Hacking probability theory