We also wish to acknowledge several interesting suggestions for future work from one of the reviewers. The ineffectiveness of the crossover operator deserves further investigation. The need for the structure to support its own weight means that reducing the overall weight of the structure requires a systematic reduction of weight throughout the structure. As a result, recombination between dissimilar parents is unlikely to produce feasible offspring. On the other hand if the parents are too similar then recombination will have little or no effect. Investigation to better understand why the crossover operation tended to be ineffective might help us design a better approach. The reviewer also suggests looking at ways to start out with initial populations that have more variety, without being completely random. We have been thinking more along the lines of introducing variation based on identifying the different “roles” of members in the design, as mentioned above, but we could certainly also look at introducing some variation around the initial solutions based on expert design rules.