47 VAR Decisions at the 2026 World Cup: The Intervention Threshold Is Not Inside the Machine

I remember exactly where I was sitting. A small editorial office on the second floor in Manchester, the evening of 16 June 2026. On screen, the Uruguayan referee Andrés Cunha stood motionless inside the Kazan Arena penalty area, one hand pressed to his earpiece, eyes down on the grass. Three minutes earlier, Joshua Risdon had lunged into Antoine Griezmann. Cunha waved play on. Then the earpiece crackled. He ran to the touchline, bent over the monitor, watched the replay twice, turned back, and pointed to the spot. Griezmann scored. France beat Australia 2-1. It was the first time in World Cup history that a penalty was created by a decision overturned through VAR. I recorded the moment the check began, the moment Cunha left the field, how many times he reviewed the incident, and how long the ball took to be back in play. One page, seven lines, nothing but numbers. The colleague next to me asked what I was writing down. I told him: I am recording the things nobody replays. The protocol is not about the ball; it is about authority One thing needs stating plainly, because most viewers do not know it. VAR at the 2026 World Cup was never designed to correct every mistake. The IFAB protocol permits VAR to intervene in exactly four categories: goal or no goal, penalty or no penalty, direct red cards, and mistaken identity. A second yellow card is not on the list. Outside those four categories, no matter how plainly wrong the referee is, VAR has no right to speak. The threshold for intervening is narrower still. The protocol uses two phrases: “clear and obvious error” and “serious missed incident”. Neither phrase has an arithmetic definition. There is no millimetre threshold, no formula, no software that outputs the conclusion “this is a clear error”. It is a human judgement, delivered somewhere between forty seconds and three minutes. Across the tournament I set aside twelve evenings for one task: rewinding every moment a referee left the field for the monitor, and logging the sequence of actions. The final dataset contained 47 decisions, tagged against seven criteria — timestamp, incident type, who called the check, review duration, how many camera angles were used, the final outcome, and how long the ball took to restart. When I sorted the table by the fourth criterion, I found something I had not expected. Forty-seven decisions, seventy-two minutes The total time referees spent standing at the touchline across those 47 incidents was seventy-two minutes. On average, one minute and thirty-one seconds each. But the standard deviation is the interesting part: the shortest was twenty-three seconds — a purely factual offside that needed a single angle — and the longest was four full minutes, in a semi-final. Duration does not correlate with difficulty. It correlates with how many angles the referee asked to see. And the number of angles correlates with something else: how firmly the referee had already made up his mind before he walked to the monitor. This is the point television debate almost always skips. The monitor does not supply the answer. It supplies additional data for a question that was already framed. The law asks about intent. The machine answers about consequence. This is the central conclusion I reached once the dataset was complete, and it accounts for almost every controversy surrounding VAR. Every foul is a question about intent; data only gives us an answer about consequence. When a defender lunges at a forward inside the penalty area, the law does not ask “was there contact”. The law asks three other questions: was the challenge careless, reckless, or made with excessive force? Was the player genuinely attempting to play the ball? Who initiated the contact? Not one of those three questions can be measured by a camera. A camera records foot position, speed of contact, angle of impact — all of it consequence. Intent sits on the far side of a line the lens cannot reach. And this is where the experimental evidence becomes uncomfortable. In 2026, Eugene Caruso, Zachary Burns and Nicholas Converse published a study in the journal PNAS showing that when an identical action is watched in slow motion, viewers rate it as significantly more intentional than when they watch it at real speed. The effect does not disappear when viewers are told they are watching slow motion. In other words, the tool VAR uses to find the truth is itself the tool that distorts the very element the law needs to assess. A referee reviews in slow motion, sees the challenge as more deliberate than it was, and arrives at a heavier decision. At real speed, the same incident looks like an accident. That is why two near-identical challenges can produce two different outcomes, and both can be “correct” by the reading of the person holding the whistle. The inconsistency viewers see on television does not live inside VAR. It lives in the fact that football law is written in the language of intent, while technology is written in the language of images. The referee's angle and the viewer's angle are two different objects I once spent a long evening with a first-tier referee in England, rewatching a match he had officiated. I asked him a simple question: when you are out there, what do you actually see? He answered with something I had not anticipated. He said he does not see the incident. He sees a moment — a player's torso shifting axis abruptly, studs braking on grass, the direction of the other man's shoulder. He makes the decision on that chain of signals, not on the image of the collision. The viewer sees the situation, the referee sees the moment, and I see the whole process. Three objects, three readings, and only one of them gets replayed on television. That gap explains most decisions branded “incomprehensible”. The viewer at home has six angles, slow motion, and thirty seconds to form an opinion. The referee on the field has one angle, real speed, and roughly two hundred milliseconds to make the initial call. And that initial call is precisely what the protocol hinges on: VAR may only intervene when that call constitutes a “clear and obvious error”. The structure contains a paradox. The more technology is introduced, the