I saw David Beck (the pitcher known as “The Creature” in Moneyball whom Oakland signed sight unseen) pitch for Cumberland, and would never have drafted him, his fastball was well below MLB average and he didn’t have the arm speed needed to generate sharp breaking stuff. He was released the next year after getting knocked around, as he should have been based on how poor his stuff was. There is a lot of good work being done using sabermetrics, but I think it should be used to enhance the job of the scout…not replace him.

I completely agree with the above. A good sabermetrician would incorporate scouting data (like arm speed, movement, and delivery) into his forecasting model. To completely ignore scouting data is to ignore half the model.