Formalize how we interact with a problem. Use two examples as case studies: 1. Stability of linearized game dynamics and its relationship to Nash optimality, 2. Conjecture learning for feedback-control systems.