June 21, 2012
In A Pickle
The War For the Ballfields
I asked an old-timer, someone who was involved in baseball decision-making 50 years ago, about Aug. 29, 2067, the day that came to be known as Red Wednesday. "There were people who tried to warn us. We didn't listen. But it's not like it came out of the blue. We brought that day on ourselves."
A confluence of technological and cultural factors early in the 21st century created an ever-rising wave of frustration, frequently boiling over to anger, at Major League Baseball over its refusal to improve umpiring. With the public availability of the PITCHf/x pitch-tracking system, fans could see their team being screwed by human fallibility in nearly real time. Large high-definition televisions and advanced camera technology allowed video replays of fair/foul and safe/out calls for the home viewer, but those same replays were not available to the umpires who could put them to the best use.
It would have been one thing for fans to sit home alone, throwing things at their televisions in frustration as the missed calls piled up, occasionally conducting extended bitch sessions at the local watering hole. The spread of technology that allowed fans to employ actual facts in their age-old complaints against officials, however, coincided with the rise of social media. Fans could put a baseball game on television (either via cable or MLB.tv through a set-top device) and have ESPN's Gameday application open on a laptop (the better to track PITCHf/x) next to a Twitter window (the better to complain to thousands of friends about the game) or a team blog's "game thread."
Screencaps and gifs of bad calls proliferated. Armando Galarraga was robbed of a perfect game by a later-contrite Jim Joyce:
A base-runner was called out when Todd Helton, then Colorado's first baseman and now one of the Hall of Fame's more controversial inductees, caught a throw literally feet away from the bag.
The cry of "robot umps" became a nightly occurrence on Twitter, often in hashtag form for easy searching.
Certain elements of the baseballerati made arguments against getting humans out of the business of judging whether a baseball crossed the plate or landed foul. Some contended that a fully equal strike zone took away from the skill displayed by some pitchers and catchers to get more strike calls from human umpires. Others pointed out that the technology was not yet ready for prime-time, given the known margins of error and necessity of calibration.
Still others, mostly people without subscriptions to Baseball Prospectus, argued that removing humans from the umpiring crew would remove the, well, human element. Joe Torre, for instance, in his role as executive vice president for baseball operations, argued, "Most in the game recognize that the human element always will be part of baseball and instant replay can never replace all judgment calls by umpires."
As it turns out, the problem with stating that "most in the game" believe something is that the population in the game changes. As the old guard faded away, so did the "human element" argument—the new generation recognized the "human element" that fans cared about was that of the players. Baseball games existed to pit Stephen Strasburg against Mike Trout, not to see whether a 53-year-old umpire could make a snap judgment on the location, to the inch, of a 96–mph fastball.
As the cultural shift happened over the early and middle decades of the 2000s, technological changes made implementation feasible. Increased processing speed allowed PITCHf/x to give strike and ball calls nearly as fast as human umpires did, but without the fuzzy edges and uncertainty that typified the human strike zone. (As it turned out, pitchers and catchers were forced to innovate on sequence, location, and hitter-deception, leaving some players clearly better than others at this aspect of run-prevention despite very similar "stuff." Pitcher and catcher skill, in other words, did not become a purely physical matter of velocity and movement.) The necessary hardware was developed to determine whether force plays were made (when did the ball hit the mitt with the fielder's foot on the bag, compared to when the runner's foot or hand touched the base?), whether a tag was applied in time, whether a ball was trapped or caught, and so forth.
There were glitches at first, as there always are. Situations unforeseen by Major League Baseball's programmers would arise from time to time and require human intervention. Fortunately, humans were not removed entirely from the equation. Games were monitored by trained umpires sitting in a central video room in New York. They could intervene if the automated system on the field failed.
MLB, incidentally, drew significant praise from the software industry regarding its programmers' speed in fixing bugs brought to light by corner-case game situations. (Some in the industry were heard to grumble that MLB was setting expectations too high. Users wanted their word processing software fixed as fast as baseball corrected its mistakes. Also, many baseball commentators noted approvingly how far baseball had come culturally given its lengthy history of refusing to acknowledge or fix mistakes.)
