Researchers offer a model for more effectively targeting wrongdoers
Learning gradually versus all at once and how we estimate value