We present a variation of the genetic programming algorithm, called Historically Assessed Hardness (HAH), in which the fitness rewards for particular test cases are scaled in proportion to their relative difficulty as gauged by historical solution rates. The method is similar in many respects to some realizations of techniques such as implicit fitness sharing, stepwise adaptation of weights and...

When presented with a hammer that is being used as a paperweight, most people judge that its proper function is to pound nails in place rather than to hold down pieces of paper. The bias to assign proper function based on historically intended function rather than on manifest or current function, can be explained as a memory effect (we have learned that hammers are used to pound nails), as abdu...