Howard Webb's Apology and the Data Void in the Manchester Derby

In January 2026, at Old Trafford, Marcus Rashford sprinted onto a through ball while standing in an offside position. He never touched it. Bruno Fernandes touched it, scored, and the goal stood. That night I logged every stride of Rashford's run into my match notebook: minute in the left column, position in the middle, and in the right column a question I could not answer — if a player never touches the ball, is he interfering with play?

More than two years later, a near-identical situation surfaced in the Manchester derby. This time English professional refereeing admitted the goal should have been disallowed, and Howard Webb, the head of the Premier League's refereeing body, publicly expressed regret over how the process was handled. Same law. Same type of incident. Only one apology.

The best sportswriter is one who knows his own notebook can lie. I wrote that line on the edge of the page before opening my laptop, because this story has too many layers for a notebook not to deceive itself. A red card. A contested goal. A late apology. And a manager standing in the middle of it, defending his players while demanding something harder to measure than three points: consistency.

The backdrop has to be built before we get to referees. The Premier League season is only five matches old, and the Manchester club sits top with a perfect record. Last weekend they travelled to Old Trafford, where, based on my experience covering matches, every technical metric is distorted by the noise of the stands. They played over seventy minutes with ten men after Phil Foden's red card, and still won.

The cost of that win was not in the table. It was in a string of decisions the authorities later had to explain, and in Erling Haaland's goal being placed under a microscope. When a derby ends and the first fifteen minutes of the press conference belong to referees, you know you are reading a different story from the football one.

The visiting manager called the officiating "quite poor" and said his side was still waiting for an explanation. Howard Webb then admitted the error. In PGMOL's language, a public admission is not routine. It appears only when an error is judged too clear to defend.

To read the story properly, go back to Law 11. A player in an offside position is penalised only if, in the referee's judgement, he becomes involved in active play. The law defines involvement through four branches: playing or touching the ball; challenging an opponent; making an obvious action that impacts an opponent's ability to play the ball; and gaining an advantage from that offside position.

The 2026 incident was argued on the first and third branches. Rashford ran after the ball — an action suggesting intent to play it — but did not touch it. Defenders and the goalkeeper had to react to a man closing on the ball, so the "challenging" argument had a basis. The goal stood. And no apology came at management level.

Context matters here. That January 2026 derby sat inside the run-in of a title race, when every point felt heavy. The goal directly changed the match and had consequences for final standings. The absence of a public apology does not prove it was correct. It only proves the threshold for reaction was different.

This time the refereeing side reached the opposite conclusion and said so publicly. From the outside the two incidents look similar enough that the consistency question becomes natural: if the criteria have not changed, why did the conclusion?

The likely answer lies in interpretation itself. The criteria in the law do not change, but the threshold for applying them does — and that is exactly the kind of movement no statistical table captures. I can count flags, VAR interventions, seconds of delay. I cannot count a judgement threshold.

The VAR protocol also needs describing properly. VAR does not re-referee a match; it intervenes only on clear and obvious errors, or serious missed incidents. "Clear and obvious" is a qualitative threshold, and qualitative thresholds shift with people and with years. In the derby, the error was located in the interpretation of offside interference, not in the technology.

The hardest part of offside law is that it asks referees to read intent. Was a retreating defender affected by a runner in front of him? Was a goalkeeper distracted coming out? No sensor measures distraction. That is why these incidents stay hot, and why they will never be fully resolved by technology.

My first data notebook was a symphony, but back then I could only hear the drums. It took years to understand that one column of numbers only means something next to another — and here the second column is missing entirely.

The basics any analysis department would ask first are absent from coverage of this derby: no expected goals, no passes per defensive action, no possession split by half, no shot map. Without raw data, every claim about a ten-man performance is description. Description has value, but it does not replace measurement.

That lesson came from another evening, further away. At a World Cup I built an entire analysis on passing data and watched it get ignored, while a colleague writing about dressing-room chaos went viral. The lesson was not that data is useless. The lesson is that data does not tell its own story; it only confirms a story someone else has already told.

Based on my experience covering matches, seventy-plus minutes a man down away at Old Trafford almost always follows a familiar pattern: a low, narrow defensive block, compact lines, conceding the middle third, and hunting chances in transition. A possession side with ten men is physically hard to imagine. That is inference, and I marked it as medium confidence in my notebook.

More interesting is the manager's comment about the League Cup. The early rounds have no VAR, and he declared he would not change his approach because of it. For someone who reads data professionally, that is a note worth keeping: a coach refusing to treat officiating conditions as a tactical variable. Philosophy over circumstance. It produces admirable consistency, and it also produces higher variance in one-off knockout ties.

