I think one of the major questions people outside of the Operations Research community have about OR is when is it useful in the real world? Obviously OR is right at home in a factory guiding production or at an airline helping determine prices. But OR can be used on a smaller, more personal scale as well. This morning Slashdot highlights a blog post about a genetic algorithm developed to optimize player strategy for the Starcraft 2 video game. Startcraft 2 is a strategy game where players collect resources and workers to build things which can be used to defeat an enemy. The genetic algorithm was created to optimize the initial build strategy based on existing resources. The chromosome is the array of potential actions that can be performed in the game. The fitness function compares the current state and the desired state.
The blog post includes analysis of an interesting proposed strategy that appears to be counter-intuitive to most popular Starcraft 2 strategies. I have never played Starcraft 2, but the post does a good job of explaining why a strategy that seems weak at the beginning yields stronger results in the end. This is one of the hallmarks of optimization: generating solutions that humans cannot or would not generate themselves including solutions that seem inferior in one area but are optimal globally. Not only is this a novel use of OR (to help a user perform better when playing a video game) but it also shows that you don’t need an Operations Research department staffed with PhDs, gigabytes of data, or fancy solvers like CPLEX to solve an optimization problem.