The Margin of Error of the Whistle: When VAR Lacks Enough Evidence to Conclude

In the 34th minute at the Tottenham Hotspur Stadium, Luis Díaz broke through, rounded the goalkeeper, and the net rippled. The assistant referee had already raised his flag. The on-field signal was "no goal". The VAR team stepped in, reviewed, and broadcast two words: "Check complete." Play restarted with a free kick. Nobody in the stadium understood what had just happened. Only late on the night of 30 September 2026 did the English refereeing body admit it was a significant human error and release the audio so the public could judge for itself.

What stands out: the cameras worked correctly. The data was not lacking. What was lacking was a question asked in the right place.

In ten years of officiating, I learned that the hardest decision is not the wrong one. The hardest decision is the one you must make when you know you do not yet have enough evidence. The mistake in Russia that year did not teach me how to referee correctly — it taught me how to live with the sound of my own whistle. And I came to realise that most VAR controversies around the world ultimately sit inside that same gap.

Context: a system designed to remain uncertain

The VAR protocol rests on four reviewable categories: goal or no goal, penalty or no penalty, direct red card, and mistaken identity. Outside those four, the referee may not refer to the VAR team. The founding principle is "clear and obvious error" or "serious missed incident". By the very way it is written, the law admits it only intervenes when the error is large enough for the human eye to catch.

Which means: in every grey-zone situation, the on-field decision stands. Not because it is right. Because there is not enough evidence to call it wrong.

Alongside that protocol runs a wave of technology. At the 2026 World Cup in Qatar, FIFA deployed semi-automated offside technology. The system uses 12 dedicated cameras mounted under the stadium roof, tracking 29 data points on each player's body, sampling 50 times per second. The official ball, Al Rihla, carried an inertial measurement unit operating at 500 Hz, allowing the exact moment the ball left the foot to be determined. UEFA adopted the system in the Champions League from the 2026-23 season. Serie A followed from the 2026-24 season. The English Premier League approved it and brought it into operation in a later phase.

Technically, this was a leap forward. Psychologically, it created a dangerous promise: that a single correct answer exists, and that enough cameras will find it.

Analysis: five gaps inside the machine

Gap one — the kick point

A 500 Hz sensor tells you which millisecond the ball left the foot. But player positions are sampled only 50 times per second, meaning each frame is 20 milliseconds apart. A player sprinting at 35 km/h covers about 9.7 metres per second. Multiplied by 0.02 seconds, he moves roughly 19 centimetres between two frames. In other words, positional error at the decisive instant can reach nearly two hand-spans.

That is why, before the 2026-22 season, the Premier League moved to drawing thicker offside lines. Thicker lines create an implicit tolerance, and that tolerance admits something rarely spoken aloud: technology does not measure the truth, it measures the closest approximation of the truth.

The core of the problem sits here: football built a judging system in the language of absolute certainty, yet operates it with tools that only reach relative accuracy.

Gap two — the paradox of the "clear" threshold

The protocol says intervention only occurs when the error is clear and obvious. Sounds sensible. But it creates a paradox I encountered many times on the pitch.

The more ambiguous the situation, the higher the intervention threshold, so the on-field decision is more likely to stand. The clearer the situation, the lower the threshold, so the on-field decision is more likely to be overturned. The final value of a phase of play therefore depends not on whether the player was offside, but on whether the assistant referee had raised the flag.

Same phase of play. Same frame. Two opposite outcomes, because one person raised an arm half a second sooner.

In ten years of officiating, this troubles me most. Fans believe VAR is searching for the truth. In reality, VAR is checking whether the original decision was wrong to an indefensible degree. Those are two very different questions.

Gap three — communication is part of the law

Back to the phase at Tottenham. The cameras were sharp enough. The offside line was clear enough. The ball sensor worked. And yet the outcome was wrong. The cause lay in a misread assumption: the VAR team believed the referee had awarded the goal, so they concluded "check complete" and did not intervene. The referee believed he had awarded a free kick, so he could not understand why the VAR team stayed silent.

Two people spoke two different languages about the same event, and neither checked their own assumption.

The technical lesson here is concrete. A judging process is only as strong as its weakest link. When the weakest link is an unconfirmed assumption, upgrading cameras from 50 to 500 frames per second solves nothing.

Gap four — when process data and results diverge

Based on my experience of watching matches, I always separate two layers of analysis: the process layer and the result layer. In the process layer, I record chances created, chance quality, and touches in the final third. In the result layer, I record goals, cards, and boundary decisions.

There are matches where these two layers tell entirely different stories. A team generates a superior volume of chances but loses to a controversial penalty. A team plays second-best but wins through a goal suspected of offside.