Fans and players needed an adjustment period, of course, not only in the strike zone, but on the bases as well. "The ball beat the runner" had been an announcer refrain for decades to wave away concerns that the umpire had missed a sneaky slide or a blown tag. Both players and fans had, over years of watching baseball, created a mental picture of what it meant for a runner to be out. They developed heuristics, in other words, and the truth, which did not always fit those heuristics precisely, took some getting used to. Eventually, however, sliding skill became a legitimate area of study, as some players showed a notable ability to reach base safely despite the throw being at the bag in plenty of time to record an out. (Major League Baseball's release of the data it used to determine safe and out calls aided this research greatly. It was later regarded as a pleasant side-effect of MLB's decision to respond to calls for an audit of its data by simply releasing that data into the wild. The main goal of said release, to bolster fan and player confidence in the accuracy of its "umpires," was also a rousing success, and was recognized as a major factor in baseball's shift back to the top of the American sporting pyramid. The contrast between MLB's proven ability to make the correct calls and the NBA's continued difficulties with conspiracy theories rooted in the referees' nearly impossible task of judging fouls could not have been more stark.)
The biggest adjustment, though, was to the lack of an authority figure on the field. Ultimate frisbee players had few difficulties—they'd been playing their sport at a high level without a referee for years—but everyone else in sports was used to a highly visible person signaling his judgments. It turned out that something was lost in removing the umpires from the field. The crowd rising in anticipation of a play at the plate as a runner attempted to score from second on a single felt let down when the result of the play was flashed on the scoreboard. A giant "SAFE" sign in center field didn't compare to the umpire determining that the catcher had applied the tag just a millisecond late, straightening from his crouch, and throwing his arms wide in a gesture that, when the home team was the one doing the scoring, matched the exultant mood in the stands.
At first, MLB thought that it was just a matter of time. Fans grew up watching umpires, so they couldn't be expected to immediately feel an emotional connection with calls flashed on the massive high-definition screens throughout the park. Perhaps this belief would have been borne out given enough time, but even after fans and players adjusted to the new strike zone and correct calls on the bases, the feeling of something missing persisted. While television viewing stayed strong, ballpark attendance began showing a decline. (TV has never relied directly on the umpire's calls the way fans viewing the game in person have.)
Up to this point, the idea of "robot umps" had been a metaphor. Nobody seriously thought that C-3PO would stand on the field and make calls. Robots were too immobile to get out of the way of balls or players, and human-replicas fell into the uncanny valley. Eventually, though, humanoid robot technology and society's adjustment to said robots' growing prevalence were sufficiently advanced to provide solutions: distinctly non-human robots became fast and agile enough to avoid objects (and people) hurtling toward them; and their artificial intelligence could anticipate the movement of players given the game situation and the path of the ball, letting them position themselves to avoid events like Carlos Lee tackling Hunter Wendelstedt:
In retrospect, it's remarkable that the players' reactions to the on-field robot umpires weren't foreseen. Automatic calling of balls and strikes was quickly put in place not just in professional baseball, but in most of NCAA as well. A few rich high schools even installed the system. (The death of football left high-school boosters needing somewhere to put their money, and the calls for help for classroom teachers went predictably unheeded.) Outside of organized baseball, however, safe/out and fair/foul calls continued to be made by humans. The costs simply were not worth the gains for schools, travel teams, independent leagues, and so forth.
What this meant was that players, even decades into The Robot Era, still grew up as ballplayers with human umpires on the field. Humans who could make mistakes. Humans with whom managers and players would sometimes have heated arguments.
Now, maybe that's just an excuse for how some players began treating the robot umpires. Maybe the players were simply expressing their frustration at being out, not actually disagreeing with an objectively correct decision made by a computer deep in the bowels of the park and only communicated by the robot umpire on the field. Whatever the reason, baseball soon had a violence problem on its hands. Or at least a property-damage problem. (The terminology and the proper use of the word "violence" when the "victim" was not a human was hotly debated.)
