In practice, the ideal way to predict what sample a CA will present in, say, a hundred techniques, is simply to run the CA rule alone for just one hundred techniques.