When a Horror Film Wore a Football Label

Last week a file landed on my desk. Its label read "football." I opened it and found not a single letter of football inside. It was a piece about Jessica Chastain's horror film "Other Mommy," the screen adaptation of Josh Malerman's novel "Incidents Around the House." Casting changes, five major alterations to the screenplay, the architecture of a horror narrative — the names of Karen Allen, Arian Moayed and Arabella Olivia Clark turning across the page. No club, no player, not one match. Yet the label stuck to the file announced in silence: this is football.

Across more than four decades of this sixty-year life, I have written football from dugouts, press boxes and the tea stalls of cities. When I left civil engineering for journalism in 2026, I did not know that one day I would have to judge whether a film file was football. In 2026, sitting in the dugout for Sylhet Sixers, I live-tweeted the batting of seventeen-year-old Afif Hossain. That day the team lost by seven runs, but Afif's innings trended in Sylhet. Voice notes sent from the tea stalls, one hundred and eighty comments — together they became a two-thousand-word diary. The lesson of that night was simple: new media is not a threat, it is a bridge. But another lesson joined it — put the wrong name on information and the bridge lands in the wrong place.

The question is, how did that wrong label arrive? An automated data pipeline runs in two stages. The first stage gathers and classifies information — which piece is football, which is entertainment, which is politics. The second stage takes that information deeper, analysing tactics, finance, public opinion. This file from Sylhet crossed the first stage carrying a false-positive signal. The football classifier mislabelled a horror-film article as football. The cause was probably one of two — the machine saw words like "change," "structure," "narrative" and decided too quickly, or the label field was filled in wrongly by someone.

I believe this is not an isolated accident; it is the picture of a structural weakness. When news is assembled by machines, the greatest risk lives in the machine's confidence — it does not know that it does not know. A wrong label does not stay alone; it spreads. If that file travels only into a football-analysis downstream, it enters the corpus that trains models and deposits false noise. And this noise is the most dangerous, because it is invisible. A reader who sees a wrong headline grows suspicious; but a reader who sees an analysis standing on a wrong corpus does not grow suspicious.

The deep-analysis report checked that file across nine dimensions — tactics and technique, club finance and transfer market, results and public-opinion cycle, league landscape, rules and governance, management and dressing-room, risk profile, media narrative, and industry transmission. In every dimension the answer came back in one sentence: "insufficient information, cannot assess." In tactical analysis there is no xG, no PPDA, no possession statistic. In finance there is no club, no balance sheet, no wage bill. In the transfer market there is no deal, no agent. In the dressing-room there is no coach, no leadership.

Here a lesson hides that is precious to me as a football writer. Saying "insufficient information, cannot assess" is a strength, not a weakness. When a machine guesses, it manufactures fiction. Football history has no shortage of fake transfer stories, where a single unsourced rumour becomes a thousand comments. I have seen for myself how one false rumour can capture a whole city's week of conversation. But here the pipeline admitted its limit. It said: I could not verify, so I will not lie.

Still the problem remains. The risk matrix flagged the biggest risk as "data contamination," or corpus contamination — that is, when wrongly labelled information slips in, the quality of the whole corpus drops. Three warnings emerged. Domain misclassification first. Downstream contamination second. The risk of fabricated analysis third. These three are woven from one thread. And the solutions are also three. Quarantine the file, correct the label, and install a "domain-consistency validation gate" before the second stage. The simplest rule is this — if at least one club, player or competition entity cannot be found, the item is rejected automatically.

My experience says this rule is like the rule on a football pitch. In a match, VAR decides by looking at the right footage, not a video from the wrong angle. In 2026 I did not go to Russia. I watched every match with three hundred supporters in a community hall in Sylhet. When Japan beat Colombia, the hall burst; on the night Belgium led 2-0 and lost 3-2, the hall collapsed together. In one night I collected forty-seven voice memos. My editor wanted tactical analysis; readers wanted to know why their neighbour wept. The lesson of that night was simple: human emotion can catch a machine's error, but a machine does not understand human emotion.

The fan zone was not a crowd; it was a heartbeat with a thousand faces. If that heartbeat is driven by wrongly labelled information, we are cheating the supporters. In Sylhet, the teenager did not just score; he taught a city how to hope again. It is because I remember that teenager's innings that I know how true the emotion of football and cricket is. And the foundation of that truth can never rest on a wrong label.

Here an inverted truth hides. The easy reaction is to say automation failed, the machine is useless. But the real story is subtler. This pipeline was in fact honest; it admitted its ignorance, it did not fill the gap with guesses. The danger is not in the machine, the danger is in the trust we place blindly in the machine. A new bubble is boiling in the football-journalism market — where speed and automation are valuable, and verification is a lagging luxury. Someone wants to make a thousand headlines a second, but no one sits down to catch a single wrong label.

This error reminds me of another reality. In sports media the word "blockchain" is now a fashion — an immutable chain of information, verifiability, transparency. But the real chain is built not in paper and pen, but in human eyes. If the power to put a label on information lies only in the machine's hands, that chain breaks at the first false-positive signal. To me, the immutability of information is not a picture in a database; it means a witness who can say — this file was not football, who claims otherwise? I have followed teams through airports, but the loudest arrivals happen in living rooms. It is in that living room that a label's truth is verified, a lie caught.

One aspect of the report is especially important. It says this file is not merely an error; it is a clean "negative test case" that serves to validate the classification layer. That is, failure can be turned into learning. Three signals were named for watching. The false-positive rate — if more than one to two per cent of non-football items get a football label, the corpus integrity is at risk. Entity-type checks — at least one club, player or competition must be present. And source-beat alignment — does the news section (sport versus entertainment) match the label?

The source deserves a look too. This piece came from the entertainment section of The Express Tribune, a general-interest newspaper, which is not a football-authority source. If a football analysis is born from an entertainment-section source, that is the first warning sign. And here my values come into play. In the transfer market I regard spending huge sums on a young player — one with fewer than fifty top-flight games — as naked gambling. In the same way, placing huge trust in automated information without verification is a kind of gamble.

The tea stalls of Sylhet still host the argument — who is best, who will lose, who will win. If the foundation of that argument is wrongly labelled information, we are handing supporters a lie. At sixty I still pack a notebook for the match, because memory needs a witness. When the new league begins next month, do not run any pipeline without a verification gate. The question is simple: will a machine decide whether a file is football — or the journalist who sits in the dugout, cap off, tea cup in hand, waiting?