvogel]

Programming a game bot that plays itself with given rules (a lot of times) to figure out with which starting configurations (piece types, piece amounts etc.) the game is the most fun, for example using heuristics such as:

• the less often the bot ties against itself the better
• the more different strategies the bot employs the better
• the more likely it is the bot can sometimes loose against an inferior opponent the better (variance), but only to a certain degree

Idea: doing experiments with it and writing something about it