Wrong Labels: When the Football News Pipeline Tags the Truth Incorrectly

Da Nang, 22:47 on March 14, 2026. Hoa Xuan Stadium. The first half of a V.League match had just ended at 0-0, and Stand B was making the kind of noise Vietnamese stands always make when nobody has been given a goal yet.

I was sitting in row eleven, in the press area, pen still in hand. I still keep a notebook even though a phone can do everything — a habit I carried over from long ago, from the days when I was still an athlete and learned that your own memory is the one material nobody can buy. I keep a notebook, and it does not record goals. It records other things.

The phone buzzed. A push notification from a sports news app I had installed two years earlier, with a red label in the corner: FOOTBALL.

I opened it. There was not a single word about football inside. No club name. No scoreline. No player. No transfer. No coach. It was a report on an entirely different subject, filed under the football section, and pushed to me — a person who had just opened the app to follow a match happening ten meters away.

The label lied. Not the person behind it. The label.

Twenty minutes later, I wrote a line in my notebook: Wrong label — 14/3/2026 — 22:47. I did not yet know where that line would lead me. But I knew one thing: if the label lies in a push notification at midnight, where else does it lie — in the transfer summary, in the data stream flowing to betting companies, in the match report nobody is allowed to read?

---

To answer that, I have to start where everything starts: speed.

In 2026, the football information industry runs at a rhythm that even the people inside it no longer control. A V.League match ends. Within thirty minutes, there are hundreds of articles. Within three hours, thousands. By the next morning, that number multiplies exponentially, and nobody can count it, because most of the new content is no longer written by human beings — it is aggregated, condensed, copied, machine-translated, and republished by a chain of anonymous tools.

This is the crux few notice: the football news industry is no longer a writing profession. It is a pipeline.

That pipeline has joints. The first joint is the reporter at the stadium — the only person actually present. The second is the newsroom, where it is edited. The third is the aggregator sites, where content is collected by keyword. The fourth is the app, where content is labeled and pushed to readers. And the fifth joint — the one nobody discusses — is the data buyers, where the soulless numbers from the fourth joint turn into money.

At each joint, a person or an algorithm decides what this content belongs to. At the fourth joint, that decision is called labeling. And at the fourth joint, the label itself is what gets sold.

The transfer window is when this pipeline runs at full capacity. During the transfer window, every rumor is a unit of goods. Every name mentioned is a unit of traffic. Every number — true or false — is bait. And when goods move at that speed, the label no longer describes the content; the label describes the demand. People stick the football label on something that is not football, because the football label sells better than the correct one.

That is why I decided to turn a note written at 22:47 into an investigation. Not because of one wrong push notification. But because I wanted to know: if the pipeline mislabels at the input, how trustworthy is its output — the numbers people use to bet, to judge players, to condemn a referee?

---

I began by measuring. Over thirty days, from April 1 to April 30, 2026, I tracked push notifications from the seven most popular sports news apps in Vietnam. I did not read the content. I only recorded two things: the label the app attached to the notification, and whether the content inside actually belonged to that label. I discarded ads. I discarded duplicates. I kept 1,284 notifications.

Result: 312 notifications carried labels that did not match their content. A rate of 24.3 percent.

One quarter. One in every four notifications I received in a month told me a lie about what it was.

That figure is not shocking if you have never looked at the mechanism. It is shocking once you know how the mechanism is designed. I reached two content-operations engineers at two of those seven apps — one in Hanoi, one in Ho Chi Minh City. Both asked me not to name them, and both said the same thing in two different ways: the system does not read content. The system reads keywords.

An article is labeled not because someone read it, but because it contains words on a list. There is a list for football. There is a list for transfers. There is a list for injuries. An article about a completely different subject only needs to contain a few words from the list — and in Vietnamese, the overlap of vocabulary between fields is enormous — and it will be dragged into exactly the category it does not belong to.

The engineer in Ho Chi Minh City told me a sentence I copied word for word into my notebook: We don't have a domain check. We have a speed check, a traffic check, no right-or-wrong check.

A pipeline with no domain-validation gate will never discover that it is carrying the wrong goods. It only knows the goods passed through, the goods were counted, the goods generated traffic. And when a wrong label is born at the input, it does not die at the input. It travels on. It gets re-aggregated by another site, quoted by another app, read by another person — and it reaches the people with money, the people who do not double-check.

But the wrong label on an app is only the outer layer. It is like a player's shirt: visible, easy to fix, low in consequence. The inner layer is where things become dangerous.

I returned to a file I have pursued for years: transfers with two layers. There are contracts signed on the pitch, and there are contracts signed in the dark.

Some years ago, I published a specific case. A Vietnamese player moved from a domestic club to a Gulf club. The contract was announced at a fee of 500,000 US dollars. That figure was reposted, quoted, commented on, and became truth within twenty-four hours. But when I traced the money — every transaction code, every date, every intermediary account — I found a different figure. Two hundred thousand US dollars left the system and went to an account in the Cayman Islands, under the name of an individual who was never registered as the player's representative.

That is two layers. The upper layer is the contract for the public. The lower layer is the contract for those who actually split the money.

And here is the point I want you to hold onto: both the upper and lower layers get labeled as a transfer. But only one of the two layers contains the truth. When a news pipeline does not read content, it cannot tell the two layers apart. It only knows there is a number, a club name, a country — and it labels it a transfer and pushes it out.

