July 1, 2010
A Better Angle on Replay
In the weeks since Jim Joyce’s missed call at first base transformed Armando Galarraga’s rare perfecto into an even more memorable faux-hitter, lots has been said and written about expanding the use of instant replay in baseball. Some have come out against any increased use of technology to correct umpire mistakes, a few with arguments seemingly cribbed from King Ludd, but most with reasonable concerns about game length, undermining authority, and the difficulty of determining where to place runners after an overturned call. Others have supported increased use of replay in various forms, from allowing managers a set number of challenges, to the installation of a replay umpire to intervene when a suspicious call is made. While polls have shown a surprising lack of support among players for replay, a majority of fans seem to like the idea, and Bud Selig has at least tepidly agreed to ask his curiously-constructed “on-field matters” committee to explore the idea.
What I haven’t really seen, however, is much more than a skin-deep look at how replay might actually be expanded, how frequently calls might be reviewed, and what difficulties might ensue from overturning calls. To that end, I decided to park myself in a virtual replay booth for a small set of games to see what might happen if my dreams of expanded replay actually came true, and how well a replay system might stand up to the problems some feel it would introduce.
What follows is a very cursory analysis of one day of games on the MLB schedule (June 22nd), as viewed on MLB.tv’s “condensed game” feature, and full game replays when slo-mo or additional views not shown on the condensed game were needed to rule on a play. While condensed games don’t include every pitch, or even every plate appearance, of a given game, the plays that aren’t included tend to be mundane outs in non-scoring innings. This means the numbers below may possibly undercount the number of plays that need review, but my sense is the difference is small, and this seemed like a reasonably representative starting place for one guy; perhaps we could crowdsource a larger sample in the future.
The Rules: For this experiment, I used the replay proposal that I personally support: an official replay umpire with access to numerous video feeds, with the authority to review any questionable call not related to balls and strikes. If this umpire’s live view of a play shows a reasonable chance that a call was incorrect, he can buzz or call the home plate umpire and request that play be stopped until a quick review can be performed—ideally no more than 30 seconds. To overturn a call, there would need to be what football calls “irrefutable visual evidence” that the call was incorrect. If the play is overturned, the crew chief (with input when necessary from other umps, including the replay ump) decides where to place the batter and/or other runners as necessary, using a few general principles (see Rebuilding The Game State below).
Lacking an actual umpire’s uniform but desiring to look more authoritative, I donned the closest thing I own—a home-made TPB security guard costume—and went to work watching 15 ballgames. I kept track of how many plays looked close enough at live speed to merit a review, and how many I would overturn after reviewing. For my purpose, I applied (and would recommend) a very conservative approach to overturning calls, i.e., only those calls which are quite obviously wrong would be undone.
In the 15 games I watched, there were 42 plays that I felt were close enough to merit a review—thus an average game in this sample sees somewhere in the neighborhood of three plays worthy of review. Of these 42 plays, on 34 occasions I was able to make a decision to affirm or overturn the original call after merely watching the original view at a slower speed. Only eight plays, or about one every two games, required a different view before I could determine whether I should overturn the original call, and only five were actually overturned.
Breaking this down by play type, the largest number of plays I reviewed involved batters beating a throw to first base, with only one that seemed clearly (though barely) wrong after a repeat viewing at a slower speed. Determining when the ball is “squeezed” by the first baseman isn’t easy even at slower speeds, and for these plays I found myself more often than not being unsure exactly when that happened. This is where my “conservative” approach had the most impact—unless I was sure the ball had been completely caught, I couldn’t overturn the umpire on the field, who possibly had the benefit of hearing the ball hit the mitt and the runner’s foot hit the bag in making his call. (Even with a conservative approach I would definitely have overturned Joyce and called Jason Donald out.)
