It’s Not Time For Robot Umpires

We all know the feeling. It comes along with the proverbial baseball situation: your team is up to bat, 2 outs, 3-2 count, runners in scoring position. Your heart is pounding, eyes wide as you sit so far on the edge of your seat that you swear the chair isn’t even there. Your hands instinctively clasped together to begin your natural prayer to the baseball gods.

And then, this happens…


“Come on! That was ball 4!” you scream, hoping somehow your anger gets transmitted through the TV straight to the umpire.

As far as missed calls go, the one above is pretty tame. It was close, probably too close for the batter to take. Some calls, however, are slightly more egregious:


The umpire clearly forgot his glasses. How else could he make a call so terribly wrong? In fact, why is he back there in the first place? I mean, the TV strike zone appears to have gotten the call right in both cases. Shouldn’t we just give in to our robot overlords and replace the umpires with computers? Surely this would make the game better.

This sentiment is growing amongst MLB fans. However, this idea is, at best, uninformed, and in my opinion, wrong. Let me explain.

The Tech

The first side of this argument comes from the technological aspect of the robo-ump. Major League Baseball already has pitch tracking cameras and radar installed in every ballpark. This tech was first used around 2006 under the name PITCHf/x. In 2015, the MLB introduced Statcast, which offered much more robust movement tracking and analysis capabilities. Between these two systems, almost every pitch of every game for the last decade has been tracked and recorded.

So the problem seems easy, right? We’re already tracking every pitch, and recording where they start and where they cross the plate. So can’t we just replace the umpires with this system? Unfortunately, it’s not that easy. It has been documented time and time again that the Statcast system isn’t as accurate as you might think. According to a 2017 article by Rob Arthur, Statcast is actually less accurate than its predecessor PITCHf/x. In particular, the system struggles with reporting the vertical component of the ball’s location, and the pitch’s break.

The inaccuracies of Statcast are also analyzed in detail by David Kagan in his piece The Physics of RoboUmpThere are a plethora of other articles that report similar troubles for Statcast (some of which I have linked to in the “References” section below) but to make a long story short, there are essentially four issues:

  • The system has significantly more trouble accurately tracking the vertical part of the zone than the horizontal part.
  • According to the rules, each batter has a unique strike zone height, making it difficult for a computer to accurately determine balls and strikes.
  • Sometimes the tracking system simply misses a pitch completely. What would happen if the robot ump didn’t recognize that a pitch was thrown?
  • The way that the TV strike zones get broadcast is via a time delay. The TV station gives themselves time to process the pitch data an overlay it onto a strike zone, it’s not instantaneous. Could this result in more delays in baseball?

The key thing to take away here is that every measurement, and thus all tracking equipment, has some sort of error. In the case of Statcast, that error might be bigger than one would think. According to Kagan, it is reasonable to believe that a machine could call inside/outside pitches as well as, if not better than, a human, but it seems as though human umpires still have the edge when it comes to north/south calls.

In fact, the way that PITCHf/x determined the top and bottom of the zone was by use of human stringers manually turning a dial for each hitter until the strike zone looked correct. Statcast, on the other hand, uses a history of past umpire calls to reconstruct what the strike zone is. In both cases the MLB is actually using human input to make a computerized system, which in turn will contain plenty of human error. Another interesting note is that the strike zones reported on TV (K-zone, FoxTrax, etc.) are simply overlaying Statcast data onto a strike zone graphic, so they are subject to the exact same errors stated above.

What About the Future?

A common counter to this technological argument is that, surely as time goes on, the technology will get better. We will build better machines, faster computers, and solve more pitch tracking problems, given enough time. Eventually, we will have programs that can use computer vision and machine learning to find the top and bottom of the strike zone on every batter. Or maybe MLB will simply change the strike zone rule so that it’s a universal size, whether you’re Jose Altuve or Aaron Judge.

New York Yankees v Houston Astros
Aaron Judge, at 6′ 7″, towers over Jose Altuve, 5′ 6″. It looks like they probably have different strike zones. Via

All of that may be true. If we don’t have capable tech now, we’ll probably have it in the not-so-distant future. But the question remains: should we use it?