What is frightening is not the divergence itself. What is frightening is how we react to it. Instead of saying "the result here does not reflect the process", we go looking for a culprit. And a culprit is always available: the referee, the VAR team, or the technology.

Gap five — fans have learned to read frames

A phenomenon I have observed in recent years: audiences are getting better at reading frames. They know which frame favours their argument, where to pause, how to draw a straight line by eye. That skill once belonged only to the VAR room. Now it sits in anyone's pocket.

The consequence is that every controversial goal spawns at least two opposing versions of "evidence", both looking persuasive, both cut from the same footage. The argument is no longer about the phase of play. It becomes an argument about which frame is the honest one. And that is an argument with no referee.

Transmission across the industry

A controversial decision does not stop at the 90th minute. It spreads along a fairly clear path.

Upstream, academies and referee training centres must rebuild their curricula. Young referees learn positioning and reading play, but spend less and less time learning how to decide when evidence is thin — because they believe the headset will cover that part.

Midstream, clubs invest in data analysis to prepare for grey zones. Some hire specialists to study the flagging tendencies of individual assistant referees and the intervention appetite of individual VAR teams. Others train offside-adjacent phases at the edge of the error margin, treating them as potential scoring points. When the margin of error becomes a tactical variable, tactics have entered the territory of probability rather than pure technique.

Downstream, the market feeds on events. Every three-minute review creates an information vacuum, and that vacuum is filled with speculation. Public pressure on managers and on referees themselves rises, even when the final result may have nothing to do with their competence. A manager loses his job after a run in which two matches were decided by phases sitting inside the tolerance zone.

And at the deepest layer, a generation of fans is being taught that every dispute has a technical solution. When the solution does not appear, they lose faith in the whole system, not just in one individual.

The counter-intuitive angle

VAR does not fix mistakes for the match — it exposes how we define a mistake.

Before VAR, a referee's error was an accident. It happened, caused a few days of argument, then drifted away. After VAR, an error becomes evidence. It persists in the footage, in the frame, in the data. We can replay it a thousand times.

But replaying it a thousand times produces no additional information. It only produces more belief that the information should have been there.

There is one thing the offside trap can never catch: the player's intention. A striker running through is not always doing so because he believes he is onside. A centre-back stepping up is not always doing so because he believes he is springing the trap. Both are reacting to a situation they only partly perceive. The offside law was written to measure position, but the argument revolves around motive, and no camera records motive.

I used to think the fans' frustration was aimed at technology. Now I think differently: it is aimed at the promise. Technology advertises itself as something that can answer every question. When the answer does not arrive, people do not blame the question — they blame the answer.

Here I must say something the analysis industry rarely admits. "Not enough information to conclude" is a valid conclusion. In research, it is an honest answer. In football, we have no place for it. The scoreboard has no box marked "undetermined". The table deducts no points for ambiguity. And because there is no place for it, we are forced to invent an answer, then argue with the answer we just invented.

That is why I no longer believe more cameras will calm the controversy. More data means more ways for two people to watch the same phase and reach opposite conclusions. The problem is not the volume of information. The problem is that we have never agreed on how much uncertainty we are willing to accept.

What I consider a sensible direction

In another corner of sport, cricket solved this problem long ago. It recognises the concept of "umpire's call": a publicly declared tolerance zone within which the on-field decision stands regardless of what the prediction model says. That zone exists not because umpires are always right, but because the model also has error.

Football could do the same with offside. Publish a fixed margin of error — 10 centimetres, for example — and declare that inside that margin, the on-field decision stands. The outcome of a phase of play would then no longer depend on whether the assistant referee raised the flag quickly or slowly, but on a pre-defined threshold, identical for every team, every match, every stadium.

This will mean some goals are wrongly disallowed. True. But it will also end something more corrosive: a few more unresolvable arguments every week.

For penalty situations, the direction could differ. Instead of trying to establish absolute truth, the VAR team could be asked a narrower question: was the original decision outside the reasonable zone. Framing it that way does not produce perfect justice, but it produces consistency.

And operationally, one lesson is already obvious after the Tottenham case: every protocol needs a step that rechecks assumptions. Before declaring "check complete", the VAR team should read aloud the on-field decision they are assuming. Three seconds of reading aloud could save three weeks of argument.

When the stands are empty, I hear the ball striking the boot clearly — something ten years of officiating never let me hear. The sound of ball on boot does not lie. It just does not say enough. Perhaps that is what modern football needs to relearn: the difference between what does not lie and what says enough. Technology is good at the first part. The second part still needs a human standing in the right place, accepting that he may have to conclude while not knowing everything — and owning that conclusion.