Have you read about the Fosbury Flop? This is an example of changing the model in order to push the boundaries of what was considered possible. Also, I agree t

Have you read about the Fosbury Flop? This is an example of changing the model in order to push the boundaries of what was considered possible. Also, I agree that there is likely a talent component to establishing the best models for training - but it's also a skill. Imparting these models is what world class coaches do, after all. You are totally right that well-exploited sports have established models that are probably optimal, especially at the elite level. I find these circumstances boring, personally. My interest lies in learning unfamiliar or complex tasks for which no optimal model is known, and also in maximizing the rate of skill acquisition by iterating on the model generation in an abstract way.