When a Data System Files a Rom-Com Under Football

THE OPENING "Have a look — the system has filed this one under football." The line came from the youngest writer in our small newsroom in Nha Trang, on a weekend morning when the sea outside was still grey. She turned her screen toward me. On it was a story about a new Netflix romantic comedy, starring Lindsay Lohan and Henry Golding, directed by Mark Waters. The film is called Return to You. I read it once. No clubs. I read it twice. No players, no scoreline, no line-up, not a single sentence about transfers. I read it a third time, more slowly, to be sure I had missed nothing. Still nothing. The nineteen information points described actors, a director, a producer, the films Lindsay Lohan has appeared in, her previous Netflix collaborations. Not one word belonged to football. Yet in the top-left corner of the screen, the content label read, plainly: FOOTBALL. She turned to me, half joking, half worried: "The system must have made a mistake." I nodded. Then I asked her a question I have been thinking about ever since: "If you hadn't noticed today, how long would it have taken someone else to find it?" A WRONG LABEL INSIDE THE DATA INFRASTRUCTURE To answer that, I have to go back a few months, to when our newsroom adopted an automatic labelling system. Every incoming item passes through a pipeline: entity extraction, category labelling, competition labelling, player labelling, and then into a shared database. Football databases in Vietnam — VuaBong.vn among them — have gradually become infrastructure rather than tools. A story without a label almost does not exist. That sounds reasonable. Football is a sport of numbers: minutes, passes, kilometres covered, transfer fees. A system that labels well helps reporters retrieve data faster, helps readers search more easily, helps a newsroom measure what people care about. I support that, and I think anyone who has worked long enough must support it. But the story about Return to You went through that pipeline and came out with a wrong label. The immediate cause is hard to pin down: perhaps the system learned from earlier articles where Netflix appeared beside sports documentaries, where an actor's name matched a player's name in some league, where a keyword was split at the wrong boundary. I do not have enough data to assert a specific mechanism, and I will not guess. What I do know is the analysis that followed. When the article was run through the very framework we use for matches — tactics and technique, club finance and the transfer market, results and the opinion cycle, league landscape and team positioning, rules and governance, management and the dressing room, risk profile, media narrative and expectations, and industry transmission — all nine sections returned the same phrase: insufficient information. The central conclusion fits in one line: the football category label is wrong, and the article should be re-labelled Entertainment/Film before being removed from football databases. That recommendation is operationally correct. But if the story stops there, we have skipped the most expensive part. LABELS ARE CHEAP, MEANING IS EXPENSIVE I entered the profession in 2026, when the paper I contributed to had just been founded and everything was still done on paper. Back then, no piece reached the page without passing through the hands of a senior editor who read to the last line before picking up a pen. That person did not label. That person was responsible. What we have just lost is not a task, but a job title. A label is generated in a few milliseconds. Meaning requires a human being to read to the end, cross-check against what they know, and then ask: is this true, is this necessary, could someone misread it? The label answers where this belongs. Meaning answers what this says. The two questions sit so close together that we mistake them for one. In football we are making exactly the same error. A post-match report stating that the away side had 65 percent possession reads very professionally. But it is only a label. It does not say where they had the ball, in which zones, to what end, or why they still lost. Labels are cheap. Meaning is expensive. Our job is to buy the expensive thing, not to sell the cheap one. A LABEL CALLED xG The clearest symptom of this disease in modern football is xG. Based on my experience of watching matches over more than thirty years, I am not against data. I am against using data as a verdict. xG measures the quality of a shot inside a model built on historical data. It answers the question: in average conditions, what is the probability that this shot ends in a goal? It does not answer why that player was in that exact position, why that defender dropped half a step, why the goalkeeper waited another three tenths of a second before diving. I was at the Kazan Arena on 30 June 2026, when France beat Argentina 4-3 and a nineteen-year-old Kylian Mbappé scored twice. Nobody in that stadium remembers his expected goals that night. We remember a long sprint, the feeling that Argentina's back line could only stand and watch, the way a young player turned half the pitch into his own private space. That day a male colleague smiled and asked what women know about tactics. I did not answer. I wrote. I wrote that when Mbappé ran through the defence, I saw youth rewriting history with speed. Had I offered only a label that night — fast player, two goals — I would have done the job of a machine and failed at the job of a writer. The lesson is not whether xG is right or wrong. xG is right, within its scope. The lesson is that we use an index