Love this article.
Its very interesting that, in pitcher counts, speed plays a small role yet in hitter counts, guessing the correct speed makes a huge difference. This makes me wonder, if fast pitches are truly associated with hitter counts, and it looks like hitters guess them pretty well anyways, would it be more beneficial to throw slower in hitter counts?
All in all, this experiment uses a very primitive model for a pitcher vs. batter scenario. Simply randomizing pitch location takes away a huge chunk of strategy that goes into the cat and mouse game. In fact, I think pitch location, more so than speed, is what a hitter would try to predict.