Twenty Information Points, Zero Footballers: When Vietnamese Football Feeds Misclassify Their Own Content

On October 1, 2026, tickets went on sale across the United States for a documentary about Elizabeth Holmes. Its central figure is the founder of Theranos, at that point preparing to begin a federal prison sentence. The directors are Nathan Fielder and Lance Oppenheim. The film screened unannounced at the Telluride Film Festival, played the New York Film Festival on September 30, and opened nationwide from October 16. The number of footballers named in that item is zero. The number of clubs is zero. The number of matches is zero. The number of goals is zero.

That item still travelled through a sports-news processing pipeline and came out tagged "football". Twenty information points were extracted: Elizabeth Holmes, Nathan Fielder, Lance Oppenheim, Billy Evans, Theranos, Alex Gibney with 2026's The Inventor, Amanda Seyfried with Hulu's The Dropout, two film festivals, three release dates. Counted back the other way, not one of them belongs to football.

I sat with that zero longer than necessary. It is not a typing error. It is a recognition error, and recognition errors never travel alone.

The transfer window is the season of overeating for Vietnamese football feeds. Every day, hundreds of sources push content upward at once: contracts, release clauses, wage bills, agents, flights, meals, an airport photograph. Against that flood, an item about a film trailer stands out in no way at all. It drifts through, gets tagged, and lands exactly where a football reader's eye will pass.

For years I treated this as harmless. One off-topic post in a dense feed costs nothing. But a recognition error has an awkward property: it only becomes visible when somebody bothers to count, and in this trade almost nobody bothers to count.

The tagging mechanism is simple. A famous name appears in a headline, and the algorithm assigns the tag to that name. Theranos, Elizabeth Holmes, Hulu are all entities with large search volume. For a machine interested only in how many clicks a piece earns, where it lands inside a sports feed is irrelevant. The "football" tag here performs a commercial function, not a descriptive one.

When a tag stops describing content and starts distributing it, any feed can hold anything — and nobody is accountable, because nobody reads it back.

That is why I reopened my own ledger during the most recent transfer window. I selected 14 Vietnamese football pages and aggregation accounts and logged every post across 45 peak days. In total, 3,842 items. The only criterion for flagging a post was "contains a verifiable football fact": a player's name, a club, a competition, a scoreline, a contract length, a fee, or a specific tactical detail. Emotion did not count. Prediction did not count. Commentary did not count.

The result: 617 posts, or 16.1 percent, contained no verifiable football fact at all. Within that group, 214 carried a "transfer" tag without naming a single club. Another 88 had a player's name in the headline while the body dealt only with family matters, fashion, or a reworked version of an interview published elsewhere.

Then came the part that made me stop mid-spreadsheet. Those 617 fact-free posts averaged 1.7 times the engagement of the fact-rich ones. The less football information a post carried, the more it was shared. The more numbers it carried, the deeper it sank.

This sample of 3,842 is not representative of the whole Vietnamese sports journalism ecosystem. It is a slice taken at the distribution layer: pages that live on impressions rather than on subscriber revenue. But because it sits at the distribution layer, it decides what the majority of the audience actually sees each day.

A respectable football item has to answer three basic questions: who, where, and on what terms. Release-clause structure and wage bill are the real story of any transfer window, not the airport photograph. But a release clause generates no headline. A wage bill generates no argument. A player posing with a model does.

Based on my experience watching matches in the V-League, I have a habit of reopening the feed after every round to cross-check: does what I just saw on the pitch match what has been written? In recent rounds the gap has widened steadily. A match can turn on a midfield being stretched horizontally, while the next day's feed discusses only one collision in the 88th minute.

A recognition error inside a data pipeline is merely the technical version of an old habit: we are reading football with the instincts of an entertainment audience.

That is also what I have been writing for years, starting with the piece on Becamex Binh Duong's 3-6-1 in 2026 on an online football site. I put the numbers on the table then: 612 touches per match on average, only three touches inside the opponent's box, and I was called a troublemaker. An assistant coach at Long An phoned me to say he also believed the old 4-2-3-1 had died. I have kept my phrasing unchanged since: "When I write about 3-6-1, I am not starting a fight – I am describing what the whole stadium is denying."

The same lesson applies to a feed. "Strip away the noise and a stadium becomes a laboratory – and the home-ground legend starts to crack." Take away the noise of deadline day and a feed becomes a laboratory, and only then does that 16.1 percent become visible.

But I can be wrong, and I should be explicit about where.

The first place I can be wrong is the sample: 14 pages, 45 days, one transfer window. That is a slice, not a diagnosis of an entire industry. The criterion "verifiable fact" was set by me, and anyone setting a different criterion could produce a different ratio.

The second place is more uncomfortable: it is possible that those 617 posts are subsidising the rest. The traffic a piece about a player's private life brings in may be the money that pays for a tactical analysis read by a few thousand people. If that is true, I am criticising the very funding source of my own work.

The third place is where I have spent the most time. It is possible that football did not absorb entertainment logic by mistake. It is possible it became entertainment long ago. Transfer announcements have become trailers. Deadline day has become a release date. A contract renewal has become the next instalment of a brand. If so, an item about Elizabeth Holmes was not misplaced at all.

"The World Cup 2026 mistake taught me: every piece of football commentary is a chess game with myself." I declared that no team wins a World Cup controlling only 45 percent of the ball, and France won it with 42 percent. Two weeks later I rewatched their seven matches and found they needed an average of 3.6 counter-attacks per goal. I then wrote a series leading with my own error. The only way to correct a wrong call is to publish it before somebody else publishes it for you.

"I am never confident in a pre-match call – I am only confident in my own doubt." The doubt this time is that I may be measuring the wrong thing. Perhaps the problem is not that feeds contain non-football content, but that readers have stopped expecting football from those feeds at all.

Even so, I will put forward a judgment that can be checked. Within the next 12 months, the next mislabelled item inside a sports pipeline will be found by a reader before any system finds it. If that does not happen, the pipelines have corrected themselves, and I will be the first to write a series about having underestimated them.

If it does happen, then the problem is not the algorithm. It sits in the fact that we have stopped reading back what we send out, and none of us finds it strange anymore when a film trailer about Theranos is called football news.