May 10, 2013

Overthinking It

Where the Value of Robot Umpires Ends

by Ben Lindbergh

Printer-friendly

Contact Author

The building I grew up in had manually operated elevators. They were quaint prewar contraptions that required an attendant to slide a metal screen across the entrance and a pull a hand crank to start the ascent and stop at the desired destination. (They looked a little like this.) When you got to your floor, you felt like you’d earned it. Or you would have, if not for the person paid to take you up and down.

Those elevators had been there as long as the building, so they had tradition and inertia on their side. And most of the time, they did the job as well as a more modern elevator. But they had a tendency to get stuck between floors, they broke down fairly frequently, and they were expensive to service. Eventually, it became clear that to complete another repair would only postpone the inevitable at additional cost, and the manually operated elevators were replaced by the boring kind with buttons. I don’t remember any outcry about preserving the historic human element of the elevators, probably because by that point the would-be preservationists were sick of climbing stairs.

The first time I entered one of the newly installed automatic models, I expected to find it empty. But there was another body between me and the buttons. The elevator operators were still there, now with even less to do. Some of them would press the button for your floor and stare straight ahead at the wall inches in front of their face. Others would turn around and study you silently, as if you were disturbing their solitude or disrupting whatever they did when the elevator was empty. It was as uncomfortable and guilt-inducing as it sounds, and if I’d lived a floor or two higher I probably would have just resigned myself to life as a shut-in in order to avoid the awkward rides.

It seems inconceivable that a building employed (and still employs!) people purely to press buttons that could have been pressed with equal ease by anyone else. But the elevator operators were union workers, and the union was too powerful for the building to let them go over a little thing like their jobs becoming completely redundant. So they stayed, and despite the absurdity of the situation, there were benefits to having a human around, even in a mostly extraneous role. They could do things that the simple algorithm otherwise operating the elevator couldn’t, like tell if someone wasn’t supposed to be in the building or hold the door open while you ran back in to retrieve your umbrella. Of course, now that I live in a different building where the elevators have always been automatic and I have to rough it and press buttons like everyone else, I don’t miss the manual way. The manual way was much less efficient.

Those vestigial elevator operators, still pressing buttons after most of their responsibility was removed, might be where home plate umpires are headed. The current system of calling balls and strikes has been with baseball since the beginning, and it works well most of the time. It’s easier to accept the occasional blown call than it is to overhaul the system, so the system stays in place. But eventually, the technology will become so advanced (if it hasn’t already), and teams and fans alike will grow so dissatisfied with mistakes, that the manual way won’t fly. The umpires could still crouch behind the catcher, partially because umpires also have a powerful union that would prevent them from being fired and partially because they’d still serve a purpose—relaying the count to the pitcher and the players at the plate, helping to confirm that the cameras are correctly calibrated, even taking over in the event of a technological failure. (Ideally, whatever gear they'd wear would make them look like Lobot.) But however it works, one day—if the evolution of elevators is any indication—the burden of calling pitches will pass to a system that’s better equipped to handle it than the human eye.

It won’t happen overnight, nor should it—you can’t screw with the strike zone lightly, and suddenly switching to a rulebook zone would have serious implications for the pitcher-catcher-batter balance, even if the technology is up to the task. But I’m sympathetic to the “robot umpire” movement, at least when it’s less about calling umpires incompetent than it is about freeing them from a task that’s impossible to perform perfectly. However, when it goes beyond balls and strikes (and, to be fair, avoidable incidents like Tom Hallion’s disputed exchange with David Price and John Hirschbeck’s ejection of Bryce Harper), it can be taken too far.

