Thursday, June 3, 2010

Against replay and commissioner revision

OK, I guess I have to weigh-in in part on this. Mike captures some of the arguments in favor of replay and I have made my views known in the past. The WSJ Law Blog picked up our dispute. And here is a nice summary (with a bent towards opposing replay, I sense) from Ted Frank at Point of Law. Josh Levin goes farther to suggest ways that umpires can be eliminated in substantial part through some technology, including a system of making video responsible for the initial call rather than for review.


This game does not change my mind about replay generally and specifically in baseball. I still am not convinced that replay will across-the-board increase accuracy. And, in any event, I continue to believe that the efficiency and workability concerns outweigh any increased accuracy. And I just am not willing to give up on the human side of it. If that makes me old-fashioned or tradition-bound, so be it. In any event, I will hide behind the argument from Fred Schauer and Richard Zeckhauser about the danger to making rules (or, in this case, changing my mind about policy) based on a specific, unique case, which is necessarily skewing as to the larger problem. This is a classic example of that--one vivid story likely will drive major changes to replay.

Instead, let me talk about the question (mentioned in the WSJ post) of whether Bud Selig should step in to undo the call and award Galarraga a perfect game, to which I say no. The commissioner does not (and I believe should not) exercise power to overturn a particular call in a particular game and there is no instance of the commissioner (or a league president) ever overturning a particular call.  The commissioner/league president review power always has been limited to matters of interpretation of the rule (such as the Pine Tar Game in 1983), not its application. And, as Ted Frank notes, MLB will not overturn an erroneous interpretation if it would not affect the outcome. [Update: Selig announced he will not reverse the call, but will look into the umpiring system and, oh no, expanded use of replay.]

There also is a danger to allowing this one unique case to over-determine the question of post-hoc revision of calls. Is it so different that the blown call occurred on the 27th out rather than the 26th? How about on the 1st out--Suppose (as I propose in a comment on Mike's post) the blown call had occurred with the lead-off hitter, Galarraga then picked him off and proceeded to retire the next 26 batters. Or suppose that Batter # 27 took what replays all showed unquestionably show (and the plate umpire later admits) was strike 3 but was called a ball, then got a base hit on the very next pitch--should the commissioner be able to go back and say the batter actually struck out? None of these situations are, it seems to me, different than what we actually have--there would have been a perfect game but for the blown call.

Finally, it is interesting that this has become the lightning rod for replay, because unlike other historically "wrong" calls (see Denkinger, Don) this one did not affect the outcome of the game, but only a historical footnote.

By the way, I would have been in favor of the umpires huddling on the play and overruling the call at the time. I am not sure if the rules allow it in that situation, but it seems appropriate there.

Further Update: Mitch Berman of Texas (a co-panelist on my "Judges as Umpires panel who is working on a book about sports/law links) makes the following interesting point:
[I]t’s rare that a call can be corrected without having either to make contestable counterfactual judgments or to replay the game forward from the point of correction.  The latter, of course, is what had to happen in the pine tar game.  But in this most recent fiasco, the miscall can be corrected and everything all wrapped up without further ado. 

Indeed, I’m tempted by the following proposal: allow the C’mish to reverse mistaken factual determinations when (1) the call was clearly erroneous; and (2) correction would require neither counterfactual judgment nor re-play.  Those two conditions are likely to obtain very rarely.  But when they do, why not correct what’s correctable?