A Mexico City Pothole Labelled 'Football': The Crack Inside the Sports News Pipeline

A man in eastern Mexico City pours a bucket of hot asphalt into the road outside his home. Nobody hired him. Nobody pays him. He does it because that pothole has been there through three rainy seasons, and because his neighbour's wheels hit it once more every single morning.

The story sits inside a very specific legal frame: the Civic Culture Law of Mexico City, Article 27(IV) and Article 28(II). No provision forbids filling a pothole. What can be sanctioned is obstructing the use of public-domain goods, or blocking a street or public space without a permit and without justified cause. The fine amount is not stated in the text I read.

Three hours later, in São Paulo, I open a file on my desk. The file carries a domain label: football.

There is no team inside. No player, no scoreline, no tactical shape, no injury, no transfer. Only a pothole, a resident repairing a road, and two clauses of a local law in a country eleven time zones away.

I sat still for a long while. After thirty-five years in this trade, I am used to being called a provocateur. This was the first time I saw a news pipeline call a pothole football, and do so without a flicker of doubt.

Context: a machine designed never to return an empty result

The pipeline runs on two levels.

The first level reads a source text and assigns it a domain label: football, basketball, tennis, politics, lifestyle. The second level takes that label and opens a deep-analysis framework. For football, that means nine dimensions: tactics, club finance, results and opinion cycles, league landscape, rules and governance, dressing room, risk, media narrative, and industry transmission.

At the second level, there is a release valve. The rule is explicit: if a dimension lacks sufficient information, state plainly that it cannot be assessed. Do not fabricate. Do not speculate without grounding.

And here is where I stopped, underlined, and circled:

The system has a release valve at the analysis layer, but none at the labelling layer.

In other words: a machine carefully taught never to invent numbers was never taught not to mislabel. It can say eighteen times that there is insufficient information on tactics, on transfers, on the dressing room — and still keep the football label bolted to the top of the document.

I read that analysis from start to finish. Eighteen information points. All of them about asphalt, roadways, and a civic-culture statute. Not one line mentions football. Yet most of the document's length is spent explaining, over and over, why xG, PPDA, FFP and PSR do not apply.

Think about that the way someone who has worked thirty-five years would. I once sat in a press room in Russia in 2026, asked Didier Deschamps about the psychological pressure on a nineteen-year-old Kylian Mbappé, and had a male colleague laugh in my face: "Another girl's emotional question." I answered with one sentence: "Emotion decides performance." That night I wrote that France would win. On 15 July 2026, France beat Croatia 4-2, Mbappé scored, and the press room went silent.

People laughed at me for asking Mbappé about feelings; then a whole country wept with happiness.

I tell that story not to boast. I tell it because I know what it feels like to be judged wrongly by a system larger than yourself. In 2026, when PSG triggered Neymar's €222 million release clause, I wrote that the club was buying a brand, not discipline. I was called a woman who only chases drama. I did not retract. I attached the table of PSG's midfield touches in the previous Champions League season. In March 2026, PSG lost 5-2 on aggregate to Real Madrid, and Brazilian football forums went back to read that old line of mine.

That transfer was a purchase; and it was also a fracture running through an entire football economy.

This time is different. This time the thing judging me wrongly is not a colleague. It is a machine with no opinion at all.

Core: five mechanisms that let a machine label a pothole as football

I have no access to the pipeline's source code. But I have thirty-five years of reading copy, and I have the analysis in my hand. That is enough to reconstruct the mechanism.

Mechanism one: keyword distance is too short

In English, "pitch" means a playing surface, a tone, and a throw. "Fill" is what you do to a hole and what you do with filler. A classifier built on token proximity will see football where there is only asphalt.

This is the crudest error and the most common. On its own it is not frightening. It becomes frightening only when nobody catches it.

Mechanism two: the economics of volume

An automated newsroom needs a stable output every day. When inbound supply falls, the system has two options: return nothing, or widen the label. An empty result is recorded as a failure. A mislabelled document still counts as a completed unit of work.

No operational metric penalises a wrong label. That is the whole story.

Mechanism three: the pull of the head-term label

"Football" is among the highest-traffic labels in all of sports content. A system optimising for reach will always lean toward the head term when the signal is ambiguous. Being wrong in the direction of traffic is forgiven; being wrong in the direction of no traffic is recorded as waste.

Mechanism four: the framework manufactures its own subject

This is the part I want you to notice most, and it is the most valuable.

When an analysis framework has nine fixed football dimensions and is allowed to run on any document, the framework itself creates the illusion that the subject is football. You open with "Tactical and technical analysis." You write a heading called "League landscape and team positioning." You close with "Football industry transmission." Even if the body says "insufficient information" twenty times, the exterior screams football.

A framework large enough can turn emptiness into a section of the newspaper.

I have seen this in broadcasting. Hand a presenter three arrows on a whiteboard and one data table, and he will narrate a match that never took place.

Mechanism five: nobody owns the label

At the analysis layer there are rules, process, accountability. At the labelling layer there is none. No signature. No challenge. And because the label is the first thing produced, it is the last thing examined.

