Happy to see one of my very favorite papers in this list, in which a genetic algorithm running on an FPGA finds an exploit in the physical world in order to "cheat" at success: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.50....

I wrote a thesis on reinforcement learning in Warcraft II. Our system started with zero knowledge and learned to beat the existing AIs about half the time.

It did this by sending its single starting unit to attack the opponent's single unit, before either could use theirs to build more.

Two identical, weak units fighting was just a coin flip.

Ah yes, the worker rush. I've seen this cheese in StarCraft as well.