On September 13, 2026, in the 60th minute of the Manchester derby at Old Trafford, Erling Haaland scored the only goal of the match. The assistant referee raised his flag. The on-field decision was offside.
The video officials then used semi-automated offside technology, or SAOT, to examine Haaland’s position. The system placed him three centimeters onside, with Manchester United defender Patrick Dorgu keeping him level on the far side of the field. VAR Matt Donohue reversed the decision. The goal stood, and Manchester City won 1-0.
The technology got that question right. The officials still got the decision wrong.
Haaland’s teammate Enzo Fernández had been in an offside position when the ball was crossed. He dived toward the ball without touching it. His movement affected Lisandro Martínez’s attempt to defend the cross and probably affected goalkeeper Senne Lammens as well. Under Law 11, a player can commit an offside offence without touching the ball if his actions interfere with an opponent.
The video officials did not evaluate that question. They concentrated on Haaland, established that he was onside, checked whether Fernández touched the ball and awarded the goal. Pro Ref, the organization responsible for Premier League match officials, later said the goal should have been disallowed.
That conclusion did not come from speculation about what happened inside the video-operation room. The relevant communications were released through the Premier League’s Match Officials Mic’d Up program. The audio allowed supporters to hear what the technology supplied, what the officials examined and what they missed.
That record makes the Manchester derby a useful counterpart to an earlier article about FIFA’s explanation of a disputed World Cup decision. FIFA Media invoked technology to certify a disputed human judgment. Pro Ref eventually did the opposite. It separated an accurate technological output from an incorrect human decision and accepted responsibility for the error.
What the Technology Got Right
The Premier League describes SAOT as a support tool. Up to 30 cameras track the ball and thousands of points on each player’s body. The system proposes the moment when the ball was played and automatically creates lines showing the positions of the relevant attacker and defender. The VAR must review the output and confirm that the system selected the correct kick point, players and body parts.
The system is designed to answer a geometric question quickly and consistently: where were the relevant players when a teammate played the ball?
In the Manchester derby, SAOT answered that question. Haaland had been flagged offside on the field. The technology placed him onside by three centimeters. Donohue reviewed that result and correctly concluded that Haaland’s position did not provide a basis for disallowing the goal.
That was not the end of the offside analysis. The Premier League’s own explanation of SAOT says that the VAR must still decide whether the referee should conduct an on-field review when an attacker does not touch the ball but may have interfered with an opponent from an offside position. The system supplies positional information. It does not decide how an offside player’s actions affected a defender or goalkeeper.
That limit did not cause the error. It identified the next question the officials had to answer.
What the Officials Missed
The published VAR transcript shows that the review began with Haaland. Donohue stated that Haaland was onside but that he still needed to check the full attacking possession phase. The officials then reviewed an earlier touch, checked whether Fernández had contacted the ball and confirmed that he had not.
They never discussed whether Fernández interfered with an opponent.
That distinction is central to Law 11. Touching the ball is one way for a player in an offside position to become involved in active play. It is not the only way. A player may also commit an offence by challenging an opponent, attempting to play a nearby ball in a way that affects an opponent, or making an obvious action that clearly affects an opponent’s ability to play the ball.
The replay showed Fernández diving toward the cross between Martínez and Lammens. Pro Ref chief refereeing officer Howard Webb later explained that the action prevented Martínez from approaching the ball normally and probably affected how Lammens prepared to defend the play. Fernández did not need to touch the ball for those effects to matter.
The officials asked whether Fernández made contact. Once they saw a gap between him and the ball, they moved toward awarding the goal. The audio records Donohue saying, “Go, go, go, so it should be a goal.” He then confirmed no touch and told referee Michael Oliver to change the decision.
Speed alone does not explain the mistake. The review lasted long enough to examine several parts of the attacking phase. The error was narrower. The officials treated the absence of contact as though it resolved Fernández’s involvement. It did not.
What the Audio Revealed
Without the audio, supporters would have been left to infer what had happened. One account might have blamed SAOT for overlooking Fernández. Another might have assumed that the officials considered his movement and decided that it did not affect the defenders. A third might have treated the incident as another unexplained failure of “VAR.”
The recording removed most of that conjecture.
It showed that SAOT corrected the assistant referee’s judgment about Haaland. It showed that Donohue understood that he needed to review the full attacking phase. It showed that the officials examined whether Fernández touched the ball. It also showed that nobody raised the separate question of interference with an opponent before the review ended.
