They did hand-wave a little when they got to the part about how MWU gets in there as an update function. It's not clear whether it's "the solution" or "a solution." Whether the strategy is guided by MWU or some equivalent method, though, I think their insights about optimizing for diversity are interesting.

Thanks for the pointer to the No free lunch theorem. That is interesting and I wasn't aware of it. Although, I wonder if it applies here, because it says:

Wolpert and Macready have proved that there are free lunches in coevolutionary optimization.

Thank you for reading and offering feedback!