A 'Football' Label Stuck on an Earthquake Bulletin

A 'Football' Label Stuck on an Earthquake Bulletin

23,000 seismic alert loudspeakers. 80 million mobile phones inside the coverage area. Mexico's Second National Drill, fixed at 12:00 on 19 September 2026. No club in it. No player. No manager. Not one pound of transfer fee. Yet on the metadata line above the data file, the machine had stamped a label: football.

I read that extract close to midnight, London time. Eleven information points. The first mentioned President Claudia Sheinbaum. The fourth mentioned her morning press conference. The seventh described five scenarios by seismic region. The last described response protocols and the decision to keep the familiar alert tone so residents would not confuse a drill with a real earthquake. Having read all eleven, I could not find a single football entity: no club, no league, no federation, no contract.

The automated analysis behind it still ran all nine professional domains. Tactics and technique. Club finance and the transfer market. Results and the opinion cycle. League landscape and club positioning. Rules and governance. Management and the dressing room. Risk profile. Media narrative. Industry transmission. All nine returned the same sentence: insufficient information, cannot assess.

That result is correct. And it is also a wake-up call.

The Labelling Machine Has Outgrown Its Checkers

Over the past decade, the football content industry shifted to an assembly-line model. A match ends, event data lands on a server within seconds. A transfer story appears, and the system classifies, labels, rewrites and splits it into ten versions for ten language markets. Nobody in that chain needs to watch football. They only need the right label.

The principle is simple: the label determines the analytical frame, the frame determines the conclusion, the conclusion determines the headline. Get it wrong at the front and you are wrong at the back. A mislabelled file does not correct itself. It merely waits for a writer fast enough and unchecked enough to turn it into an article.

Based on my experience tracking matches and data pipelines over ten years, the machine always moves faster than the checker. That gap is where junk content breeds.

I have seen the consequence of that mechanism at a much smaller scale. In 2026, as a seventeen-year-old intern at a London local paper, I went through the academy accounts of Leyton Orient. After three weeks of cross-checking, I found thirty-seven sponsorship contracts with unusual refund clauses, money routed through a shell company in the British Virgin Islands. When I took it to my senior editor, he smiled and said girls tend to watch football with emotion, not with ledgers. I did not argue. I re-ran the data, wrote it up, and posted it on my personal blog. That night the piece passed twelve thousand reads.

The lesson from that year was not the readership. It was this: a single label — emotional, not expert enough — can halt a verification process before it even begins. Today's labelling machine does exactly what that editor did, only at the scale of a million files a day.

Football is the perfect habitat for this error, because it generates mountains of disconnected data: scores, cards, distance covered, transfer fees, wage bills, broadcast rights, squad values. Plenty of data, a huge audience, a news cycle measured in hours. No other industry is easier to flood with junk.

Nine Analytical Domains, Nine Blank Returns

The twenty-page analysis I read that night had one genuine virtue: it did not invent anything. On tactics, it stated plainly that there was no data on system, formation or playing style. On finance, it recorded no transfer fee, no wage bill, no loan, no financial fair play indicator. On rules and governance, it confirmed that the governing framework in the source was a national civil-protection system, not the rulebook of any football federation.

That is the correct handling: state the missing information instead of guessing.

Now place the same file inside a pipeline that lacks that rule. The story unfolds differently, and I can reconstruct it almost step by step.

Step one, the system borrows 23,000 as the capacity of a mid-sized stadium. Step two, it turns 80 million into the broadcast reach of a major league. Step three, it reads five scenarios by region and translates them into a five-man back line. Step four, it spots an authority figure named Claudia Sheinbaum attached to a major decision and files her as a club owner or sporting director. Step five, it takes response protocols and turns them into a high pressing scheme.

By step six, a headline exists. And that headline carries enough numbers to look credible.

This is the part I want to keep longest. The danger is not that the machine knows nothing about football. The danger is that it knows too many football sentence patterns, enough to fill every blank with a plausible template. Crude fabrication gets caught. Fabrication with figures, citations and clean formatting does not.

