(Credit: © Paul Yates - stock.adobe.com)
In a Nutshell
- In 2024, South Korea’s KBO became the first pro baseball league to fully automate ball-and-strike calls, with a computer deciding each pitch and relaying it to the umpire through an earpiece.
- Star batters who had won the league’s Golden Glove award saw their plate-discipline stats slip against lesser-known hitters once the system took over, consistent with losing favorable calls from human umpires.
- The drop showed up only in stats tied to ball-and-strike decisions, not raw hitting power, and a placebo test ruled out a natural slump as the cause; high-status pitchers showed no such decline.
When a legendary player steps to the plate, something subtle happens that most fans never notice. An umpire, consciously or not, hands that star a little extra grace on close pitches. A ball that should be a strike gets called a ball. A borderline pitch goes the batter’s way. Over the course of a full season, those tiny gifts pile up. For decades, nobody could prove it, let alone stop it.
Now they can. In 2024, South Korea’s professional baseball league became the first to fully hand ball-and-strike calls to an automated system, a setup that decides each pitch and pipes the verdict to the home plate umpire through an earpiece. Researchers at the University of Michigan studied what happened next, and star batters who had long enjoyed favorable treatment from human umpires saw their numbers slip against lower-profile players. The playing field, it turns out, had never been quite level.
Published in the journal European Sport Management Quarterly, the study drew on data from South Korea’s Korea Baseball Organization (KBO) for the 2023 and 2024 seasons, one year before the robot umpire arrived and one year after. High-profile batters, those who had previously won the league’s most prestigious honor, the Golden Glove, posted measurable drops in plate-discipline stats such as walk rate, strikeout rate, and on-base percentage compared with lesser-known players. That gap between famous and ordinary hitters shrank once the machine took over, consistent with a hidden advantage quietly disappearing.
How Korea’s Robot Umpire Works
KBO’s system uses three cameras, positioned at first base, third base, and the mid-outfield, to track each pitch and map an individualized strike zone for every batter. Software calculates whether the ball passed through that zone, then relays the ruling to the home plate umpire through an earpiece, and the umpire announces the call. No second-guessing, no reputation to weigh, no crowd noise, just geometry.
That marks a sharp break from more than a century of tradition. Human umpires rank among the best in the world at a punishing job, yet they remain human. In Major League Baseball, roughly 34,000 incorrect calls were made during the 2018 season alone, about 14 per game. Some of those misses come with the difficulty of the work. Others, this research indicates, run deeper.
What the Robot Umpire Exposed About Star Batters
To gauge whether fame was swaying umpires, the researchers used Golden Glove history as a status marker. In Korean baseball, the Golden Glove is a celebrated end-of-year award chosen by votes from KBO-related media officials, and winning one gives a player a reputation that follows him everywhere, including into the batter’s box.
Song and colleagues drew on 148 batter records and 112 pitcher records from players active in both 2023 and 2024. Batters who had won a Golden Glove before 2023 counted as high-status. For pitchers, the team relied on award nominees, since only three Golden Glove-winning pitchers turned up in the sample.
Batting results were clear. After the automated system arrived, high-status hitters saw their walk rates fall, their strikeout rates climb, and their command of the strike zone erode, all measured against lower-profile players. Their on-base percentage dropped by about 0.025 relative to lesser-known hitters, and their walk-to-strikeout ratio fell by more than a third. These were steady, meaningful shifts across several stats tied directly to ball-and-strike calls.
The decline showed up only in the stats tied to ball-and-strike calls, like walk rate and strikeout rate, while raw hitting power stayed the same. Stars had not forgotten how to hit. They had simply stopped getting the benefit of the doubt.
To rule out other explanations, researchers checked whether the stars were simply due for a natural slump that happened to coincide with the new system. Rerunning the analysis as if the machine had arrived a year early, they found the opposite pattern: before the real rollout, high-status batters were actually trending upward against their peers. Only once the automated calls began did the decline appear, which points to a loss of favorability rather than ordinary regression. The results also held under different sample sizes and status definitions.
Why Star Pitchers Weren’t Affected
High-status pitchers showed no such drop. Celebrated and lesser-known arms fared about the same after the switch, and the authors offer a few reasons. Pitchers meet the automated system on every pitch they throw, and that steady stream of feedback may help them adjust faster. Struggling pitchers also tend to get pulled from a game, which caps how far a bad night can drag their stats, while batters stay in the lineup and absorb the full effect. Pitcher numbers also bounce around more from game to game, making small shifts harder to spot.
What the Findings Mean Beyond Baseball
This research travels well beyond a ballpark in Seoul. The authors point to other arenas where subjective human judgment decides who gets ahead. Blind auditions in symphony orchestras, where judges cannot see the musician, raised the odds of women being hired. Similar effects turn up in job screening, where an applicant’s name, race, or gender can tilt decisions even when evaluators believe they are being fair.
Their broader argument is that automated tools, whether robot umpires, blind résumé screens, or computer-assisted reviews, can flatten playing fields that people, for all their good intentions, tend to tip. Korea’s system did more than make the strike zone consistent. It quietly stripped away an edge that reputation had been handing the game’s biggest names every time they came to bat.
For Major League Baseball, which folds an automated ball-strike system into its challenge format for the 2026 season, Korea’s experience offers a preview of what America’s stars may soon face. The technology will change how pitches are called. It may also change who benefits, and who no longer does.
Baseball has always sold itself as a game of clean numbers, a tidy ledger of balls, strikes, and outs. That ledger was never as honest as advertised, and a bank of cameras in South Korea just started keeping a more accurate count.
Paper Notes
Limitations
The authors flag several limits. Their status measure, Golden Glove history, captures only one slice of how reputation might sway umpires, and it ignores the status or strategy of the opposing player in any given at-bat. The sample comes entirely from the KBO, so the results may not carry over to leagues with different player pools, cultures, or officiating setups. Sample sizes were fairly small, especially for pitchers, which can affect the reliability of some estimates. The team also could not directly test which psychological mechanism drives the bias implied by its numbers. They call for future work that repeats the analysis across more leagues and seasons and looks at which players are most affected.
Funding and Disclosures
The authors reported no potential conflict of interest.
Publication Details
Authors: Jimin Song, Ji Hyuk Kang, and Richard J. Paulsen, all with the Department of Sport Management, School of Kinesiology, University of Michigan, Ann Arbor.
Paper Title: “Technology adoption and bias in officiating: automated Ball-Strike System implementation in Korean Baseball”
Journal: European Sport Management Quarterly
Published Online: July 9, 2026.
DOI: 10.1080/16184742.2026.2681111. It is an open-access article published under a Creative Commons Attribution-NonCommercial-NoDerivatives (CC BY-NC-ND 4.0) license.







