These pelicans are clearly indicative of good RL training algorithms.