This has been a bad week for umpires. Joe Maddon was ejected on Tuesday after arguing an incorrect out call by Marty Foster on a slide at home plate, then ejected again on Wednesday when Tim Welke said that what would have been a groundout had actually been a foul ball off a batter’s foot. Also on Wednesday, an incorrect home run call was allowed to stand by Angel Hernandez even after replay review, despite video evidence showing that it had cleared the yellow line that marks home run height at Progressive Field and bounced off a railing and back onto the field. And on Thursday, maybe most egregious of all—pending any further revelations that might clarify the ruling—Fieldin Culbreth allowed Astros manager Bo Porter to bring in lefty reliever Wesley Wright to face left-handed hitter J.B. Shuck, then remove Wright in favor of righty Hector Ambriz when Angels skipper Mike Scioscia sent up right-handed pinch hitter Luis Jimenez. That decision appeared to violate Rule 3.05(b), which states:

If the pitcher is replaced, the substitute pitcher shall pitch to the batter then at bat, or any substitute batter, until such batter is put out or reaches first base, or until the offensive team is put out, unless the substitute pitcher sustains injury or illness which, in the umpire-in-chief’s judgment, incapacitates him for further play as a pitcher.

The Angels played the rest of the game under protest, though—surely to Major League Baseball’s relief—they won anyway, making the protest moot. But the Angels’ victory didn’t sooth savage Twitter, which was pushed too far by the quick succession of obviously wrong rulings.

I’m singling out Joe only because I happened to see his tweet in my timeline; plenty of other people expressed similar sentiments. I don’t disagree that we deserve better; what I take exception to is the idea that any of those incidents could have been avoided by adopting robot umpires. I think we’re overestimating the abilities of these theoretical robots.

Computers are better at some tasks than others. They’re awesome at, oh, adding, for instance. But if you’ve ever had to conduct a lengthy interview and look up excerpts later, you know that computers aren’t much help when it comes to transcription, at least not now—they still need humans’ help with that. Similarly, computers would be of much more help in getting certain kinds of calls correct than others. They’re very good, though far from perfect, at distinguishing between ball and strikes. But interpreting rules, determining whether they’re being adhered to, and enforcing them is an even more difficult matter

A robot ump could be programmed to sound the alarm when a pitcher is substituted, then removed from the game before throwing a pitch. But first, someone would have to tell it that the pitching change had occurred, and that the second substitution wasn’t the result of an injury. And that person would have to enter that information well before the second substitute pitcher threw his first pitch in order for the robot ump to A) recognize the incorrect ruling and B) dispatch someone else to the field to notify the meatbag umps of their mistake in time to prevent it. It’s also hard to see how a robot ump would be able to review a replay or tell whether a runner got his hand on home plate before being tagged. To keep track of the game, a robot ump would be reliant on human scorers, and human scorers make mistakes. Exchanging #robotumps for #robotscorers would only move the weak link to a longer chain.

Doing away with human umpires isn’t a viable option today, given the demands of the job and the available alternatives. As Colin Wyers put it, "Until we replace the PLAYERS with robots, some human needs to be involved here." What makes the last few umpire flubs more frustrating than a forgivable blown call behind the plate is that they're mistakes even imperfect umpires shouldn't make. So the more helpful response to anything other than a blown ball or strike call isn’t a reflexive “#robotumps,” but “How can we make human umps better?” That’s an area where technology can and has helped, and could help more. MLB should use technology to support the umpires, and persuade or coerce the umpires into accepting that support.

Umps are already graded on their strike zone accuracy, and the good ones get rewarded with more important assignments. Replay, Angel Hernandez aside, has made a positive impact, and it could be made better by updating the existing replay review system (which could have contributed to Hernandez’s mistake) and making replays admissible for more types of calls. It’s hard to recommend specific training measures, because we don’t know for sure which training measures are already in place. (More transparency might improve the process, too.) But umpires obviously need to know the rules, and they need to be trained not to provoke arguments with players. And if they can't comply with those reasonable requirements, they need to pay a price. After all, they’re only human. And they'll have to be for the foreseeable future.

Ben Lindbergh is an author of Baseball Prospectus. Follow @benlindbergh
Click here to see Ben's other articles. You can contact Ben by clicking here