When Mexican health data gets tagged as football: a systems error in sports content

This month my assistant forwarded me a file tagged "football". I opened it expecting an xG table, a PPDA pair, or at the very least the name of a club. Instead I got seventeen information points on breast cancer mortality rates in Mexico, ranked by state, broken down by age cohort. No club. No player. No competition. Not a single match.

I sat still for about thirty seconds. At 42, I still write as if every match is the last one I will ever get to live through. This time there was no match to write. What chilled me was not the file's contents — the contents were decent, sourced, fully quantified. What chilled me was the label on top of it.

Someone, or something, had decided that a Mexican public-health bulletin was football.

I handed the file to three people on my team. All three came back with the same question: so what is this article about? Nobody asked out of curiosity. They asked because they had been assigned to write, and they needed to know what they were writing about. That was the moment I understood the problem was not in the file.

A file done very well, mislabelled in exactly one place

In fairness, the extraction was meticulous. Seventeen information points, each individually sourced, with data and author opinion kept cleanly apart. Two Mexican government statistical platforms were named explicitly: Plataforma Demográfica and Sistema Nacional de Vigilancia Epidemiológica. The mortality figures belong to 2026. Case counts run from 1 January to 12 October 2026. The state ranking runs from Chihuahua, Baja California Sur, Coahuila, Sonora, Jalisco and Colima down to Guerrero and Oaxaca at the bottom. The highest-risk age cohort is women aged 75 and over, followed by 60-64, then 55-59.

That is a good health bulletin. It is not a football bulletin.

When you feed it to a system built to answer nine questions about football — tactics, club finance, the transfer market, standings, regulations, the dressing room, risk, media narrative, industry value chains — the system will do exactly one thing: it will fill the blanks.

In the report I read, it did not fill them. It wrote "insufficient information, cannot assess" across all nine dimensions, then stopped and raised a red flag against the label itself. That is correct behaviour. But it was only correct because a human was sitting there reading it with the authority to say no. Now imagine the same file moving through a pipeline with no reader and only a deadline.

Four faults, and only one of them is a data fault

The first fault is the label. A label does not describe content; a label selects the analytical frame. A file tagged "football" gets weighed on football scales, and every empty cell has an incentive to be filled with football material. That is the mechanism that produces fabrication nobody intended. Nobody lies; the frame simply demands an answer, and the frame is not afraid of blanks.

Try imagining those nine dimensions filled in. The tactics section would discuss "structural instability" — when the only thing unstable here is a mortality curve by age. The finance section would discuss "wage-bill pressure" — when the real pressure is the cost of late detection. The dressing-room section would discuss "generational transition" — when the generation in question is women aged 75 and over. Every sentence grammatical, every term correct, every claim wrong.

The second fault is far subtler, and I like it because it sits close to my own trade. The file contains a geographic hierarchy: Chihuahua, Baja California Sur, Coahuila, Sonora, Jalisco and Colima at the top; Guerrero and Oaxaca at the bottom. To an ordinary reader, that is a ranking of mortality rates by state. To a pipeline that has been taught that Jalisco, Sonora and Chihuahua are names with a football ring to them, it looks exactly like a league table. Chihuahua sits top of the table not because they won the title; Chihuahua sits top because they have the highest mortality rate. A machine that cannot tell those two things apart will produce a "form analysis" that is formally plausible and substantively empty — precisely the kind of piece I once had to delete after forty minutes.

The third fault is dating. Mortality figures are from 2026. Case counts run to 12 October 2026. Those two markers do not reconcile in a competent statistical bulletin, unless the original was a rolling surveillance release with a mismatched publication date. I once wrote that Italy scored 9 goals in the group stage when the true figure was 7, and an Italian fan caught me within two hours. I laughed it off, then hired a fact-checker a week later. A wrong number in my own article is my problem. A wrong number inside a pipeline becomes the truth of ten thousand other articles.

The fourth fault is source tier. The bulletin's most valuable conclusion — "highest rate in a decade" — was initially attributed to an image caption. An image caption is the weakest citation tier of all citation tiers. Only by point three does Plataforma Demográfica underwrite it. Anyone who read point one and ran had already carried a headline further than the source could hold it.

Put the four faults together and what do you get? A system capable of producing fluent football commentary — with numbers, with rankings, with judgements about "pressure", "narrative cycles", "systemic risk" — all of it about something that does not exist. The report calls that the highest-level risk. I will call it in my own language: a perfect hot take with no match behind it.

And this is where I think my trade now stands, after twenty-six years watching this industry operate. The economics of sports content no longer pay for accuracy; they pay for volume. The old newsroom needed a person to sit and read and say "this isn't football". The new pipeline needs one data field with two words in it. And a data field does not know how to refuse.

Maybe I am reading the label wrong

Now the part where I argue with myself, because otherwise this is just a grumble.

There is one possibility I have not ruled out. Maybe the "football" label was once correct. Mexican football has a tradition of breast cancer awareness rounds in October — pink shirts, pink armbands, players standing beside patients before kick-off. If the original piece was about those campaigns, the health material is only context and the football material is the meat. But across the seventeen points I read, that meat does not appear in a single word. No club, no player, no campaign, no stadium.

So I have to choose between two explanations. One: the pipeline mislabelled the record. Two: somebody intended to write a football piece, the football part fell out before it reached me, and what survived was the bare context. The second is scarier than the first. If the part that fell out was the most important part, then an entire process ran without anyone noticing the main course had vanished.

I do not have enough evidence to choose. I say so rather than picking at random. Fans hate me because I am right; they come back to read because I dare to be wrong. But there is one kind of wrong I am not allowed to commit: the lazy kind, when there is a file sitting in front of me to check.

The label is the new editor

Summer 2026 taught me one thing: people remember the shock merchant more than they remember the contract. Nine years on, I have learned something far less glamorous: people will also remember the articles with no match behind them, simply because they appeared fast enough and read smoothly enough.

In my trade, the old editor was the person who read the piece and asked "is this real". The new editor is a data field containing two words: "football". That field knows nothing about this sport. It knows only that if the label says football, there must be a football answer.

I do not hate the pipeline. It lets me watch more matches, faster, and occasionally it spots a detail I missed. But a pipeline is good at finding and bad at judging. The judging still needs a person.

As for me, I still side with the shocks. But I want every shock of mine to have a real match behind it, a real player, a real 88th minute. Once the label has been applied, the only remaining job is to check what is behind it. If what is behind it is seventeen points about breast cancer in Mexico, the correct job is to peel the label off — not to keep writing.

And if anyone asks why I wrote about a file with no football in it, the answer is simple: because this was the first time in years I could see clearly that the frame was conning me.