Nested Look-Ahead Evolutionary Algorithm Based Planning for a Believable Diplomacy Bot

With regard to literature, improved estimations for the number of possible moves and placements are provided, showing that the complexity of Diplomacy is enormous, making it a good candidate for machine learning and evolutionary learning techniques. To enhance the playing strength of an existing Diplomacy bot and alleviate the distance to the presumed best current bot, a look-ahead planning component based on nested evolutionary algorithms, is then implanted into an already existing bot. The experimental investigation shows that the resulting bot is significantly improved.