Near-Optimal Play in a Social Learning Game