From Kazan to Wembley: A 1,247-Decision Referee Ledger and the Compensation Effect

Wembley, 7 July 2026, the 104th minute. Raheem Sterling goes down inside the box after contact with Joakim Maehle. Referee Danny Makkelie points to the spot within four seconds. He does not call for the monitor, does not move from his position, does not consult the assistant. Harry Kane shoots, Kasper Schmeichel saves, Kane scores the rebound. England reach the Euro final, Denmark go home at 1-2.

In Seoul, I sit in front of the screen with my notebook already open before kick-off. I do not write down the goal. I write the interval between Sterling receiving the ball and the whistle. I write the distance between the referee and the point of contact. I write that the far-side assistant was the only person who never changed position during the entire sequence. Those three lines sit beside each other in the same column, and they do not agree with one another.

I do not watch the goal; I watch the camera that watches the goal. By the 104th minute of a semi-final, the camera has stopped being a tool and become the only witness both sides can accept or reject depending on what they need.

A night in Kazan and a 47-page notebook

To understand why a collision in the 104th minute caused such fierce argument, you have to go back to another night, three years earlier.

Kazan, 27 June 2026. South Korea meet Germany in the final round of World Cup group matches. In the second minute of stoppage time, Kim Young-gwon puts the ball in the net. The assistant raises his flag for offside. The referee lets VAR intervene. The goal stands. Four minutes later, Son Heung-min scores into an empty net. South Korea win 2-0 and Germany are eliminated in the group stage — the first time in that nation's history they have left a World Cup at that point.

I was seventeen that year, a high-school student in Seoul. My friends celebrated; I downloaded all 64 matches of the tournament and started logging every incident where VAR changed a decision. Two weeks later I had a 47-page journal devoted entirely to video interventions. Kazan erased a goal, but it opened an eye.

That notebook shaped how I write to this day: note the minute, note the score at the moment of the decision, note the referee's position, note the number of replays. Every conclusion has to stand behind the raw record.

In 2026, when the pandemic froze almost every pitch on earth, I took the Kazan notebook back down and turned it into something larger. I coded 1,247 refereeing decisions from World Cup 2026 and three K League 1 seasons. The 2026 season had no spectators, but it had one enormous ear: every whistle in an empty stadium was captured intact, unburied by crowd noise. Listening back through the footage, I realised referees had lost an important signal — the signal telling them how serious their own decision had just been.

Four categories and a name: the compensation effect

The 1,247 decisions were split into four groups by reversibility. Group A held 183 incidents where a goal was overturned for offside or handball. Group B held 402 in-box contact incidents leading to penalties. Group C held 268 second-yellow-card incidents. Group D held 394 fouls outside the box with a direct effect on a goal.

The coding rules were strict. An incident counted as "wrong" only when replay showed a decisive element the referee could not have seen from his position, or when the referee reversed his own call after the monitor. Any incident where two camera angles produced two different answers was filed as "inconclusive" and dropped. I accepted losing data in order to keep the rest standing.

The result made me stop for a few days. After a team suffered a wrong decision, the probability that the same team would be awarded a low-controversy penalty in their next two matches was 89%. I called it the compensation effect.

The mechanism behind the compensation effect

The compensation effect is not cheating. It is the sum of three things.

The first is cognitive load. A referee processes ball position, the position of 22 players, the clock, cards already issued and pressure from both benches simultaneously. When a major decision has just caused controversy, that load is not released — it carries into the next match as an invisible debt.

The second is game management. Referees are trained to keep a match under control, and the cheapest way to cool down an aggrieved team is to hand them a favourable situation. A penalty in the 60th minute carries fewer consequences than one in the 90th, so the drop point usually falls early in the second half.

The third is moral arithmetic. Humans flatten things out, and referees are human. They do not count errors in favour of one side; they count them by a felt sense that the balance has tipped.

The striking part lies elsewhere. The compensation effect appears more strongly in matches with large crowds and weakens markedly during the empty-stadium period. Without a crowd, referees lose their external ruler for how serious a decision is, and the psychological debt stops being recorded the old way.

Of the 402 Group B incidents, 356 saw referee and VAR agree within ten seconds. Such fast agreement on incidents that depend on the feel of a challenge is itself a signal about how VAR operates: it does not re-examine the decision, it confirms it.

The compensation effect and the lesson from Qatar

World Cup 2026 in Qatar was the first time I took the model out of the office for testing. I predicted referees would restrict cards in the group stage to preserve match flow, on the assumption that a November tournament would generate a denser playing rhythm than usual.

The model was right in 26 of 36 group matches. Then it collapsed in the Netherlands versus Ecuador match on 29 November 2026 at Khalifa International Stadium. The referee issued 8 yellow cards and awarded 2 penalties — rates my spreadsheet could not produce. After the tournament I spent three weeks interviewing two former FIFA referees. They showed me what I had missed: when a host nation exits early, pressure from the stands and from the organisers becomes a different kind of tension, and referees respond by tightening their grip on the match.

That is a social variable, not a technical one. It does not sit in a spreadsheet, and it cannot be measured by the number of monitor checks. Data never makes a mistake; the writer is the one who gets the card.

The blind spot is not the referee

Most of the argument after England versus Denmark landed on Makkelie. I think that is where the eye is pointed wrong.

The first blind spot is the broadcast director. Throughout the semi-final, the person deciding what viewers saw was not the referee and not VAR — it was the director. A camera angle skipped, a slow-motion frame cut early: those choices shaped the verdict of millions, and no mechanism audits them. We audit referees with footage, but nobody audits the person choosing the footage.

The second blind spot is the law. The laws say contact in the box may be a foul, but they do not quantify how much force is enough. In that gap, every decision is simultaneously right and wrong. That is why no data can refute the 104th-minute penalty: refuting it would require a threshold, and the threshold has never existed.

The third blind spot is crowd emotion. Viewers react to images, not to laws. A fall that looks heavy reads heavier than a fall that looks light, and the eye grades before the law speaks. The law is the only thing that never enters stoppage time, while emotion enters from the first minute.

There is a paradox few accept. VAR was introduced to reduce controversy, yet controversy grew. The reason is that VAR did not create the grey zone — it simply turned a light on a grey zone that already existed. Technology does not make the law clearer; it makes the law look clearer, and the distance between those two things is what produces the sense of being cheated.

Sterling fell in the box; I stood up in the lecture hall. The 2,100-word piece I filed after that night did not defend the referee and did not accuse Sterling. It showed that both sides were reading the same event through two different rulebooks.

What should happen next

A sport that publishes its refereeing ledger openly would change the way arguments are conducted. If every league published a weekly decision log — minute, incident type, number of replays, agreement time with VAR — models like the compensation effect could be tested in public instead of contested by feel. That log would also force broadcast directors to answer for the frames they choose.

For K League 1, which has used VAR since 2026, this is work that could be done in a single season. The data already exists in the video operations room. The question is not technical; it is whether the door gets opened.

As I write this, a new season is running and new controversies will keep arriving. What I want to know is not which referee will err next round, but whether anyone will publish enough data for us to stop arguing on faith. Discipline is not punishment; discipline is a way of reading the match.