A 'Football' Label Glued Onto a Tamaulipas Power-Outage Notice: Anatomy of a Verification Failure

On September 24, 2026, between 09:45 and 17:45 local time, the power grid in Nuevo Morelos, Tamaulipas, Mexico will be switched off so that CFE (Comisión Federal de Electricidad) technical crews can work on equipment. That is the entire content. No club. No player. No scoreline, no xG, no PPDA, not a single line about transfers, wage bills or financial fair play.

And yet that document arrived in my processing pipeline with a single label: Domain — football. In the summer of 2026, I saw the Opta ghost — and from that day, my eyes stopped believing what they saw. In 2026, the ghost came back wearing a utility notice.

The label arrives first, the truth arrives later

In 2026 I was fifty-nine, leaving a traditional print newspaper in Barcelona for a newly launched online sports platform. My first analytical piece covered Valencia's 3-0 win over Las Palmas on matchday two of La Liga. Valencia scored three goals from an xG of just 1.4. Las Palmas posted an unusual PPDA of 7.2 — pressing hard, but with a defensive line so high that every lost ball was a death sentence. A young colleague laughed: you read the spreadsheet before you watched the game. I stayed quiet, then spent three weeks building a homemade xG model and running it across the first seventy-six matches of the season. The question I was answering was not about Valencia. It was this: if the number is right but the label attached to the number is wrong, what exactly are we trusting?

I asked that question in 2026. In 2026 it knocked on my door in the form of a CFE outage advisory.

A pipeline without eyes

Modern sports newsrooms run on four layers. Layer one collects: agency copy, club leaks, press releases, social media, vendor data. Layer two classifies automatically, attaching topic labels such as football, basketball, tennis or general news. Layer three ranks: an algorithm decides what reaches the front page. Layer four is human — or rather, whatever is left of the human after the first three layers have finished.

Only layer two goes unchecked. It is boring. It returns a short string. Nobody wants to be the sixty-eight-year-old man inspecting labels one by one while deadlines drain away.

The CFE notice is a second-type label: the kind that weighs more than the text it is glued onto. Its content, in full: an outage from 09:45 to 17:45 on September 24, 2026; routine maintenance and equipment upgrading; the affected area is Nuevo Morelos, Tamaulipas, bordering Nuevo León; residents should charge devices in advance; businesses with refrigeration or terminals should prepare contingency plans; restoration may shift depending on operating conditions; crews will work on de-energized installations for safety. Twenty-two information points. Not one of them about football.

In the modern sports pipeline, the most dangerous thing is not fake news. The most dangerous thing is true news wearing the wrong label.

Nine drawers, all empty

When I applied the nine-dimension framework — tactics and technique, club finance and the transfer market, results and the opinion cycle, league landscape and positioning, rules and governance, management and dressing room, risk profile, media narrative, industry transmission — every drawer returned the same answer: insufficient information.

What matters is not that the drawers were empty. What matters is that I had to write out the emptiness nine times. Because the first reflex of anyone ten years into this trade, handed an empty input, is to fill it.

I know exactly how that temptation works. You read Tamaulipas, you read Nuevo León, you remember Monterrey, you remember Rayados and Tigres. Could the outage affect a match? Could a stadium be inside the affected zone? Could a training session be interrupted? Three questions. Three bridges. Not one of them with a pier underneath.

That is how this industry produces most of its content. Not by inventing events, but by building bridges between a real event and an unrelated topic, then letting the reader walk across and believe the bridge was always there.

In the tactics drawer there is not one tactical term. In the finance drawer there is not one monetary figure — the only cost mentioned is business interruption for electricity users, an operating cost, not a football variable. In the results drawer the sample is zero. In the league drawer, the place names are administrative units. They are real. They have borders. They do not have a table. In the governance drawer, the only regulation-like content is a technical safety procedure. In the dressing-room drawer, no individual is named; the only actor is an institution. In the risk drawer, all six standard football risk categories return empty, and the seventh — systemic risk — returns high, but its content is input-integrity risk, not sporting risk. In the narrative drawer, the document is a neutral public-service announcement. In the transmission drawer there is no academy-to-broadcast path — but there is a real one: a data-pipeline transmission, in which a non-football item travels through a football-labelled pipe.

When the stadiums fell silent in 2026, I understood something: football never died, it merely took off its clothes and revealed its skeleton. At sixty-two I had real-time data access to a second-tier Catalan club playing to empty stands. Home win rate fell from forty-six per cent to thirty-eight per cent. Passes into the final third rose eleven per cent. I wrote a long essay about lost space and digitised psychological pressure. It was not about players' feelings. It was about structure.

The CFE mislabel also reveals a structure. Only that structure belongs to us, not to Mexico.

When correlation claims to be causation

I have spent most of my career fighting one sentence: this team won because of spirit, that team lost because of a lapse in concentration. I do not deny spirit or lapses. I deny using them as explanations when better data exists.

But there is a symmetrical error that we in the data wing rarely confess. It is using correlation to prove causation when the correlation is beautiful.

In the summer of 2026 I predicted France would win the World Cup, based on the share of passes into the final third by the French U21 cohort and on Antoine Griezmann's average shot xG of 0.21, above the benchmark for elite forwards. France won. A Spanish editor told me: you were right, but nobody reads the way you write. I went home and wrote one line in my notebook: correct data told badly is dead data.

