The Octopus of Progreso and the Classification Gap in the Football News Feed

The Octopus of Progreso and the Error No VAR Could Catch

In a vertical clip barely a minute long, a fisherman in Progreso, in the Mexican state of Yucatán, is trying to peel an octopus off his face. The animal wraps its eight arms around his head, the suckers gripping the skin, while the man pulls hard forward with both hands. Seawater drips from the tentacles. Laughter, breathing, and voices mingle. Nothing about the scene involves a match: no player, no referee, no score, no card. Yet it landed in a feed labeled "football."

The person who filmed and reposted it is the journalist Hiram Hurtado, who shared the footage on X, and within a single day it spread across timelines. Early users shared it out of curiosity. News pages shared it for views. Then, somewhere in the meshes of a system, a machine that reads text stamped the story with a sports tag. I read this technical analysis on an evening in Berlin, and what drew my attention was not the octopus. What drew my attention was the gap between what the event actually is and what the system says it is.

I have sat in a broadcast booth listening to arguments over whether a ball had crossed the line. I have written about contract clauses nobody bothered to read. But I had never seen an event placed so thoroughly in the wrong drawer, and, more seriously, placed there quietly, with no objection and no whistle.

Context: A Door Opened by Mistake in the Digital Newsroom

To understand why a video about a fisherman sits beside a transfer story, you have to look at how the feed operates. Most sports pages no longer rely on humans as their only gatekeepers. They run an automated chain: gather sources, extract named entities, classify topics, then push content into sections. Entity extraction, in the trade's language, is the machine identifying who and what appears in a sentence. Topic classification is the machine deciding whether a story belongs to football, music, current affairs, or lifestyle.

When that machine read the octopus footage, it did not see eight arms. It saw a string of characters. And within that string, words familiar to football may have been mixed in: "catch," "target," "attack," "defense," "cling," "tussle." An octopus "catches" a fisherman's face. A creature hunts its "target." The man "attacks" the animal to break free. To a system that merely counts words, this story reads exactly like a report about a duel for the ball.

That is the classic trap of automated classification: keyword collision. One word sits in two entirely different fields. "Catch" in football is intercepting the ball. "Catch" in life is holding on. "Target" in football is the goal. "Target" in life is an objective. "Defense" in football is the back line. "Defense" in life is self-protection. Let just a few such words slip in, and the machine nods and tags.

Once affixed, that tag outlives the story. The video ends, the views cool, but the data stays. It enters the weekly statistic on football content share. It is counted alongside genuine transfer reports. It blurs the picture an editor uses to decide what to publish tomorrow. A small error at the input becomes a distortion at the output, and nobody notices, because nobody rechecks every label.

Analysis: Dissecting an Error With No Whistle

In refereeing, the first thing I learned is this: a wrong decision does no harm equal to a wrong decision nobody checks. On the pitch there are at least four assistant referees, a VAR team, twelve cameras, and seventy thousand fans screaming at you when you err. In the feed, a classification error has no one to scream. It drifts by in silence.

Let us start by pinning down the true nature of the event. The source contains thirteen information points. Read all thirteen and you find no club, no player, no competition, no governing body, no deal, no financial statement. The only subject mentioned is a fisherman and an octopus. The only geographic setting is Progreso, a coastal town in Yucatán where fishing is a livelihood. The only media figure is Hiram Hurtado, who posted the images. None of that belongs to football.

Yet the "football" label exists. That teaches three lessons about data.

First, a label is not a fact but an assumption no one has yet refuted. When the system tags the octopus video "football," it is not asserting the event is football; it is merely saying no one has stood up to say otherwise. The machine's default is to believe itself, and that default breaks only when a human checks. If the process has no human check, the default becomes reality.

Second, the input error is larger than the output error, and harder to see. If a feed carries one octopus story among ten football stories, a reader's eye may skip it. But if it is one item among ten thousand, and the machine uses that data to forecast trends, the error multiplies. The editor did not err. The machine did not lie. A single mesh in the net tore, and the whole net dragged crooked.

Third, viral content flows anywhere it likes, while sports content must fight to be noticed. The octopus video spread for its novelty. A solid tactical analysis rarely spreads that way. So sports feeds, already crowded with rumors and emotion, get pushed further toward whatever grabs attention. The door opened by mistake did not just admit an octopus; it admitted a habit.

To gauge the severity, place two numbers side by side. One: the sporting value of the octopus video, which is zero, absolutely zero. Two: its reference value as a data-quality signal, which is high, because it is concrete, verifiable evidence of a defect in the pipeline. The video is worthless as football news but valuable as a bug report. That paradox is why I am writing this piece.

