When a Sports Database Labels a Crime Report as Football

In the archive I have built over more than twenty years as a journalist, there sits an item tagged "football" that contains not a single player's name. No club. No competition. Not one expected-goals figure. Only the name of a Mexican mayor who was assassinated, the number 32 detainees, and a string of security agencies: the State Prosecutor's Office, SSPC, FGR, the Defense Ministry, the Navy, the National Guard. A young editor sent me the link with one line: "Nam, why did the system file this under football?" I opened it, read it to the end, and sat still. Not because the content shocked me, but because of the way it walked into my house.

Skim it, and you might nod and move on. A security report about the assassination of the mayor of Uruapan, in the state of Michoacán, should sit in the current-affairs section. Instead it sits here, among my tactical breakdowns. The question worth asking is not "why was there an error," but "how many gates did this error pass through before it reached me."

The report says that Carlos Manzo, mayor of Uruapan, was assassinated on November 1, 2026, while in office. Mexican media covered the case widely. According to the report I read, police had detained the 32nd person in the investigation, a suspect referred to by the identity-protecting name Alexis "N". Omar García Harfuch, Secretary of Security and Citizen Protection, stated officially that the agency had identified the operating structure of a criminal group with a presence in Uruapan. Another suspect, Jorge Armando "N", is accused of carrying out the attack directly.

Nothing in any of that belongs to football. So why was it tagged as football?

I sat down, opened the classification log, and found the culprit. The system has no intelligence. It has keywords. The report contained words like "attack," "operation," and "structure" — the same words my football archive uses to describe wing attacks, transfer operations, and squad structures. A keyword filter met "Uruapan," "assassinated," "attack," and decided this was an article about football. It is a naive error, but it exposes the real problem: our archive knows how to match words, not how to understand meaning.

At 69, I do not believe in the spectacular fall; I believe in the quiet crack from the previous season. Misclassification is the same. It did not appear on the day it made me sit still. It was built over years, each time we accepted a convenient shortcut.

When I built my personal database for each player, I learned one principle: an event is only trustworthy when it has a timestamp and someone accountable. For a news report, the same holds. The Uruapan report has a timestamp, but most of its information comes from unnamed sources. In my notes, I counted more than ten fields marked "source: unspecified," with only SSPC and Secretary García Harfuch named. The photo carried a single credit, @SSPC. A report with more than half its details anonymous should not be treated by any system as verified data, no matter which section it belongs to.

Here is where I differ from the crowd. People will say: it is the algorithm's fault. I disagree. A machine only scans keywords the way people scan a scoreboard; the problem lies at the gate above it. A sound system must have a checkpoint: at least one football entity — a player, a club, a league, a stadium — must be present before a football tag is applied. Without one, it must not be allowed in. That gate does not exist, or it exists wide open. And when the gate is open, everything walks through, including an assassination.

A single error is not frightening. Its spread is. Each time a mislabeled report passes unchecked, the wrong label is written into history. Next time, the system learns from that history and grows more confident that football includes reports like this. I call it source contamination. It is silent, like a crack in a wall — until the whole wall falls, and no one remembers when they first ignored it.

This reminds me of an old case of my own. In the 2026-18 season, when Mike D'Antoni's Houston Rockets fired three-pointers almost without limit, editors pushed me to write a piece celebrating the "air revolution." That, too, was fast labeling. The label "revolution" was applied first, the data found afterward. I refused, sat for three weeks, filtered 1,200 games, and found that teams taking more than 40 threes a game won only 62 percent of the time — barely different from teams taking 28 to 35. Had I labeled by feel, I would have let bad data into the archive forever.

A sports archive lives on accuracy. An article filed under the wrong section kills no one. But an archive that mislabels again and again will rot the very foundation the crowd leans on.

There is one subtle detail I want to say slowly. In that report, the two people named are not players: García Harfuch is a security secretary, and Alexis "N" and Jorge Armando "N" are suspects. None of them has a player file in my archive. If my system were built properly, it would never pen them alongside the people I store by season, by usage rate, by minutes played. This is not about some noble distinction. It is about data order.

I think again of Bojan Bogdanović. In September 2026, at the European qualifiers of the basketball World Cup in China, I flew to Tel Aviv to watch Croatia face the host team. Bojan shot 3 of 14; Croatia lost 78-88. Young reporters in the press area immediately blamed fitness. Once again, the label was applied first, the data found afterward. I spent ten days, reviewed all 47 Croatian attacking possessions, and found their pick-and-roll system had gone stale, was read 19 times by Italy, and that Bojan received the ball 8 meters from the rim instead of the 6.5 meters he was used to in the NBA.

The collapse of a star and the collapse of an archive share one mechanism. The most frightening thing about Bojan's collapse is that it was so silent we grew used to it. The most frightening thing about an archive infected with wrong labels is the same: it raises no alarm, it simply sits there, and we slowly treat it as normal.

The 2026 transfer window brought me a related lesson. In January 2026, when Ben Simmons was traded from the Philadelphia 76ers to the Brooklyn Nets, most of the press chased the noise and called it "the deal that saves the Nets' future." I did not rush. I reopened the file I had kept since 2026, when Simmons refused to shoot threes throughout the playoffs, when his usage rate dropped 12 percent entering the fourth quarter. His defensive numbers were only good when his team led by ten or more. I wrote a warning built on three metrics that contradicted the media story. The result: Simmons played 42 games, looked pale, and the Nets exited the playoffs in the first round.

What I mean is this. Numbers do not lie. The way we grip them in our hands is what lies. When we grip a security report into the football section simply because it contains the word "attack," we are gripping it wrong. And we grip it wrong, usually, because we want to be fast.

Somewhere, a gate was skipped. That is the real problem.

Here, someone will push back: a wrong label is just a wrong label, what is the big deal? I think that calm is the problem. In my trade, an article filed under the wrong section costs no one money. But the habit of accepting small errors is the habit that erodes every system. If we ignore a wrong label today, we will ignore a wrong statistic tomorrow, and by next season we will lean on an archive whose correctness we no longer know.

I held the numbers of an assassination case under every light, and realized the problem was not the report. It was the gate that let the report through. Fix only the label and we have treated the symptom. We need a harder question: how many other articles passed through the same gate, and how long has our system silently counted them?

I never stop writing about the small cracks, because we stopped asking why long ago. A misclassification is not news. But it is evidence.

I will not remove the report from the archive. I will keep it, place it in a separate drawer, tag it correctly, and note the day it walked in. In my trade, memory is a legal code, not a nostalgic warehouse. I keep it so that next time, when someone tries to fling the gate open, they must see the first crack.

The question for all of us is not where the machine went wrong. It is when we stopped asking the question.