Either way, players began throwing helmets at the robots, shoving them to the ground, and occasionally engaging in even more creative action. As with most "epidemics" of this type, the actual number of incidents was always less than the perceived problem. Remember, after all, the entire reason for the existence of the on-field robots was because fans missed the humans. For these human-replacements to be subject to the kind of attacks not seen in the major leagues since the early days of the game upset people irrespective of the relatively low real cost of the "problem."
The union fought bitterly over the lengthy suspensions that MLB first attempted to hand down and eventually won a landmark arbitration case on the issue. The near-invulnerability of the umpires figured significantly in the decision. Who exactly was being harmed by this behavior to the point where suspensions should equal those of human vs. human fights?
Nobody is quite sure who proposed electrification. All we're sure of is that someone in the league office devised a system whereby robot umpires would become literally untouchable any time the ball was dead. (To avoid accidental shocks, the Umpire Defense System was turned off during the course of play—the robots were fast, agile, and smart, but occasional incidental brushes with players still happened.) The implementation of the shock-bots (upheld by another landmark arbitration decision after the union filed a grievance) eliminated the problem of players damaging umpires almost overnight. The resulting disappearance of on-field arguments, an organic shift away from "defending teammates" via beanball wars and takeout slides, and a general cultural trend discouraging agitation and the public display of anger combined to create the most peaceful era baseball had ever seen. It lasted almost three decades.
Ground zero of Red Wednesday was Cincinnati. The Cardinals were in town for a late-season game with no playoff implications. The Umpire Defense System had gone unused for so long that many fans had forgotten it existed. Players adapted new, more suitable, less visible outlets for their frustration. What nobody considered was robot frustration.
The theory now is that the problem was the central umpiring computer's learning code, which had been gathering data on weather, ball movement, player behavior, and more, all with the goal of improving its ability to satisfy its prime directive: to eliminate human error. This had been the rallying cry of the movement for robot umps from its early days. Unfortunately, in the quest for communicative simplicity, specificity was lost.
The inciting incident was an error on a routine ground ball. Nobody knows why this error on this day in a game between these teams did it, but the result, ironically, was that the central computer arrived at the same erroneous conclusion that Joe Torre had all those years ago: it was unable to separate umpire human error from player human error. The computer, embodied in the on-field robot umpires, having come to understand that the only way to eliminate human error was to eliminate the humans, proceeded to do what was necessary: it ejected every single human player from every single game. The umpires, first in Cincinnati but quickly spreading to all other parks in which a contest was being played (as well as a few west coast stadiums where teams were just taking batting practice), swiftly moved from player to player, delivering a robotic version of the ancient heave-ho signal.
In a few cities, the teams attempted to play on, pulling Little League umpires and the like out of the stands to judge balls and strikes. They quickly grew frustrated with the inconsistent strike zones and gave up the attempt. In any case, the groups of electrified robots (to them, the ball was dead and any baseball action was completely unlicensed) criss-crossing the field ejecting players were a serious distraction.
The story doesn't end here, of course, but the War For The Ballfields played out in predictable fashion: the central umpiring computer resisted efforts to shut it down, robot umpires occupied America's professional ballparks, and Major League Baseball quickly came to regret not implementing Isaac Asimov's Three Laws of Robotics. It's been told before, and there's little to learn from telling it again. From the tale of how we got to the war, though, there is still much to be absorbed, from arrogance about technology to the dangers of overreaction to the dark side of a highly connected world.
"You know what they say about unintended consequences," the old-timer tells me. "Baseball is complicated. Or was complicated. Technology is complicated. Life is complicated. I've studied the history of this thing a little bit. Things really weren't that bad. There were some notable missed calls, sure, and the strike zones were occasionally infuriating, but honestly, a little human-driven instant replay would have made everyone happy. This didn't have to happen."