Now I am no Luddite, nor do I play one on TV. If it weren’t for Statcast and PITCHf/x data, and the incredible job the MLB has done with their public database Baseball Savant, this blog would be astonishingly boring. I study, analyze, and play around with Statcast data all the time, and the technical advances that organizations like MLB Advanced Media have made are unparalleled in other sports. No matter how much resistance new technology meets in its beginnings, it generally seems to stick around, often for the betterment of everyone’s lives. Technological advancement is a force that is hard to stop, which is why once the MLB makes the switch to robot umpires, it would be near impossible to turn back, even if it proves to be a mistake initially.

For that reason, I ask you to pause and think about where the game of baseball is, how human umpires fit into that situation, and what you want the game to become. Go ahead, take a second to think, or a minute. Heck, come back next week if you want. I’ll be here.

Culture Shock

Several internet articles over the past few years have claimed that human attention spans are getting shorter. There is much debate surrounding that notion, and if it were true, it could have a lasting impact on sports watching. However, what appears to be less controversial is that when we get a notification on our phones or tablets, our brains receive a small hit of dopamine. Something about the lights flashing and the idea that you may have gotten a like on your photo is very pleasing to our brains, and addicting.

When the TV broadcast has a little white box projected in front of the catcher, and the pitcher throws a ball in the zone, a filled-in white circle flashes onto the screen. A strike! Our brains get the same instant gratification as when we receive a notification on our phones, often before the umpire can even make a gesture. These microsecond interactions may seem insignificant, but I worry that they feed into our possibly shortening attention spans.

Chris Guccione is seen wearing a white wrist band last season to protest ‘abusive’ behavior from players. Via Chris O’Meara / AP

You see, baseball is not a game for short attention spans. The nature of the game, whether it’s being played or watched, simply doesn’t play well with a need for instant gratification. It’s a game without a clock after all. It’s because of this that it’s rather difficult to compare the “pace of play” or amount of “action” in baseball to other sports. Baseball is just, for lack of a better word, different. My fear is that computerized strike zones similar to the ones broadcast on TV will change our psychosis around the game. We will be so focused on the small flash of light of the strike being called, that our sense of the game and our willingness to pay attention to it could drift away, no matter how accurate the call.

The other fact of the matter is that human umpires, whether we like it or not, are ingrained within the culture of the game. Ever since the game’s creation there has seemingly been a constant battle between the players and the umpires. The pitchers throw more strikes and try to “get away with” borderline calls. The catcher tries to befriend the umpire to get him on “his side.” They learn the fine art of framing, trying to ever so slightly fool the umpire’s eyes. Batters stare disapprovingly into the black, wire mask behind home plate. They ask questions. They “learn the zone” of that particular umpire. And every once in a while, if the fans are lucky, they will see a manager storm out of dugout, either out of frustration with the umpire’s lack of vision, or simply to light a fire under his players’ butts.

That’s not to say that all parts of this culture are perfect, or even good. After all it sounds a bit inherently unfair doesn’t it? Umpires shouldn’t be on a team’s “side” and players shouldn’t have to “learn a zone,” they should know their established zone, right? Clearly, there are a lot of judgements that could be made about the current culture surrounding human umpires. What’s indisputable, however, is that replacing human umpires with robots would instantly and drastically uproot this century-old culture. Could the hole that’s left in the ground be filled in easily?

Given the current state of pitch tracking technology, we would get marginal improvement in calls behind the plate at best. Is this small improvement worth changing the game in such a significant way? Some of you may say yes. Right now, I would disagree. But the important part is that we think deeply about the repercussions surrounding these big changes to the game and its culture.

The truth is, baseball is an incredible game. We’re living in an era where it’s changing, regardless of who is calling balls and strikes. The so called “launch angle revolution” has seen soaring home run numbers, and also soaring strike out rates. There seem to be some fundamental shifts in hitter’s mindsets, and a lot more literal shifts on the infield. To me, the possibility of maybe someday getting more accurate strike calls seems like small potatoes compared to these more foundation-shaking recent trends.

We all want to find ways to make sure this game not only survives, but thrives. I love baseball, and all of its quirky culture. It’s imperfect, yet impossible to master. I also love technology, and all of the fascinating opportunity it brings. I don’t believe that these two sides need to be mutually exclusive. There feels like there should be a way to reconcile them, as long as we approach it carefully. So if we change the game, let’s change it for the better. But first, we need to figure out what “better” actually means.



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s