Conclusion

There is apparently something in my cellular automata that encourages rule arrays to efficiently turn the test arrays towards the same final state every time. I do not know why this happens, but I have a few guesses. Perhaps my cut off for the cellular automata development (1000 steps) is not long enough to allow most rule arrays to finish developing test arrays towards a proper majority. Therefore, the algorithm will favor any rule that achieves a quick final state. Even if that final state is always the same, the rule array should still have an approximate success rate of 50than that of an array that never manages to finish its development within the development limit.