Operationally, the early League Cup rounds have long been a laboratory for big clubs. No VAR means errors go uncorrected, which raises result variance. A side rotating seven or eight positions with no technological safety net is far more vulnerable than its record suggests.

On fixtures, midweek brings an away League Cup tie at Norwich, a lower-tier opponent. At the weekend, Sunderland at home in the league. To be clear: this is not a crisis schedule. It is an ordinary two-match window in an ordinary season. But it sits immediately after an emotionally charged derby, which is the classic recipe for a stumble.

The Sunderland match means something different. For a side on a perfect run, every domestic opponent is approached the same way: control the ball, stretch the pitch, make them run. That approach only works while the physical base holds. Three days after a cup tie, that base is a question mark.

One item deserves separating from the refereeing story: Foden's red card. In my notebook it is the only hard fact of the week — a disciplinary event with defined consequences, independent of interpretation. The length of the ban determines how many matches the attack loses a link, and that affects the winning run far more than any VAR debate.

Notably, most coverage treats the red card as a supporting detail for the "won with ten men" narrative. It becomes background, when in reality it is an unpaid disciplinary debt.

Now the difficult part. The "double standard" argument carries emotional weight and has a principled basis. But its evidence is thinner than it appears. Two incidents more than two years apart, two different officiating teams, two different match contexts, and an interpretive threshold that may have shifted. "Same law, same outcome" is logically sound, but it ignores that the law is written in qualitative language — "clear", "impact", "advantage".

The strongest part of that argument is the asymmetry of the apology, not the two incidents themselves. If a situation is judged clearly wrong recently and publicly apologised for, then the absence of a similar apology in 2026 is either the product of a different threshold or a different attitude. Both possibilities require PGMOL to publish its criteria, not just its apologies.

In my notebook, each offside-interference incident gets three lines: position, action, consequence. The third line is always hardest, because the consequence of an action without a touch often exists only in an opponent's head. Logging it means logging a guess, and I always mark such lines with a question mark in the margin.

There is a further layer rarely mentioned. A new manager, newly empowered, publicly attacking officiating and demanding explanations. In a dressing room, that is usually read as protecting the collective, and it strengthens internal belief. It is also a way of directing attention outward. Both can be true in one sentence.

When data fails, I start believing the story has its own power. And this story has a very specific problem: the chain of events as retold does not match itself.

In the aggregated accounts I cross-checked, the Manchester club's manager is described with a title that does not reflect reality — the man was an assistant at that club, then worked elsewhere, and never managed the side in the derby in question. Another name, belonging to a player at a different club and a different national team, is assigned the role of "interfering with the ball" in that same derby. Details like these are not harmless typos. They show where the verification chain broke, and once it breaks, readers can no longer separate observation from inference.

This is the clearest point I want to make: internal contradictions in the reporting do not make the manager's complaint baseless. They simply make the reporting unusable as evidence. Those are two entirely different things, and mixing them is the fastest way to lose both.

There is a structural reason such stories outlive their substance. Controversy generates engagement. Aggregation platforms have an incentive to keep a controversial story circulating regardless of whether it has run out of information. A story's heat and its evidential weight run on two different clocks.

I think about that whenever I see a prompt inviting readers to set a site as a preferred source in a search engine. It signals a traffic-driven business model, and such models tend to favour speed over accuracy in the opening lines. Nothing wrong economically. But it explains why a perfect winning run is retold mainly through a debate.

On social media, reaction splits into two familiar camps: one says the referees are biased, the other says the manager is looking for excuses. Both miss the hardest part — that football's laws are written to accept a degree of interpretation. The volume of a debate is not proportional to the quality of its evidence.

Back to football. Five wins is a good start, and a ten-man away win is a fact worth recording. But both belong to the low-repeatability category. A heroic result is not proof of invincibility; it is a small sample told in a loud voice. I wrote that in my notebook after the match, under "risk", not under "achievement".

Coverage also barely touches injuries or workload. For a side that just played seventy-plus minutes a man down and now faces two matches in seven days, that is a far more worrying blind spot than a refereeing debate that will fade within days.

The coming week will answer much of this. Whether PGMOL publishes specific criteria for interference from an offside position, or stops at an apology, will decide whether the admission becomes reform. Foden's ban length is the only immediately quantifiable number in the story. And the cup line-up midweek will reveal where the club's real ambitions sit.

I do not write to predict; I write so that someone reading this later will know how we lived. If this notebook survives a few more years, I want today's entry to be about a league learning to write its criteria down.