Pureness also makes it easier to remove calls to a function. If f is a pure function, then I know that if x, y or z doesn't change, then I never need to call f(x, y, z) again.