There is something I never published. That prediction was partly luck. Run the same model across ten World Cups and I might be right four times. If I do not admit that, I am doing exactly what I criticise in others: building a bridge from correlation to causation and letting the audience cross it.

On the Moscow night I did not sleep. Not because of football, but because the numbers were whispering a prophecy. Only in 2026 did I turn it into a sentence: a number speaks only what it speaks once you have verified where it belongs, when it was measured, by whom, and under which definition.

The number I received in September 2026 was a label. It said football. An uncalibrated machine would start generating football from it. I know exactly what it would write, because I have read hundreds of such pieces this transfer window. It would open with infrastructure pressure. It would mention Monterrey a few hundred kilometres away. It would ask whether a match might be affected. It would close by saying Mexican football must prepare better for infrastructure incidents. Every sentence grammatically correct. The whole piece ontologically wrong.

The transfer market is a monastery where the numbers chant; I merely transcribe what they pray. And every summer there are services where nobody checks which religion the officiating priest belongs to.

Three traps the wrong label conceals

A wrong label never travels alone. It drags three consequences, and all three are existential for this industry.

First, esports. A professional player's career is shorter than a footballer's — peak at nineteen, exit at twenty-five — while the youth system and post-retirement support are close to zero. When a mislabelled document is quietly pushed downstream, I think of those players. They are also pushed downstream in silence. Nobody labelled them, so nobody checks.

Second, the commercialisation of women's sport. Women's competitions are folded into sponsorship packages as a corporate social responsibility line, counted in ESG reports, photographed beside executives. When broadcast infrastructure fails, when fixtures pile up, when budgets are cut, that line is the first to vanish. It happens because women's sport is given a label — a moral label — rather than a market label. A wrong label always has a price. The person paying is never the person labelling.

Third, injury transparency. Clubs publish only the injuries that suit their asset values. A knock that would depress the price of a player being shopped is described as minor. An injury that raises pressure on the coaching staff is disclosed days late. Medical confidentiality is a sound principle, but it has been converted into a screen for a communications strategy. That strategy runs on one assumption: that nobody will check.

These three are versions of one error. An entity is given a label that does not match its nature, and the label is allowed to travel further than the entity.

I am sixty-eight, but the data is younger than I have ever seen it — every season it grows another set of teeth.

The other side of the desk

When I was young I believed the problem with sports journalism was a shortage of data. In 2026, joining the sports desk of Belgrade Television, we had three sources for a match: the eyewitness, the tape, and the reporter's memory. Memory was the weakest source and the most used. I told myself that if we ever had complete data, the trade would be clean.

We now have complete data. The trade is not cleaner. It is only faster.

Between 2026 and roughly 2026 I hosted and produced a football night programme. I learned that a sports broadcast does not sell truth. It sells tempo. A correct frame that is slow dies in front of an incorrect frame that is fast. I thought that was a television problem. It is a system problem.

The football label on CFE's outage advisory is the natural output of such a system. Nobody intended it. Nobody fabricated anything. A classifier simply ran faster than verification. In a system that rewards speed and does not reward checking, correct classification becomes luck rather than process.

I once believed in instinct. After Opta, I believed in probability. After COVID, I believed in structure. After September 2026, I believe in the gate.

The gate

Three proposals, offered not as technical fixes but as ethical conditions of the trade.

One: every topic label must carry a negative condition. A document may be labelled football only if it contains at least one minimum entity — a club name, a player name, a competition name, or a verifiable match event. If it contains none, the label is suspended and routed to a human. The cost is near zero. The benefit is an entire family of errors eliminated.

Two: when input is insufficient, the correct output is an empty answer, written explicitly and pushed back up the chain. In my trade, saying not enough information is a professional answer, not a confession of weakness. But for that answer to be permitted, the newsroom must reward it. If an editor docks me for returning empty, I will fabricate. That is the entire mechanism. There is no mystery.

Three: when one mislabelled item is found, audit the batch. A classifier that fails once almost always fails repeatedly, because it fails systematically, not randomly. I checked. In the same batch I received that week, the error rate among items filed under sport was high enough that any careful person would stop.

None of this requires better artificial intelligence. It requires one person willing to read.

What I keep

A beautiful number is like a perfect pass: it needs no explanation, only to be seen. A beautiful label is different. A beautiful label does not need to be seen. It only needs to be believed.

I have kept the CFE notice in a folder named by date: September 24, 2026. Beside it I keep three other things. A printout of my first 2026 analysis of Valencia and Las Palmas, creased where a colleague laughed. A page of notes on home win rates in the empty-stadium season. And a slip of paper taped to my desk throughout this transfer window, carrying one sentence: a label is not data.

I do not know whether that label will appear again next autumn. I know what I will do if it does: reopen the folder, reread the notice, and write once more that there was no football club in Nuevo Morelos on September 24, 2026.

The question I leave to successors thirty years younger is not how to make the classifier more accurate. It will become more accurate. The question is: when a machine glues a football label onto a power-outage schedule, who in your newsroom is slow enough to stop it.

In fifty-two years in this trade I have never seen anyone paid for being slow. I have seen the price of not being slow, and I have just seen it for the second time.