Αποστολή με SMS: Design of Experiments for Reinforcement Learning