The Crucial Role of Problem Formulation in Real-World Reinforcement Learning