more important the referee's initial decision becomes — because it is the reference point. Technology does not replace human judgement; it turns human judgement into an anchor. The 2026 World Cup had the highest rate of referees leaving the field for the monitor of any major tournament in history. By the 2026 World Cup the trend had reversed sharply: referees were encouraged to stand by their original decisions, and most checks were handled remotely, without the referee leaving the pitch. Technically this is an improvement — the game is interrupted less. Procedurally it creates a new problem: when the referee does not review himself, the person making the final call is someone who has already heard VAR's conclusion before seeing the images. Tactical fouls: when a referee has three seconds to read intent In the summer of 2026, the Premier League was suspended by the pandemic. With no matches to report on, I sat through all 380 fixtures of the 2026-20 season, focusing on Liverpool's tactical fouls under Jürgen Klopp. The number I recorded: an average of 10.2 fouls per match, mostly in midfield, and mostly committed within five seconds of losing possession. The number is not the interesting part. The handling is. An identical act — a shirt pull, a body block, a clip of the heel — draws two different treatments depending on where it happens on the pitch and what the tempo of the match is. In the tenth minute it is an ordinary foul. In the eightieth minute, with the score level, the same act is whistled instantly, sometimes with a card. Referees do not measure behaviour. They read intent, and intent shifts with context. A foul in midfield in the first half is tactical calculation; a similar foul outside the box in the final minutes is a latent goalscoring threat. One movement, two meanings. VAR cannot read that second layer of meaning. It only sees contact. Penalties: the one place where everything is measurable The Euro 2026 final between Italy and England at Wembley is a case study running in the opposite direction. I was assigned to analyse the shootout, and I spent most of that time on something nobody discusses: which elements of a penalty can actually be verified with data. The penalty spot is 11 metres from the goal line. At an average strike, the ball takes roughly three-tenths of a second to travel from the spot to the line. Gianluigi Donnarumma chose the correct direction on most of the shootout attempts in the knockout rounds, and that was not luck — it was the product of studying each taker's standing-foot habit and hip-opening angle. Marcus Rashford and Jadon Sancho were sent on late, primarily to take penalties, and neither scored. Bukayo Saka, who had started the match, also missed. Those three misses fell inside a window no referee could enter, because shootouts are effectively exempt from VAR. Yet what VAR actually checks on a penalty is a very short list, and almost all of it is factual: did an attacker enter the area before the ball was struck, did the goalkeeper keep at least part of one foot on the line until the moment of the kick, did any player encroach before the ball left the spot. That is worth pausing on. Of all the incidents VAR handles, the penalty is the least disputed once the outcome is known — because most of the check is measurement, not judgement. Nobody argues about whether the goalkeeper's foot was on the line. They only argue when the question turns to intent. The laws never stand outside the match; they are the second match played in parallel. The counter-intuitive point: VAR does not fix errors If you read the entire dataset I collected, the most honest conclusion is this: VAR does not fix errors, it only changes who carries the responsibility. Before 2026, when a wrong decision was made, the responsibility belonged to the referee on the field, and it ended there. After 2026, responsibility is distributed across a chain: the on-field referee, the assistant VAR, the video referee, the replay technician, the person selecting the angle. Nobody in that chain carries full responsibility, and that is precisely why the controversies have not diminished — they have only changed shape. The most common demand I hear on discussion programmes is: “make VAR consistent”. That demand cannot be met within the current laws, and not because referees are poor. It cannot be met because the object under review is intent, and intent has no unit of measurement. One piece of evidence sits in the history of the laws themselves. The handball law was amended repeatedly between 2026 and 2026 — each tightening, each loosening producing a fresh wave of controversy. Those amendments were not responses to refereeing errors. They were symptoms of a legislative crisis: the people writing the laws had never agreed among themselves on what the law was meant to protect. A referee is the only person on the pitch not permitted to be guided by emotion. But a referee is also the only person who must answer to a document whose own authors have not agreed on its purpose. Which is why, when a crowd is furious at a referee, most of the time it is furious at a text. Football has no VAR, only hidden angles waiting to be exposed. Looking ahead, I do not think the right reform is more cameras. The 2026 World Cup showed that semi-automated offside technology can convert a judgement question into a measurement: twelve cameras under the roof, twenty-nine data points per player, sampled fifty times per second, plus an inertial sensor inside the ball recording the moment of contact. But even once the measurement layer is fully automated, what remains is human: whether a player is “involved in active play” is still a judgement call, and always will be. What should be published is not the footage but the reasoning. If every VAR decision came with a short explanation — what the referee saw, which threshold was applied, why this incident crossed it and that one did not — the public argument would shift from “right or wrong” to “consistent or not”. That is an argument capable of resolution. And the question I leave behind: if the lawmakers themselves cannot define “clear and obvious error”, what exactly are we demanding of a person holding a whistle, in three seconds, inside the noise of seventy thousand people?