The next most frequently reviewed plays were stolen bases and pickoffs. I’ve been critical in the past of umpires calling players out when a throw beats them to a base, regardless of the tag, and the two plays overturned here were exactly that—high tags on players who were safe but called out. It was more difficult, however, to be sure that any players called safe were actually out—replay could show conclusively that a tag was missed or was late, but it was harder to use a two-dimensional replay to be sure exactly when a tag was applied.
I was surprised to see only one “neighborhood” play (where a middle infielder isn’t in contact with second base when performing a force out) worth a review, and only because it was obvious that the shortstop had stepped off the bag to avoid a sliding runner before the ball got to him. One valid criticism of replay is that neighborhood plays are necessary to avoid injuries and perhaps shouldn’t be overturned, but perhaps they’re not as common as we think. Personally I’d rather see this addressed by a rules change which allows force outs when a fielder is within, say, a foot of the bag.
The only other play I thought would be overturned was a force play at second in which a runner pretty clearly slid in ahead of the throw but was called out anyway at the start of a double play. The interesting thing here is that the runner then slid past the bag, and theoretically might have been tagged out had the umpire called him safe—exactly the sort of dilemma an umpiring crew would face after an overturned call.
The Time Factor: What can we make of all this? First of all, to their credit umpires get most calls right, or at least not obviously wrong, on the vast majority of occasions. Most plays don’t require replay—let’s generously say up to five per game do—and the vast majority of those can be adjudicated after a single replay. If a single viewing can be performed in under 30 seconds, and multiple viewings in less than a minute (both of which seem reasonable to me), the average game might take around three minutes longer under this review system, perhaps a little more if there’s a complicated reversal—in my opinion a small price to pay to avoid the possibility that an umpire mistake affects the outcome of a game.
Booth Reviews vs. Manager Challenges: Some have suggested a “manager challenge” system to ensure only a few “important” calls get reviewed to save time, but if the cost of ensuring all close plays are reviewed is only a few minutes, isn’t that a better solution? The drama surrounding Ozzie Guillen charging onto the field and throwing a challenge flag might very well take more time than a full game’s worth of self-initiated booth reviews. Moreover, any out at any point in the game can make the difference between no runs being scored and a crooked number inning, so who’s to say which plays are “important”? Wouldn’t it be better to make sure all plays are called as correctly as possible? Worst of all, a challenge system would become a managerial tactic, and a “challenge” would be an adversarial moment between managers, players and umpires. Self-initiated booth challenges, however, are akin to umpires huddling to get a call right, and would probably cause less friction—consider the difference in how hockey coaches react to non-reviewable penalty calls and reviewable goal/no-goal calls.
Rebuilding The Game State: Another reasonable argument against instant replay is the difficulty in determining where to place runners after a call is overturned. Sure, for some plays this will be difficult, but it’s not unprecedented. For overturned home runs and fan interference, the crew chief is already directed to place runners where he feels they would have ended up had the interference or incorrect home run call not occurred. Rule 7.05(g) details how to place runners when a ball goes out of play or otherwise becomes dead. No one expects umpires to be recruited from the Precrime program, able to foresee exactly how a play would have progressed had a bad call not intervened, but a mixture of common sense and a few basic guidelines should suffice to fairly rebuild the game state. Here are a few possible guidelines:
This is just food for thought—these suggestions wouldn’t be perfect, but they would be better than doing nothing. The point is these aren’t intractable problems, and since fans are used to seeing ground-rule doubles that keep fast runners on first from scoring, and consider them the same sort of “bad break” as a bad-hop single, rules of this sort shouldn’t be hard to accept.
None of this is meant as a comprehensive replay proposal, nor as a comprehensive study of the impact of replay that MLB itself could and should undertake. However, after spending a single virtual day in the replay booth, I’m more convinced than ever that a replay system could be implemented that judiciously corrects flagrant mistakes, without significantly lengthening games or undermining umpire authority. Umpiring mistakes are unavoidable, understandable, and rare, but anything that can be done to help reduce the chance that a mistake helps decide an important game is worth looking at seriously.