>
>
> The map is a hexagonal grid of "prisoners", each one playing
> prisoner's dilemma with it's neighbours. Red stands for Defect,
> Green for Cooperate
>
>
The payoff matrix is:
* D C - neighbour
* D 3 13
* C -1 7
The map starts at maximum zoom out with the entire map displayed. The
selection button doesn't do anything apart from select at the moment.
The prisoners use two primitive neural networks, both take the last
four interactions as input and each turn the neural network with the
highest output wins, this decides whether to cooperate or defect.
Reproduction is into neighbouring cells, both asexual and sexual with
neighbours. Evolution is by crossover and mutation. Cells of the same
colour are of the same species.
So far, the program seems to show that evolution of cooperators is
achieved, in stable groups of the same species or friendly species.
However there are what I call crime waves which spread until evolution
wins over them. I have seen a crime waves destroy all the prisoners,
(non-cooperators cannot survive on their own with this payoff matrix)
