Multi-objective assessment of pre-optimized build orders exemplified for StarCraft 2

Matthias Kuchem, Mike Preuss, Gunter Rudolph
2013 2013 IEEE Conference on Computational Inteligence in Games (CIG)  
Modern realtime strategy (RTS) games as Star-Craft 2 educe so-called metagames in which the players compete for the best strategies. The metagames of complex RTS games thrive in the absence of apparent dominant strategies, and developers will intervene to adjust the game when such strategies arise in public. However, there are still strategies considered as strong and ones thought of as weak. For the Zerg faction in StarCraft 2, we show how strong strategies can be identified by taking combat
more » ... by taking combat strength and economic power into account. The multi-objective perspective enables us to clearly rule out the unfavourable ones of the single optimal build orders and thus selects interesting openings to be tested by real players. By this means, we are e.g. able to explain the success of the recently proposed 7-roach opening. While we demonstrate our approach for StarCraft 2 only, it is of course applicable to other RTS games, given build-order optimization tools exist.
doi:10.1109/cig.2013.6633626 dblp:conf/cig/KuchemPR13 fatcat:o4csbjojsbguvazx4uefjepek4