Did the SA author fail to distinguish between one-sided and two-sided testing? My read of it was that he was making a general point - not actually trying to speak statistics. And given his assumptions I think his conclusion is roughly accurate.