Well, teaching's different from doctoring/lawyering in a few ways. If you judge, say, a couple defense lawyers strictly on their success rate, put one on OJ Simpson's team and give another a long backlog of public defender cases, then Johnny Cochran's going to be the best layer ever. Does that mean he would've done better with the cases where you're set up to fail? Not to say that metrics are useless, but they can definitely be flawed.