The Mislabel Audit: A Metro Incident That Entered the Football Database

Two in the morning in Rangpur. A laptop open on the table, a cup of tea gone cold beside it, a CSV file on screen. I put the cursor on row four thousand four hundred and seventeen and read the column again. Domain label: football. The column next to it: Mexico City Metro, Line 1, Moctezuma station, a metal crutch falling onto the electrified track, a short circuit across a 750-volt DC supply, a loud bang, smoke, passengers running from the platform.

No club. No player. No coach, no competition, no transfer, no federation. The label says football anyway.

I counted the nulls in the analysis report that came with the file. Nine times it says ‘N/A — insufficient information’. Those nine empty cells are the real document here. Not the crutch.

Years of watching matches has given me a habit: I do not look at the scoreboard first, I look at the team sheet. Who signed it, who is missing, whose name has been struck out, who is listed with a fitness complaint. I read a data feed the same way now. In this file I found the sheet, no players, no signature, only a label.

Sports news is no longer written by reporters alone. Agency feeds, wire summaries, CMS tags — three layers hold up the modern sports desk. One field decides where an item goes, which desk receives it, which model swallows it. The label is the load-bearing wall of that building. Put the wall in the wrong place and everything built above it arrives at the wrong address.

I am not going to discuss formations, pressing structures or expected goals here, because the source material contains no football. Instead I want to open the document itself and see how an urban transport incident acquired football citizenship, and whose signature is on the papers.

The event, briefly. Mexico City Metro Line 1 runs from Pantitlán to Observatorio. At Moctezuma station a metal crutch slipped from a passenger's hand onto the electrified track, contacted the guide bar carrying 750 volts of direct current, and short-circuited. Noise, smoke, alarm. Nobody was injured, according to the transport authority. Power was cut, the object removed, track condition verified, service restored, and a train sent to the workshops for a preventive inspection — a sequence stated in four separate places in the report and internally consistent every time.

One sentence carries the most weight: the incident at Moctezuma ‘once again highlights a frequent problem’. Alongside it, ‘hundreds of cases’ between January and October. How many hundreds, in which year, against what denominator of passengers or trips — none of that is there.

I read transport reports regularly, for personal reasons. In 2026 a file of Bangladesh Premier League gate receipts landed on my desk and looked just as clean at first. The totals reconciled. The categories reconciled. Only one thing did not: the ratio between the people in the stands and the money in the bank. Three franchises had underreported revenue by 40 percent, moving 2.3 million dollars, and twelve players were short by 340,000 dollars between them. Those numbers survived an audit because they had denominators — how many spectators, how many matches, how many contracts. That is the only reason anyone ever reached the audit stage.

So I stopped at four places in this metro report.

First stop: the label is a document, and nothing on the document is signed. Every file I have chased in my career carried a human trace somewhere — a clerk's initials, an office stamp, a date box. The 2026 file had three signatures, three different people, three different departments. This file has a box reading ‘football’. Who wrote it? Under what rule? Which words in the item convinced the system? No name, no log, no correction path. A classifier most likely decided from a newsroom URL path or a shared CMS feed — meaning the label came from a signal outside the document rather than from evidence inside it. A document without a signature carries no liability, and a document that carries no liability cannot be held to account when it is wrong.

Second stop: the evidence was in the empty cells. Nine fields marked ‘N/A — insufficient information’, each followed by a warning: do not read these as ‘no risk detected’, read them as ‘no subject present’. That is not a technical footnote. That is a governance ruling written in technical language. Because whoever sits downstream and sees an empty table will conclude that the absence of a team means the absence of risk. An empty cell gets read as safe, never as missing. In my trade that mistake has flesh on it. At the 2026 World Cup in Russia, one team's players were taking injury timeouts at exactly twenty-three-minute intervals across three matches. Which box was empty in the medical file? The reason box. No reason was written, so nobody saw a risk. The empty cell is the place where an institution spends the least effort hiding its responsibility.

Third stop: the causation rests on a single source. The crutch explanation comes entirely from the transport operator and the capital's transport authorities. No fire service statement, no independent electrical inspector, no named eyewitness. Every attribution is institutional — ‘authorities indicated’, ‘the operator explained’. The explanation is plausible. It has no crack in it. Metal object, 750 volts, short circuit, bang, smoke. Four information points, consistent. And that is exactly where my profession's oldest trap sits: consistency is not corroboration. The masking agent I wrote six thousand words about in 2026 was not on the banned list. The protocol was followed. The paperwork had no gap. The whole apparatus was internally perfect, which was the problem. I delayed that piece by two weeks to re-verify every source. Where are those two weeks here? The cause travelled the world within hours on one source's word.

Fourth stop: recurrence without a denominator. ‘Hundreds of cases’ from January to October, with no exact figure, no year, no comparison against total passengers or total trips. My trade has a name for this move: turning a number into a mood. A reader hears ‘hundreds’ and believes. An analyst can model nothing. Imagine the 2026 franchises had written that ‘attendance was somewhat low in some matches’. Would there have been an audit? There would not. The audit happened because 40 percent and 2.3 million dollars had denominators. A number without a denominator is not a charge, it is weather.