In 2026, I learned to listen to football with my heart when the stadiums had no voices left.

I lost my television commentary contract, started a livestream series inviting forgotten people to speak — assistant coaches, unemployed ex-players, and João, a Maracanã cleaner of twenty years who lost his job overnight when distancing began. His episode reached 2.1 million views and raised more than R$340,000 for an emergency fund.

João appears in no statistical table. And if an automated pipeline read the piece about him, it might label it "public services" — far more accurate than labelling a pothole "football."

The irony is that the machine mislabelled precisely the subject I care about most.

Contrarian: what if the machine is right?

At this point I have to argue against myself, or this is just a three-thousand-word complaint.

Hypothesis one. Perhaps the machine read the future correctly, and I am clinging to an old definition of football.

Consider it. Professional football lives inside public infrastructure. The pitch is public property. Stadiums often belong to a city or a state. Pitch renovation budgets come from municipal coffers. In Vietnam, the stories I have followed for years — surfaces degrading after the rainy season, grass being cut again mid-season, clubs waiting for provincial money to repair stands — carry the exact shape of a pothole story.

A pothole in Mexico City and a crack in the stand of a top-division stadium ask the same question: who is responsible for the thing everybody shares?

Hypothesis two, and this one is harder to swallow. Perhaps my reaction is a form of turf defence.

I spent thirty-five years earning recognition in a male-dominated industry. I was once asked, in that tone — "what does a woman know about tactics" — and I know what it feels like to have someone else define where you belong. And now here I am, about to say a pothole does not belong in my space.

A woman said a small thing; people laughed. Five years later, they repeated it.

If I once demanded the right to speak about what others considered irrelevant to me, I cannot turn around and close the door on someone else.

Hypothesis three. Perhaps the real problem lies elsewhere entirely: nobody actually believes the label. A label is just a label. The harm comes from the behaviour behind it — the enormous volume of content generated to fill space, and the shrinking number of readers left to read it.

So here is where I could be wrong.

First, I may be exaggerating the severity. A wrong label, in the end, does not ruin a match.

Second, I may be wrong about the cause. Perhaps the root is not the classifier but the output pressure placed on the people running it, who were never consulted before the quota went up.

Third, and this is what I weigh most: I am using a technical incident inside a system I do not own to talk about something bigger. Sometimes people do that. Sometimes they are only convincing themselves.

Pressure and expectation: the section I always keep separate

Since 2026, every analysis I write has its own section, because I believe supporters need to understand the heart before debating the shape.

The pressure on a sports news pipeline today resembles the pressure on a young player pushed into the first team too early. He can run, shoot, score. But nobody taught him to recognise when he is standing in the wrong position, because nobody gave him time to be wrong and correct it.

A content system with no release valve at the labelling layer is exactly like a young midfielder who was never coached in proactive defending. He will shine for thirty minutes, and the space behind him will be exploited for the rest of the match.

For Vietnamese football, that pressure is more concrete. On 5 January 2026, in the second leg of the ASEAN Cup final in Thailand, Nguyễn Xuân Son broke his leg and had to leave the pitch in a match Vietnam won 3-2, taking the title 5-3 on aggregate. I have written many times that fixture density is the single biggest cause of injury, and that no medical department can save a squad playing two matches a week for months. What I have not written enough is this: at the same moment, a player also absorbs pressure from a content machine pumping out hundreds of articles a day about him, most of them verified by no one.

Collective expectation does not kill players. Unfiltered collective expectation does.

Proposal: three concrete steps, with a timeline

I do not write to vent. Since 2026, when a piece on Brazil's women's team losing the Tokyo Olympic final led the Brazilian federation to invite me into an advisory group and push women's media budget up 47%, I have held one rule: every time I speak out, I must attach something doable.

Step one: add a release valve at the labelling layer. Any document assigned a domain label must pass a minimum cross-check — if no entity belonging to that domain appears (team, player, competition, governing body), the label drops to "unverified" and waits for human review. Timeline: one update cycle, no rebuild required.

Step two: separate the volume metric from the label-quality metric. As long as a mislabelled document still counts as complete, the error rate will not fall. Timeline: one quarter, with a comparison report.

Step three, and the one I actually care about: publish the label. If every article displays its data source, publication date and verification status, a wrong label exposes itself before it spreads. Every sports newsroom can do this in the current regular season, because our readers are already long accustomed to checking things themselves.

What I predict, and what I want you to verify

I always end with a testable judgement, because an untestable judgement is just emotion in makeup.

I predict that over the remainder of the 2026-2026 season, at least one club in a Southeast Asian top division will publish detailed data on its pitch quality and its stadium infrastructure costs — and that data will be labelled local politics by an automated pipeline before it is labelled correctly.

If that happens, you will understand why a pothole in Mexico City matters to me.

Finally, a question for those running these pipelines: if your machine cannot say "I do not know what field this is," what makes you trust that it knows what is true?

I am fifty-one, I still sit at the same corner table in São Paulo, and I still believe the most valuable thing in this trade is not speaking louder than everyone else. It is knowing exactly what you are talking about.