if all((state in state_samples) for _, state in move_states):
                log_total_samples = math.log(sum(state_samples[s] for s in move_states.values()))
                move, state = max(move_states,
                            key=lambda _, s:upper_confidence_bounds(state_results[s],state_samples[s], log_total_samples))
            else:
                move = random.choice(list(move_states.keys()))