Put those four stops together and a structure appears — what I call the Loophole Eleven, eleven institutional exemptions that let today's sports data system avoid accountability. First: nobody performs the basic entity check, whether an item contains at least one club, player or competition. Second: institutional statements are accepted as quotes without a name attached. Third: single-source causation passes, with no requirement of independent corroboration. Fourth: numbers without denominators pass. Fifth: vague durations pass — ‘several minutes’ — so the reader who wants to know how long service was suspended never finds out. Sixth: relative dates pass — ‘last Tuesday’ — so the file permanently loses the absolute date. Seventh: incidents are counted and incident costs are not. Eighth: nobody publishes a mislabel rate. Ninth: there is no correction or retraction window. Tenth: there is no dividing line between ‘the authority's account’ and ‘verified fact’. Eleventh: there is no signature.

All eleven work identically in football, and I have seen the precedent for each. In transfers, the fee a club announces is the headline number; the real story sits in the sell-on percentage in the footnote. The sell-on clause was a footnote; the footnote was the story. Automated pipelines do not read footnotes, they lift the headline. In age falsification, the birth certificate is a label, and when the label is wrong the entire decision chain beneath it — the scout's report, the trial invitation, the family's loan — runs in the wrong direction. For years I have watched scouting networks in poor countries discover genius on one side and break households on the other through the football lottery. The first step of that breaking is often a single wrong cell: a date of birth. And in 2026, at a mid-table La Liga club, I found two hundred staff furloughed while fourteen million euros went out in agent fees. The money had not vanished; it had been walked through three shell companies in Cyprus and Malta and relabelled ‘agent fees’. The missing gate receipts were not missing; they were renamed. The metro report does the same thing in miniature — the incidents are not missing, they have been renamed into a vague category called ‘hundreds’.

One more thing struck me, and it is directly relevant to sports economics. The metro authority counts incidents and does not count the cost of incidents. Yet every event carries a fixed cost: cutting power, a specialist crew, an inspection, possibly a train withdrawn to the workshops, thousands of passengers losing time. That is a classic low-probability, high-frequency, moderate-cost risk. Football clubs make precisely this error with injuries. They count injuries; they do not cost them. How many matches lost, how many points dropped — that is the whole ledger. How much medical spend, how much asset value depreciated, who signed the contract — in nobody's file. This is where the doctor enters. — Root: The Doctor. The man who signs the fitness certificate is also a salaried employee of the club; that dual position appears in no document, because no document is required to record it. The seventh exemption of the Loophole Eleven is the most expensive of the eleven.

Watching matches for years has built another habit: before kickoff I watch the warm-up. Who has tape on an ankle, who is not putting weight on the left foot, who is stretching longer than the others. I have transferred that habit to data feeds. When an item arrives I ask who is inside it — names, roles, positions. Then who is speaking — a named human, or ‘authorities’. Then where the denominator is. Three questions kill half of football's ‘exclusives’ on contact. A feed that cannot answer those three questions is not news. It is volume.

Now let the objection come, the one I hear most on the desk: this is one bad tag, fix the tagger and move on. The objection is reasonable and stands in the wrong place. Fixing the tagger is cheap; auditing it is expensive. And in a system where volume is the product, auditing is a cost centre, not a revenue line. A newsroom pushing three hundred items a day does not care about the accuracy of each item; it cares about the average, and the average is never visible to the reader. That is exactly why no one publishes the mislabel rate. Publish it and the problem stops looking like one row and starts looking like the system.

The second thing critics miss is more uncomfortable. The analyst who wrote ‘N/A’ nine times did the most honest work in the entire chain. He refused to fill an empty cell with a story. My profession punishes that honesty. A report full of ‘none’ gets called weak by an editor. A report with a twelve-page money flowchart takes three months and reaches a tenth of the audience of a hot take. I know, because I have done both. The arithmetic still favours caution: one retraction costs more than one long investigation. In 2026 the veteran reporters in the press box dismissed me until the audit arrived. After the audit nobody said anything. That was the only answer I needed.

The third thing everyone will forget is the crutch. Everyone will remember the crutch; nobody will remember platform screen doors. Yet the report itself says hundreds of objects fall every year, and lists the crutch alongside canes, umbrellas, keys and thermoses. After that many incidents, no policy commitment on platform-edge protection is quoted anywhere. The question is raised and not chased. Football does this with everything off the pitch: everyone discusses the missed penalty, nobody asks who signed the medical clearance. And one detail is absent from every account — a crutch means a passenger with limited mobility. How that passenger boarded, who stood beside them on the platform, is nowhere. I keep two files, one of money and one of people. This report has no money file, and the people file has no names in it. At the centre of an incident involving a crutch, a person was certainly present. Nobody counted them.

So what comes next? Three demands, all verifiable. One: a mandatory entity gate before any football aggregate — the system answers whether an item contains at least one club, player, coach or competition; if not, the item does not reach stage two, it goes to general news. Two: publish the mislabel rate every quarter, with the number of corrections. An institution that cannot print its own error count has no signature behind its accuracy claim. Three: beside every institutional statement, a field asking whether it was independently verified. Where the answer is ‘no’, the reader at least knows whose voice they are hearing.

The file is still open on my table. Row four thousand four hundred and seventeen is still sitting there labelled football. Next month there will be more rows like it — a train delay, a road accident, a budget speech, all turned into football, because someone writes the label and nobody reads it. The question is not a technical one. It is this: when a report is published in perfect structure, in exact language, at the correct length, about an event that never happened on a football pitch — whose name goes under it as the reporter?