The Eye and the Data: Who Really Makes the Call on a Tennis Court?

Third set, score 5-4, receiving game. A serve is called out. The player slams the racquet down, points at the line, and asks for a review. On the big screen, the ball path flashes up, touching the line by perhaps two millimeters. The crowd holds its breath. The chair umpire looks at the screen, nods, overturns the call. The match turns on a dime in twelve seconds.

I have sat through those moments many times. And each time I ask myself the same question: what just happened — a correction, or an entirely new decision being created? In tennis, the line between those two things is thinner than people assume.

More than a decade ago, the authority to judge on court belonged to the human eye. Line judges stood along the lines, deciding in a split second, and their calls were almost impossible to reverse. The challenge system arrived as a pressure valve, letting players contest a ball — but only a limited number of times, and only when they believed they were right. From then on, Hawk-Eye stepped in like a higher court.

Then the game changed again. The majors gradually removed line judges entirely, moving to live Electronic Line Calling. The US Open led, the Australian Open followed, and by 2026 even Wimbledon — the most tradition-bound of all — announced it would play without line judges for the first time. A long chapter of tennis closed. But removing line judges does not mean the end of disputes. It only moves the dispute from the court line into the technical room.

Three layers must be separated inside an electronic decision. The first layer is the sensor: how the camera system is calibrated, how it is positioned, and what the manufacturer's stated "average error" really means. Hawk-Eye has been recorded with an average error of roughly a few millimeters — a number that sounds small, until you remember that many decisive balls sit just millimeters from the line. When the margin of a decision sits against the margin of error of the tool, "certainty" becomes far too strong a word.

The second layer is processing. From discrete frames, an algorithm reconstructs the ball trajectory and extrapolates the point of impact. It is a model, not a photograph. It is right almost always, but "almost" does not mean "absolutely."

The third layer — and the one where I spend most of my time — is the human operator. Who calibrates the system? Who decides when to display the image, how many frames to show, and when the image is so blurry that the chair umpire must be called in? The tool is neutral. The operator is not. When data conflicts with the eye, trust the data — but never forget to check its origin. That is the sentence I repeat to myself every time I sit in front of a screen.

I once spent three days re-watching an entire amateur match, counting every collision and cross-checking against the official report, all because I doubted a single number. The discrepancy I found taught me something: official reports are not wrong out of malice, but because they inherit the error of whoever entered the data before them. Numbers do not generate themselves. They are written by someone, at some time, with some degree of care. The same holds in electronic tennis. Every "in" or "out" on screen is the final result of a chain of technical choices — choices the public never sees. And that dark space is exactly where every real dispute lives.

There is another angle rarely discussed: the psychology of the challenge system. When a player has only one review left, they are no longer reacting to the ball — they are reacting to risk. They save the review for a more important point, sometimes letting a clear error in a mid-game go, in order to hold the right for set point. This means technology does not merely correct errors; it creates a new strategic game, in which the right decision is postponed out of calculation rather than belief.

Traditionalists say: machines ruin the soul of the match, erasing the moment when human disputes human. Modernists say: humans judge wrongly far too often, machines are fairer. Both are half right, and both miss the same point.

The missed point is this: the problem was never the tool, but that we hand the tool a power the tool does not have. Hawk-Eye is not an all-knowing judge. It is an estimate with an error bar, placed inside a ritual of display that makes it look like truth. When the crowd believes the projection is absolute truth, then any system error becomes a scandal — and any decision contradicting the projection is treated as stubbornness.

The paradox sits right there: the more we trust technology, the less we forgive the humans who operate it. In reality, if we want more fairness, we must publish the error margins, publish the calibration process, and admit that some balls fall into a "grey zone" that no tool resolves completely.

A tournament is a system. Every referee decision is a variable. And the analyst's job is not to cheer for a variable, but to check whether the system operates consistently. I log every warning, every serve-clock overrun, every point penalty for conduct. Because a wrong number repeated three times becomes a fact in the end-of-season report.

In tennis, penalties rarely carry the weight of a football red card, but warnings, point penalties, and game penalties for time or conduct violations all leave a mark on the match's flow. A warning issued at the wrong moment can turn the psychology of an entire set, and a trailing player can break stride simply because a decision was made off-beat. My first mistake was not the card I wrongly reported in an old article. It was believing I would never report one wrongly.

So where does the future lie? Perhaps not in returning to line judges, nor in canonizing machines. The future lies in transparency: showing error margins, recording the process, and keeping one human ultimately accountable — but a human who clearly understands the limits of the tool being used.

Because in the end, fans do not need a machine that never errs. They need a match in which every decision can be questioned, and every question can be verified.