The upshot of the No Free Lunch theorems is that averaged over all fitness functions, evolutionary computation does no better than blind search (see Dembski 2002, ch 4 as well as Dembski 2005 for an overview). But this raises a question: How does evolutionary computation obtain its power since, clearly, it is capable of doing better than blind search? One approach is to limit the fitness functions (see Igel and Toussaint 2001). Another, illustrated in David Fogel’s work on automated checker (...) and chess playing (see, for instance, Chellapilla and Fogel 1999 and Fogel et al. 2004) and, more recently, given a theoretical underpinning by David Wolpert and William Macready (2005), is to limit optimization problems to search spaces consisting of agents that play competitively against one another. In this brief note, I focus on the latter attempt to get around the force of No Free Lunch. (shrink)