Webb described the mistake as “tunnel vision.” Donohue became focused on the player identified in the on-field decision. The other members of the review team did not redirect his attention to Fernández’s effect on Martínez and Lammens. The IFAB VAR protocol provides several people, camera angles and stages of review. In this instance, none of them supplied the missing question.
The disclosure therefore produced a real explanation. The goal was not awarded because the technology classified Fernández’s conduct incorrectly. The technology did not make that classification. The goal was awarded because the officials stopped after answering a narrower question.
What Pro Ref Got Right
Pro Ref’s response was not perfect, but its later explanation addressed the actual failure.
The organization acknowledged an error of judgment, contacted Manchester United and said the goal should not have stood. It provided the club with the audio and footage from the review. Webb then used Match Officials Mic’d Up to explain both the correct positional finding and the missed interference question. Pro Ref also began reviewing which parts of the process had failed, and Donohue and assistant VAR Blake Antrobus were stood down from the next round of matches.
None of those steps changed the score. They did, however, give supporters a factual account of the decision:
- SAOT placed Haaland onside by three centimeters.
- The video officials reviewed and accepted that result.
- They checked whether Fernández touched the ball but did not assess whether he interfered with the defenders.
- Pro Ref concluded that the omission was an officiating error and that the goal should have been disallowed.
That account preserved the distinction between the system and its users. Pro Ref did not argue that a correct SAOT measurement made the final ruling correct. It stated that the measurement was correct and the ruling was wrong.
The contrast with FIFA is direct. In the World Cup incident discussed in the earlier article, FIFA Media cited connected-ball data as proof that the referee had correctly disallowed a goal, even though the technology could not decide every question required by Law 11. That language made the system appear responsible for the complete ruling. Pro Ref’s later explanation identified the system’s limited contribution and left the remaining judgment with the officials who made it.
What Could Have Been Better
The initial public explanation repeated part of the review team’s mistake. The Premier League Match Centre stated that the VAR had determined that Fernández did not play the ball. That statement was true but incomplete. The controversy concerned what Fernández did to the defenders, not merely whether he touched the cross.
A useful real-time explanation would have said that SAOT placed Haaland onside, while the VAR also determined that Fernández had not interfered with an opponent from his offside position. That would at least have identified the judgment being made. It also would have made the error easier to recognize immediately.
The fuller account arrived later through a television program. Pro Ref could improve the process by publishing the same chronology promptly and in writing whenever it acknowledges a significant error. A written account should identify the technological output, the questions considered by the officials, the omitted issue and the corrective action. Supporters should not need to reconstruct that sequence from several news reports and video clips.
Those shortcomings do not erase the value of the later disclosure. They show the difference between issuing an apology and explaining an error. Pro Ref did both, but the explanation was the more useful act.
Why Disclosure Improved the Diagnosis
An automated support tool can fail. The person using it can also misunderstand its output, ask the wrong question or stop the review too soon. Those failures require different corrections.
If SAOT had placed Haaland onside when he was actually offside, the response would concern camera coverage, calibration, kick-point selection, line placement or validation. That is not what the published record shows. SAOT corrected an inaccurate flag. The failure occurred when the officials applied Law 11 to another player in the same sequence.
That diagnosis directs attention toward the review process. The officials needed a reliable way to move from the player named in the on-field decision to every potentially relevant attacker in the full attacking phase. They also needed someone in the review team to challenge an analysis that had narrowed to contact alone. Training, review sequence and team communication address that problem. Recalibrating the offside system does not.
The audio made that distinction available to everyone, not only to Pro Ref. Supporters could compare the replay, the system’s positional finding, the officials’ conversation and Webb’s subsequent explanation. Disclosure did not ask them to trust that an internal review had found the answer. It gave them enough of the record to understand the answer for themselves.
What Organizations Should Explain
The lesson extends beyond soccer. Organizations increasingly use automated measurements, alerts and recommendations in decisions involving employment, health care, insurance, education and public benefits. When someone challenges a result, a general description of the system is not enough. The organization must explain how the system was used in that case.
The explanation should answer five questions:
- What question was the system asked to answer?
- What did the system report?
- What questions remained for people to decide?
- What evidence did those people consider, and what did they miss?
- What will change because of the error?
The Manchester derby record answers each question. The system measured player position. It reported that Haaland was onside. The officials still had to evaluate Fernández’s effect on the defenders. They considered whether he touched the ball but overlooked interference without contact. Pro Ref identified the mistake and began reviewing the process.
That is what useful transparency looks like. It does not require an organization to claim that its technology is infallible or that its employees never make mistakes. It requires enough information to identify which part of the process succeeded and which part failed.
