[Using evolutionary algorithms to make a walkthrough for the light-bot game with C# at Chris' Babbling Blog](http://splinter.com.au/blog/?p=38).

This is cool. I’ve always been very interested in genetic/evolutionary algorithms and one of the ways I’ve wanted to practice is writing a simulation of a creature that figures out how to walk a path. This guy used this same idea to solve a real game. Nice stuff, although I wish there were some more technical details. (He does provide the code though).

A cool thing he sees is something a lot of GA research finds, that the algorithms often evolve unintuitive or extra-complex solutions. Just as in real life, evolution is only interested in solving the problem.