At dinner this weekend (at a hip salad joint), my friend (a machine-learning expert) talked about a conceptual framework for thinking about decisions: exploration versus exploitation. Just add the term "synergy," and that would be the most California sentence ever written. But the principle is important to think about, because it could help you shift your training paradigm to disrupt the running space. "Machine learning" is how computers can understand tasks without being explicitly programmed for them (think recommendation algorithms on Netflix that seem to know you a bit too well). "Reinforcement learning" is a subset of machine learning inspired by behavioral psychology that addresses how actions maximize cumulative rewards. "Exploration" is trying new things, branching out into uncharted territory.