Marisa completed her thesis in the spring of 2009. She studied Wolpert and Macready's 1997 paper on NFL theorems for optimization, extended one of their theorems (for choosing procedures) in a natural way, and then showed that, given a reasonable assumption (that has nothing to do with the underlying optimization algorithms or the function to be optimized over), there exists a choosing procedure that is better than random guessing. This stands in stark contrast to the typical conclusions that have been drawn from NFL theorems for optimization and gives some reason to question them.