On July 15, 2026, thirty-seven minutes before kick-off of the World Cup final at Luzhniki Stadium, I sat in the control room for the goal-line and VAR systems, re-checking the calibration layer of the virtual offside line on the monitor. The average deviation between the camera signal and the actual pitch: 0.43 metres. I filed a correction report. The organisers had to re-verify the entire system before France and Croatia walked out.

In the stands, nobody knew. To more than 78,000 spectators, the offside line is an absolute line. To the person sitting in the control room, it is a band of error that has to be compensated for — measured by a human, entered by a human, owned by a human.

Eight years later, the 2026 World Cup expands to 48 teams, 104 matches, spread across 16 cities in three countries. Different cameras, different software, and a match load nearly half again as large. The question remains intact: when the decision is handed to the machine, who ultimately pays the price?

From Russia 2026 to North America 2026: the same problem, a different shell

VAR made its full debut at the 2026 World Cup in Russia. The first penalty review in tournament history took place on June 16, 2026, in France versus Australia: referee Andrés Cunha initially awarded France a direct free-kick after a challenge on Antoine Griezmann, then changed it to a penalty after reviewing the footage at the pitch-side monitor. Around the same period, I took part in reviewing the own goal by Aziz Bouhaddouz for Morocco.

At the 2026 World Cup in Qatar, semi-automated offside technology arrived. The system uses a cluster of cameras under the stadium roof to track 29 points on each player's body, combined with a sensor inside the ball that records the moment of contact at high frequency. In principle, this was a major leap: humans no longer had to manually select the frame in which the player touched the ball.

The 2026 World Cup places that system in a far harsher environment. 104 matches instead of 64. Stadiums at different altitudes, climates ranging from Houston's humid heat to Toronto's cold. Each venue is its own calibration problem, and every move between host cities means the system has to be set up again from scratch.

That is the context the audience never sees. All they see is a line appearing on the broadcast, accompanied by a number measured in centimetres.

Dissecting the decision chain: four points where it can go wrong

The line never lies, but the person who draws it can. In more than a decade of working with refereeing data, I have always broken an offside incident into four points where error can enter — and only when all four are clean does the conclusion hold.

First, camera geometry. In the same three-dimensional space, fewer cameras mean larger occluded zones. With semi-automated offside, if a single player blocks the sightline of two or more cameras at the exact moment of contact, the system has to interpolate — and interpolation is controlled guesswork, not truth.

Second, the moment of contact. This is the point I once argued about most fiercely with colleagues. Before the era of in-ball sensors, the operator had to manually pick the frame in which the player touched the ball. At 50 frames per second, one frame of error equals 20 milliseconds. A player sprinting at 8 metres per second covers 16 centimetres in that window. Sixteen centimetres is the distance between a goal and an offside offence, between the next round and a ticket home.

Third, calibration drift. Cameras mounted on the stadium roof are subject to vibration, wind and thermal expansion. A line established at 10 a.m. is no longer correct at 9 p.m. unless there is a re-verification procedure. The 0.43-metre figure at Luzhniki was the result of exactly this kind of drift. It was not a software failure. It was an operational failure.

Fourth, the people in the room. This is the point major tournaments discuss least. The VAR team is not just one referee. There is an assistant VAR, a replay operator, a calibration technician. The main referee only issues the final ruling; the person who selects the frame, the person who selects which camera angle appears on the monitor, the person who decides whether to freeze the frame — these are all decisions that bear directly on the conclusion.

From the 2026 dataset: when crowd noise disappears

In 2026, when the Chinese top flight restarted with empty stadiums, I analysed 212 matches before and after the outbreak of the pandemic. The home win rate fell from 41.3 per cent to 35.2 per cent. Average yellow cards per match dropped 17 per cent, from 3.8 to 3.15.

The press at the time wrote about the collapse of home advantage. I disagreed with that interpretation. An empty stadium does not create ghost football — it creates storytellers: people who need a compelling narrative more than they need a verified variable.

The real cause lay elsewhere: referees lost a reference signal. Crowd noise — the swell of reaction that rises and fades after every challenge — is a silent yardstick that helps referees recalibrate their foul threshold over the course of a match. Without that signal, the threshold shifts, and it shifts in different directions for different referees.

My white paper was later used as referee training material. But the larger lesson was methodological: to draw a conclusion about a change, you must separate the environmental variable, the human variable and the tooling variable. Blending those three together is the fastest way to produce a wrong analysis.

Storytelling experience, data signs the record

In 2026, I was a mid-level VAR analyst at the data centre of the Chinese football federation. I spent six weeks compiling 47 penalties across 15 matchdays and found that one referee leaned toward the home team in 68 per cent of 50/50 situations. The report was rejected on the grounds that refereeing intuition mattered more than statistics.

In August 2026, the federation changed how the handball law was applied, based on similar data. My report was reinstated and became an internal document.

I retell this not to claim I was right. I retell it to point at a mechanism: when a correct conclusion is buried, it is buried by a process, not by a malicious individual. And when it is restored, it is restored by a process too. The human is the ultimate unit of analysis, but humans act inside the framework the system builds for them.

That is why I do not watch matches; I read the rhythm of a match through individual frames. A goal is merely the final output of a chain of decisions, and the chain is what deserves to be dissected.

The contrarian angle: VAR is not wrong; expectations of VAR are

What most viewers fail to realise is that VAR was never designed to make every decision correct. It was designed to correct clear and obvious errors, or to catch serious incidents the on-field officials missed. The high intervention threshold is deliberate. Lower it, and football becomes a game in which every challenge is reviewed, with the on-field referee reduced to a presenter rather than the person running the match.

But there is a more serious problem, and it is rarely discussed: review time.

A review lasting three minutes does not merely interrupt the match. It alters the psychological state of 22 players, alters the tempo, and alters the referee himself — a man left standing and waiting, knowing that every spectator in the stadium is watching him on the big screen. Making a decision after three minutes of waiting is not the same as making one in two seconds.

And here an asymmetry of accountability appears. The on-field referee has a name, a face, and takes the criticism. The replay operator inside the VAR room does not. Nobody knows who selected which frame, who selected which camera angle to show the main referee. That anonymity protects people from media pressure — but it also protects mistakes from being corrected.

One further point needs to be stated plainly: most viewers judge a VAR decision by their emotional attachment to a team, while the system judges a VAR decision by the consistency of a process. These two yardsticks almost never coincide. The gap between them is the source of most of the controversies we see after every major matchday.

Conditional conclusion

If major tournaments do not publish the intervention logs of the VAR team — the timestamp, the camera angle selected, the review duration, the pre-match calibration report — then every controversy will keep circling without accumulating anything. Data that is not published cannot be verified, and what cannot be verified cannot be improved.

A refereeing system only gets better when someone checks the checkers.

Data collection method: the 2026 figures come from internal analysis logs across 15 matchdays; the 2026 figures come from 212 matches before and after the pandemic, cross-referenced between the organiser's aggregate sheets and personal logs; the 0.43-metre calibration detail comes from an internal technical report dated July 15, 2026, at Luzhniki.