Investigating Techniques for Balancing Complex Multi-Agent Systems


Consider a multiplayer game where two players compete. These players may vary in skill, but each expect they will have an even chance of victory. Each carry unique items and abilities that they can trigger and use at their choosing. Both players have assessed their chances before engaging, and both believe to have an equal chance of success. After the battle is over, one player has defeated the other easily.
Is the winner's victory a result of their skill, or did they have an unfair advantage? The answer falls within the attributes and rules of the game, which require testing and tweaking by a designer to achieve fairness. However if the qualities that dictate the battle are quantifiable as attributes , then it is reasonable that the fairness can be deduced. The difficulty is in the complexity of interaction between players and their available options. Even the mildest change may have a large emergent effect on the game. Because of this, fine tuning of attributes and game rules is a long process called balance.


Conference Papers (peer reviewed)

[1] Owen Makin and Shaun Bangay. Orthogonal analysis of starcraft ii for game balance. In Proceedings of the Australasian Computer Science Week Multiconference, ACSW '17, pages 30:1-30:4, Geelong, Australia, 2017. ACM, New York, NY, USA. [PDF] [BibTeX]

[2] Shaun Bangay and Owen Makin. Generating an attribute space for analyzing balance in single unit rts game combat. In CIG 2014 : IEEE Conference on Computational Intelligence and Games, August 2014. [PDF] [BibTeX]

[3] Owen Makin and Shaun Bangay. Establishing competitive domination cycles for peer-to-peer game combat. In 2013 IEEE International Games Innovation Conference (IGIC), pages 152-156, Sept 2013. [PDF] [BibTeX]

[4] Shaun Bangay and Owen Makin. Modelling attribute dependencies in single unit game combat settings. In 2013 IEEE International Games Innovation Conference (IGIC), pages 20-26, Sept 2013. [PDF] [BibTeX]