describing the past to draw conclusions about the present, and then call it analysis. That is precisely what the labelling system did with the Return to You story. It recognised familiar signals, attached the FOOTBALL tag, and felt entirely confident. Being confidently wrong is the most dangerous kind of wrong in this trade. THE TRANSFER MARKET AND THE GAPS BEHIND THE NUMBERS On my desk sits a file on the transfer market, something I have followed for years. There, another version of the same disease appears. A transfer is announced with a figure, and that figure immediately becomes the whole story. People argue over whether the player is worth that much, while the player is living the hardest days of a career: packing up a house, saying goodbye to old team-mates, explaining to a child why the school is about to change. I have written before, and still hold the view, that signing-on fees for free agents are a more harmful form of spending than transfer fees, because they sit outside the core scrutiny of financial regulations. When a player's contract expires and he joins a new club, most of the cost does not appear as a transfer fee but as signing-on payments, agent commissions, and milestone installments. The figures published outside therefore look better than the reality. The label is clean; the meaning is not. This connects directly to this morning's story. When a system looks only at labels, it will read every entertainment item as football and every transfer as an arithmetic exercise. When people look only at labels, we lose the ability to see that a transfer is not a set of numbers, but a parting trying to find its own goodbye. MEMORY IS NOT IN THE DATA TABLE I live in Nha Trang, and my work takes me through many stadiums. Vietnamese football in recent years has been measured more than ever: more cameras, more indices, more spreadsheets. That is progress, and I do not want to go back to the era of a single camera angle. But ask a supporter at My Dinh what remains from the night of 15 December 2026. They will not tell you about possession statistics. They will tell you about Nguyen Anh Duc's early goal, about an entire stand bursting open, about a city pouring into the streets after the final whistle, about the first AFF Cup in ten years of waiting. In 2026, when the pandemic stopped every competition, I came back to Nha Trang and stood in the empty 19 August Stadium. The sound of a ball rolling on grass was audible beat by beat. I called the groundsman who had worked there for forty years, and he told me that for the first time in his life he heard birds on the stands. That empty stadium taught me that football is the voice of people. No model predicted those birds. No database stored them. And no label can hold them. On 12 June 2026, at Euro 2026, I watched Christian Eriksen collapse in the Denmark–Finland match. The players formed a circle around him, the crowd sang his name in the dark, and when he woke and waved, I cried in a virtual press room. Eriksen fell, and the whole world realised that football is just one way for us to hold each other. A veteran editor called me the next day and said he had been wrong to think women do not understand football. I was not angry. I only thought: if he read the match data, he would see a postponed game. You have to sit to the last line, you have to stay in the stadium after every reporter has left, to understand what actually happened. THE COUNTER-INTUITIVE VIEW: THE FAULT IS OURS The machine was wrong. But it was honestly wrong. It labels based on what it has seen appear together. Netflix appears with sports documentaries, a proper name appears with a club, a keyword appears with a category. That way of learning is crude, but it does not pretend. It never claims to understand football. We are the ones pretending. We built pipelines that count articles, count labels, count quick-answer blocks for search engines. We taught a twenty-four-year-old writer that a labelled piece is a finished piece. We praise speed and call care slow. Under those conditions, a romantic comedy slipping into the football section is an inevitable consequence, arriving a few months earlier than scheduled. The frightening part is not that the system mislabelled something. The frightening part is that nobody in the newsroom found it strange. That morning, the person who spotted it was the youngest writer in the room. She is new, she has not yet acquired the habit of trusting labels, so she still reads content. The paradox lives there: the more structure we build, the more we need people willing to ignore structure. The more indices we have, the more we need people who will sit to the last line. WHAT I WANT TO KEEP A wrong label can be fixed with one click. The Return to You article was moved to the correct category, and the football database was clean again before any downstream analysis could be built on a false foundation. But a newsroom that loses the habit of reading to the last line needs years to get it back. And in that time, mistakes like this morning's will stop being exceptions; they will become how we work. I am not proposing that we switch the system off. I am proposing that we keep one person at the gate with a single task: read to the end. No labelling, no summarising, no optimising. Just read, and then ask what is actually being said. If one day a machine labels a Vietnamese match as a romantic film, I hope it is because the match was beautiful enough to deserve it. Until then, I still want someone sitting there after every reporter has gone home, listening to the grass, remembering that this trade begins with a human being.