By the same mechanism, an expenditure of 12.4 billion dong for "coaching consultancy services" once sat in a V.League club's financial report with no accompanying contract. The 12.4 billion never sleeps, but it can vanish — and it vanished precisely at a moment when nobody was looking at the report, because the season had turned the page, because the transfer window had opened, because there was a bigger new rumor to discuss.

A wrong label does not merely add noise. A wrong label creates smoke. And inside that smoke, money can move through without being questioned.

At the fifth joint — the one nobody discusses — the consequences become serious in a different way.

When football data is digitized, it becomes raw material for the betting market. This is not new. What is new is the speed and the granularity. Today, every pass, every duel, every minute a player stands on the pitch can become a parameter. And that raw material flows through the very pipeline that mislabeled you at 22:47.

This is the darkest side effect of digitizing sport: data is supplied directly to betting companies, and the quality of that data depends on the quality of the entire pipeline upstream. The second blood sample does not lie; only people lie. But if you mislabel a data sample at the very input, then by the time it reaches the user, nobody can tell the real sample from the fabricated one.

And then there is VAR.

In my list of wrong labels, the second-most common category after transfers is referees. Not notifications about wrong decisions. But notifications about right decisions — statements asserting that VAR checked, confirmed, found no error.

The problem is not VAR. The problem is that the in-stadium explanation mechanism barely exists at the level of the spectators. When a decision is reviewed, the person sitting in the stand is not told why. After the match, they receive only a label: VAR checked.

That is a label, not an explanation. And this is the biggest blind spot of football's technology era: transparency is only a slogan when it does not come with an explanation mechanism. Fans become the most forgotten subjects in the VAR room — they are asked to believe, but not granted the right to understand.

At this point I must say something clear about method, because if I do not, I will commit the very error I am criticizing.

Three years ago, I investigated a case involving a track-and-field athlete I had once admired. I did not believe the published result. I checked it myself. And I discovered that the athlete's national-team doctor was a name that had appeared in a document I received from a foreign source years earlier. When an idol collapses, I no longer believe in victory. It took me three weeks before I could write a single complete sentence again.

The lesson I drew from that is not "trust no one." The lesson is: do not believe a conclusion merely because it has been repeated many times. A conclusion is only trustworthy when you can follow the path of it from beginning to end — and when no joint along that path labels itself.

So who benefits when the pipeline keeps mislabeling?

The first beneficiary is the seller of traffic. A football label on non-football content still generates an open, a view, an interaction. In the attention economy, that is revenue. Nobody is fined for a wrong label.

The second beneficiary is the seller of ambiguity. During the transfer window, a rumor labeled "transfer" is worth more than a rumor labeled "unverified." Ambiguity sells better than accuracy. So nobody has an incentive to be clear.

The third — and this is the most important link — is anyone who needs a layer of smoke to cover a money trail. An expenditure with no accompanying contract, a transfer fee with two figures, a payment routed through an account that belongs to no one — all of them require an environment in which information is systematically misclassified. Because when the pipeline cannot tell truth from rumor, the fraudster does not need to hide the truth. The fraudster only needs to let the truth be buried among the rumors.

---

Before I close, I must make room for the other side — because an investigation with no other side is just an indictment, and I have no authority to write indictments.

One could tell me that at the current scale, automation is mandatory. Nobody can read hundreds of thousands of articles a month, and keyword labeling is the only practical solution. This is right about the objective and wrong about the conclusion. Humans do not need to read everything; humans need only a domain check. This is a technical problem that has had a solution for years in large content systems. Whether that gate exists depends on whether it is treated as a priority, not on whether it is feasible.

One could also tell me the label does not matter, because readers can tell the difference themselves. Modern readers recognize rumor from fact. This argument is half right. For a long-time reader, perhaps. But for a newcomer — a fourteen-year-old opening the app for the first time, receiving the first notification — the label is the first truth that child is given. And the person paying to place a bet is not a long-time reader; that is a data buyer, buying at scale, with no time to check.

And one could add that mislabeling is harmless, that it is just noise. For a single notification, perhaps harmless. But at the scale of a pipeline, noise is no longer noise. Noise becomes the environment. And a noisy environment enables precisely the things nobody wants: transfer fraud, data manipulation, and a layer of smoke that stretches across multiple seasons.

All three counterarguments hold at the surface. All three fail at the system level. And this is what I want to say as a man who has worked the beat long enough to have lost interest in finding a name to point at: the problem is not the labeling clerk, not the editor, not any individual. The problem is that no one is assigned responsibility for the joint through which everything passes.

---

I went to Moscow to watch football, but I left with a different life. I have told that story many times, and every time I tell it, I have to remind myself that what changed me was not a secret, but a habit: from then on, whenever a pipeline hands me a label, I ask it what it read to attach that label.

Vietnamese football is moving faster than at any point in its history. That is good. But speed always carries a price, and the price of informational speed is accuracy. If we accept that a quarter of our notifications lie to us about what they are, then we are accepting that a quarter of the information infrastructure around this sport operates with no one accountable.

The question is not who mislabeled a notification at 22:47.

The question is: who benefits when this system keeps operating exactly as it operates now — and how much longer will we keep calling it news?