The traditional method of learning is to try to learn the correct "textbook" way of performing a skill, and to do lots and lots of repetitions of that exact movement. But there's some increasing research in the last 20 years which suggests that this may not always lead to the best learning or performance. And that doing a bunch of goofy "wrong" repetitions that one would never do in performance could actually lead to more consistently and high-level execution of that skill. What?! Evidence That Too Much Consistency in Practice Could Potentially Hinder the Learning Process (What?!)