When the Football Feed Receives the Wrong News: A Morning in the Newsroom

I have a habit of arriving at the newsroom earlier than anyone else. Rome in early October, mist clinging to the glass, and in the silence of a room with the lights still off, I read back the lines that will go on the page that day. That morning, an odd item appeared in our football feed: a report about the death of a Peruvian journalist, shot while broadcasting a campaign event live in Caraz, in the Áncash region. The classification tag clearly read "football."

I read it a second time, then a third. There was no club. No player. No coach, no competition, no tactical diagram, no transfer list. There was only a woman who had died, three people in preliminary detention, and an investigation that had just begun. Of the twenty-five data points sent through the system, not one mentioned a ball.

This was not the first time I had encountered something like this. But it was the first time I had seen it so plainly.

The labelling machine and its harmless keywords

The story behind it is very simple, and very troubling. Modern newsrooms, sports newsrooms included, run on content pipelines. Every day, thousands of items pour in from everywhere, and no one has enough hands to read them all. So we hand the machine the job of tagging: this item belongs to football, that one to politics, another to economics. The machine does this by counting keywords — having learned from millions of previous articles that words like "club," "transfer," "coach," and "qualifier" mean football.

But in the item that should never have belonged to us that morning, the machine encountered different words: "regional committee," "campaign," "movement," and a proper name that sounded as though it had been written for a pitch — "China Polo." None of those words carry a football meaning. Yet in the vector space the machine operates in, they sit close enough together to push the item through the wrong door.

I have spent years watching teams. I am used to reading a movement off the ball to guess where a player will go. And I recognised that the machine is like an untrained observer: it sees motion, but it does not understand the touch. It sees a crowded event, voices raised, someone addressing a crowd, an unfamiliar name — and it calls that sport.

The trouble is not in the item, but in the pipeline

What made me stop was not the report itself. That report is a serious piece of journalism, belonging to the security and press-freedom section, and it deserves to be read by people doing the right job. What made me stop was the place it passed through: a pipeline that received it, stamped "football" on it, and passed it to people like me — people who would sit down to analyse it in the language of the pitch, and who would fail, because there is no pitch here.

I called a colleague in the technical department. He told me this happens more often than I think. Not because the machine is poor, but because the machine is trained to be fast, while people are taught to be slow. Between those two speeds there is always a gap — and in that gap, an item about a person's death can be filed beside a transfer deal.

I thought of the Trigoria training ground in November 2026, when I filmed a three-minute clip of Nicolò Zaniolo and watched it reach half a million views within an hour. That day, I understood that speed had changed my profession. But only that morning did I understand that speed had also changed the way our newsroom knows what it is talking about.

What the machine cannot count

There is something anyone who has followed a team long enough knows: what decides a match rarely lies in the numbers. It lies in the silence between two passes — the moment a defender turns his head toward a teammate, and that glance says everything. No statistics table records that glance.

Likewise, an item about a journalist's death can be classified by a machine that counts words. But no machine can count what is lost when a woman who both practised journalism and stood for office is shot in the middle of a live broadcast, before the eyes of those watching. No algorithm can touch the fear of other journalists in the same land, who are asking themselves whether tonight will be their turn.

In the document the system sent me, one detail made me pause longest: that the victim, before her death, said she was being harassed and intimidated, and that she had received calls from three separate detention facilities warning of a possible attack. Three facilities. Three different places, saying the same thing. If that was recorded anywhere, it lies in prison call logs, in visitor records, in the very things a football classifier never looks at.

The truth lies where the machine turns away

I am writing these lines not to tell a story about a technical error. I am writing because I believe that error tells us something about ourselves. When we hand the machine the job of deciding what belongs to whom, we are quietly admitting that we no longer have the time to ask. And when we no longer have time to ask, we begin to receive wrongly not only the news, but the responsibility.

I am fifty-nine years old. I have commentated on football for radio and television since 2026, from Belgrade to Rome, from Olympic Games to athletics championships. I learned to hear a match through the noise of the stands and through the silence of empty stadiums during the pandemic. This profession taught me that the most important things always lie where the camera is not pointed.

The classifier does not look there. It looks at keywords, and it names things. But we, seated at the end of the pipeline, must be the ones who look a second time, and say: this does not belong here.

The next signal

There will come a day, within the next few weeks, when the investigation moves from preliminary detention to formal conclusions, and we will learn whether the person who ordered the crime has been found. That is the real milestone, not the footage circulating online. And on this side of the world, in newsrooms like mine, a question will also be answered: will we set aside enough time to ask where each item belongs, or will we keep letting the machine name things on our behalf?

I still arrive at the newsroom early. Mist still clings to the glass. But since that morning, when I read back the feed, I always read one beat slower — just enough to hear the breathing of a story that is not mine.