I once mispronounced Perišić's name three times on camera in Moscow in 2026, and a local commentator mocked me, saying I should go back to the keyboard. I was humiliated. But I did not correct it by shouting louder. I corrected it by opening the English FA's financial records and reading. There I found a 2.7 million pound compensation match fee paid to a sponsor that never appeared in the official report. I have mispronounced Perišić, but I have never been wrong about what I witnessed — and that is the only line I hold.

That line is eroding at industrial scale. An earthquake bulletin labelled as football causes no immediate harm. It just creates one junk file in the archive. But when thousands of such files flow into the same pipe, readers begin to lose the ability to tell analysis apart from a template poured into a blank space.

And the damage does not stop at football. Mexico's national drill has a very specific purpose: keep the familiar alert tone so residents do not confuse a rehearsal with a real quake. Confusing a false signal with a true one, in an alert system, can kill people. Confusing false content with true content, in an information system, only kills trust. But both start in the same place: a signal read wrongly, and nobody checking it again.

The Machine's Reasonable Side

It would be unfair to stand on only one side. I have followed automated data pipelines long enough to know they have done work no reporter could.

Before them, a fifth-tier English match could finish without a single line in any paper. A women's league in a small country had no public table. An eighteen-year-old in the Balkans had no profile page for a scout to look up. Those blanks used to be a privilege of silence, and that silence served a very small group with an interest in it.

The machine fills those blanks. It puts lower-league football on the data map, puts women players into searchable lists, puts unwatched matches into a verifiable archive. In the lower leagues, people do not need glory; they need a roof when the storm hits, and most of that roof comes from data.

The problem lies elsewhere. Human error has an address. Pipeline error does not. When a reporter gets it wrong, an editor can name the person, and that person answers for it. When a labelling system gets it wrong, nobody is named. Nobody apologises. No newsroom issues a correction, because that data file was never treated as journalism — it was only an input.

In 2026, I spent a semester analysing a three hundred million pound rescue package for small clubs, including fan-owned AFC Wimbledon. I found six National League clubs excluded from the list purely because of administrative registration codes, losing a combined 1.4 million pounds. If I had only exposed it, the story would have ended there. Instead, I called the six clubs into an online meeting and drafted a joint petition to the Department for Culture. Four of the six got their money back within eight weeks.

I tell that story not to boast. I tell it because it shows the only real difference between a system and a newsroom: the capacity for collective accountability. A rescue package only truly exists when someone dares to ask: where is the money? A data pipeline is the same. It is only trustworthy when someone dares to ask: where did this file come from, who labelled it, and who checked it before it became a headline?

Who Checks the Data Before It Becomes a Headline?

In 2026, in Doha, I interviewed 2,154 workers building World Cup stadiums. Thirty-eight per cent of them were not receiving the twelve per cent wage premium their contracts required. The organisers pressured us and threatened to pull accreditation. I kept one rule: every testimony had to match a payslip, a bank line, a timestamp. A Bangladeshi worker named Rahim had three months of wages withheld. I recorded his account, then sat with him to reconcile each pay period against his bank statement, turning testimony into verifiable data.

Football does not end at minute ninety; it runs to the last line of the bank statement. And a bank statement only means something when someone stands behind it.

The machine that labelled the Mexican earthquake bulletin will not be fined. It will not lose credibility. It will keep running, and that junk file will sit in the archive, waiting for the day it is dug up and turned into an article with a tidy headline, plenty of figures, and not a single fact.

A contract exists only on paper, the money vanished long ago — I still use that line for inflated deals. With data it is even truer: a label exists only in the metadata, while the real content drifted away long before anyone read it.

Fans pay the bills, but they are usually the last to see the books. Readers are the same. They are the final link in a whole content chain, and the last to learn that the file they just read may never have been about football at all.

The question I leave behind is not for the machine, because the machine cannot read questions. It is for the people who apply labels, who approve pipelines, who sign off on processes. Someone tagged an earthquake bulletin as football. Someone let it through. And if none of them is ever named, next time that file will not be about an earthquake — it will be about your club.

An academy can produce talent, but it cannot produce honesty. Neither can a data pipeline. It can simulate every football template, but it cannot simulate a human being taking responsibility for what was just published.