There still seems to be the problem where two genes are
referencing each other. Could they possibly be doing it
in a roundabout way? like g1->g2->g3->g1-> ? I know the
probability is very low, but if the program is run with
a high enough mutation rate it comes up quite a bit.