Following Ashlock et al it would be very interesting to see tournament score as a function of number of nodes or parameters.