The 'Sanction' That Was Never Football's: A Case File on a Vocabulary Collision

Two in the morning. I opened the last file of the stack — the Stage-2 deconstruction, the paper that sits inside the pipeline. The first page carried a domain label: football. Not one of the eighteen information points beneath it contained a club, a player, a coach, a league or a fixture. What it contained was a pothole in Mexico City, a few articles of civic law, and a question ordinary people actually ask: if I fill a pothole myself, can I be fined?

The label lied. It did not lie for no reason, and that is the case file worth opening.

The story itself is harmless. The Civic Culture Law of Mexico City does not say that filling a pothole is an offence. Liability attaches not to the repair but to the conduct around it — occupying a roadway or public space without authorisation, obstructing movement. Article 27, fraction IV, and Article 28, fraction II, carry that wording, along with a carve-out: where there is justified cause, no infraction arises. The text cited by the capital's Congress is the current text. Neighbours who organise to repair a road, and an administration that wants a permit first — that friction is the article's real subject.

That is all. What happened next in the pipeline is the actual document.

The classifier tagged it as football. Hunting for the cause, I found eight words that sit identically in two entirely separate rulebooks: sanction, infraction, fine, regulation, authorisation, jurisdiction, appeal, review. They live in football's disciplinary code. They also live in municipal administrative law. When keywords match, a classifier does not err — it obeys its own definition. The fault is not the machine's. Nor is it the fault of the person who first borrowed that vocabulary.

Over two decades, football administration has redrafted its own rulebook to read like municipal law. Financial fair play in Europe, profit and sustainability rules in England, 115 charges against Manchester City, points deductions for Everton and Nottingham Forest, independent commissions, appeal tribunals, a court of arbitration for sport — these are no longer merely football rules. They are administrative-law texts. A reader who picked them up would assume he was holding a city council's bye-laws. Tagging a pothole case as football is therefore not solely a machine's mistake; the two worlds genuinely speak one language.

I know that language. I spent eleven years in marine insurance — I know a clause when it bites. In 2026 I left a stable salary to work freelance on football governance. In March 2026 I was sitting in the Anfield press box, one of two women among forty seats. The IFAB annual general meeting rewrote the handball clause of Law 12 and formally embedded VAR in the Laws of the Game. I published a four-thousand-word close reading — no editor, no outlet — showing that the word deliberate had been quietly replaced by unnatural silhouette, and that this had happened eighteen months before anyone said it out loud. Nine hundred people read it. Four hundred of them were referees.

The clause was buried on page ninety-four. That is where the match was lost.

Why does that substitution matter so much? Because when the word moves, the burden of proof moves with it. Deliberate requires the referee to enter a player's head and hunt for intent. Unnatural silhouette does not — he simply looks at the shape of a body. Liability stops resting on the person who doubts and starts resting on the defender who raised an arm. It is a silent, almost invisible reallocation, and no league table records it.

The logic in the Mexican document — a general rule, then specific conduct, then a conditional infraction — is exactly the logic embedded in football's disciplinary code. The interesting part is where the exception sits. Mexico's article says no liability where there is justified cause. Football's equivalent sentence is clear and obvious error. Both sentences do the same work: they hand the decision-maker a licence to exercise judgement. Where such a sentence exists, every ruling is ultimately an interpretation, and interpretation is jurisdiction.

When VAR arguments erupt, people talk about technology. I do not. On 16 June 2026, in Kazan, referee Andrés Cunha waved away Antoine Griezmann's tumble against Australia, then went to the monitor and reversed himself — the first VAR-awarded penalty in World Cup history. France won 2–1. I had forty minutes to file. I ignored the was-it-a-penalty argument entirely and wrote eleven hundred words on the burden of proof Cunha had just shifted. The on-field decision stopped being a decision and became a hypothesis.

The crowd saw a foul. I saw a question of who had the authority to say so.

That is a jurisdiction question, and it is also a classification question. Which world a sentence belongs to determines which article, which committee, which territory governs it. A wrong label is a wrong jurisdiction, and a wrong jurisdiction is a wrong verdict.

In 2026 the Bundesliga returned behind closed doors and I stopped watching football as sport and started watching it as a dataset. I logged all eighty-one matches of the nine-round restart. The home-win rate fell from 43 per cent to 33 per cent. And IFAB's temporary five-substitution amendment — meant to protect players after a ten-week layoff — was being used tactically before half-time in sixty-one of those matches. I published the table with one sentence: this rule is not temporary. IFAB made it permanent in 2026.

Numbers are a language. But a number without a jurisdiction is a rumour with a decimal point. That is precisely where the cost of this mislabel sits.

If a municipal fine enters a football-finance model and nobody verifies it, the dashboard does not lie — the dashboard tells the truth flawlessly, in the wrong world. Multiply that error a few thousand times in a training set and the model learns that sanction always means football. When the border between words collapses, the border between numbers collapses too.

Transfer windows spread the same disease faster. Sanction, fine, release clause, authorisation — these words now circulate daily in the rumour machine. One report says a fee is agreed, another says the deal is done, and nobody measures the distance between fact and inference. Information without jurisdiction — which party's document says it, whose signature is on it, from when it takes effect — is not merely worthless, it is a declaration. The real story lives in the clause structure and the wage bill, not in the headline.

Now consider the conventional explanation: the classifier misfired, keyword ambiguity is to blame, fix the label and the problem ends. That explanation is correct but incomplete. Basketball did not do this to itself; football did. If regulation, authorisation, compliance and infraction sit on the first page of football's own rulebook, a machine will identify them as football, and blaming it is unfair. The machine only handed our language back to us.

Second, and more uncomfortable: the article's own sourcing was anonymous. The sourcing quality was already in question before the label went wrong. The anonymous field is the safe corner where bad labels survive. Weak sourcing and weak classification travel together; you cannot fix one and leave the other.

The largest risk, though, is not this article. It is a good dashboard standing on top of it. A bad article is visible. A bad dashboard is not.

And before pointing at machines, remember that humans make the same error. Analysts disappear down the footnote, clutching one fascinating sub-clause while the match was decided on the adjacent page. The process is then immaculate and the conclusion is in the wrong world.

So my proposal concerns rules, not robots. Any classification carrying a football label must contain at least one football entity — a club, a player, a competition, a fixture. If that condition is not met, the item returns to the civic and legal shelf and never enters the football dataset. Add an updated list of ambiguous keywords, and audit a sample of football-tagged items regularly. If the machine will not stay quiet, the paperwork should not either.

One question remains. If a reader cannot tell which rulebook's verdict he is reading — a municipality's or football's — then the mistake was never the classifier's. The mistake was our language's.