When Clause 17 lies on the deliberation table, I remember the way Neymar stepped across the law without looking down at his feet. Here too: the machine steps across the definition of football without looking down, because that definition is not in its keyword table. VAR is not wrong. What is wrong is the way we believe it can replace a night when the referee makes a mistake. And automated classification is not wrong in the sense we usually assign. What is wrong is the belief that an automatic label can replace an editor who sits and reads.

One more detail in the file deserves notice: the octopus caused no serious injury to the fisherman, and the man was able to continue his fishing day. This is the only data point that can count as a "result," and it belongs to coastal workplace safety, not the scoreboard. That only confirms it: every sporting analytical dimension in this file is empty. No match rhythm. No lineups. No form. No standings. No transfer market. No deal, no release clause, no wage bill. No playing rule violated. No dressing room disturbed. Every one of those boxes sits in a state of no information, and truthfully, that is the correct conclusion.

What I want to stress to those in the trade is this: a wrong label does not only soil one feed. It corrupts the measuring capacity of an entire system. If an octopus counts as football this week, next week a cooking video containing the word "defense" will count as football too. The noise ratio rises. Report credibility falls. And readers, already weary of fake news, gain one more reason to trust nothing. A small technical error, accumulated long enough, becomes a crisis of trust.

There is a line I once wrote about building audience trust: saving a club is not about football, but about a city that had lost faith in the whistle. Today the whistle has moved into the hands of algorithms. And when the algorithm blows it wrong, no one hears anything left to argue about.

The Contrary View: The Machine Is Not Alone in This Error

To blame only the algorithm is to ignore the truth that humans put the algorithm there. A system tags automatically only when someone decided tagging does not need a human. The machine does exactly what it was told: react to keywords. It does not know what football is, nor does it need to. The person who told it that knowing or not knowing does not matter — that person is the one who erred.

Here lies an economic pressure the sports-news trade rarely states plainly. Views are money. An octopus video yields clicks faster than an analysis of a contract clause, even if the analysis is a hundred times more useful. When posting speed becomes the measure of productivity, people tend to loosen the checking stage. And the checking stage is the only stage capable of catching a classification error. Cutting it to run faster is blinding yourself in a corridor full of doors that open by mistake.

On the other side, there is a justification that sounds reasonable: audiences want the octopus video, so running it serves the audience. I do not buy it. Audiences want many things, and most of them do not belong on a football page. Serving the audience does not mean handing them anything that makes them click. It means keeping the promise of the section. A football page promises that when you open it, you will get football news. Breaking that promise for a few seconds of attention is a losing trade, however green the ad board.

For someone like me, who has followed matches from the stands and the studio, this error carries a professional meaning too. I am used to a situation being misread in an instant, but there is always a way to fix it: review, consult, admit. The sports-data industry needs to learn that very habit. A machine that admits error and fixes a label is not a weak machine. It is a machine capable of self-correction. Honesty about error has been, is, and will remain the thing that decides the credibility of any information system.

People once told me the worst refereeing decision is the one where you dare not change your mind after seeing the monitor. The worst classification system is the one that has seen the error but dares not remove the tag. The error lies in believing that data, once correct, needs no rechecking. On the night I faced VAR, I learned that technology is not at fault. Its operators are. That holds for the monitor in the VAR room, and it holds for the monitor in the newsroom.

A Proposal: How to Keep the Octopus Out of the Feed Next Time

There are small fixes, doable now, that need no grand overhaul. First, set a check threshold for any content the system tags automatically before publication: if no football entity is recognized — no club, no player, no competition — the tag must be suspended pending human review. A minimal entity list is enough to catch most errors of this kind. Second, split ambiguous keywords into fields by context rather than sharing one vocabulary. "Catch" in football and "catch" in life should not sit side by side in the same table. Third, regularly sample already-tagged items and recheck them by human eye; the cost is small, but it is a safety net for the whole chain.

Beyond tools, a professional principle is needed. A label is not decoration for a section; a label is a promise to the reader. Once promised, it must be kept. A miscategorized post, however small, is one breach of faith with readers — no grand betrayal, but repeated often enough, trust erodes like a sandy shore.

And finally, I propose what in refereeing we call the post-match report. Each week, those who produce sports content should keep a record of classification errors that occurred, how they were fixed, and what was learned. Not as an administrative ritual, but as a habit of professional learning. Referees review the tape after every match, even a smooth one. News people should review the path of their data, even when the chart looks good. For the most dangerous thing is not being wrong, but being wrong without knowing it.

An Open Thought

The octopus of Progreso left the fisherman's face long ago. But the wrong label is still there, sitting in some database, waiting to be counted into some report. The right question is not about the animal, but about the person who tagged it, and about those who trusted the tag without looking again. When a sports feed grows long and fast enough to contain an octopus, that is the moment to ask ourselves: if one day a real match were mislabeled and vanished from sight, would we slow down enough to notice?

Saving a feed is not about the data. It is about a reader who once believed that opening a football page would show football.