Thirteen Against Fifteen: The Two Points Where Truth Hides, and the Wrong Label That Sends Us All Astray

From the Rajshahi print desk, I learned that silence also has a deadline.

In those seconds before the midnight proof page went into the press, the silence wasn't courtesy from the writer—it was the sound of the editor's wristwatch. Inside that silence we made the real decision: which number to print, which to drop. A wrong number earned a phone call the next morning. A wrong label earned nothing, for years. Nobody interviews a label.

Last week a file reached my desk. One word sat on top of it: football. Inside were twenty-two information points. I read all twenty-two twice. No club. No player. No coach. No transfer. No federation, no table, no governing body. What it held was a reality elimination competition, five nominees, a nomination plate, a public voting system, and one outside poll. Not one of the twenty-two points is football.

But the numbers were there. And the numbers were strangely honest. Thirty-one, twenty-one, twenty, fifteen, thirteen.

What actually arrived on our desk

The story: La Granja VIP, a Mexican reality elimination format produced by TV Azteca and broadcast on Azteca UNO. The season's third gala, the third elimination. Sunday, 27 September, 20:00. Five nominees on the plate, final decision by public vote.

Ahead of it, an outside polling outfit called Vaya Vaya published a support measurement across the five nominees: Manelyk González 31%, La Bebeshita 21%, Carlos Trejo 20%, Julio Camejo 15%, Kenny Avilés 13%.

Stop at the arithmetic. Those five figures sum to exactly 100. Not 99, not 101—100.

Five shares summing to exactly 100 means the question was forced-choice, not open-ended. Respondents were not asked who they love; they were asked to pick one of five names. These percentages are relative shares, not absolute popularity. That sounds like a technical footnote. It changes the colour of the entire picture, because in relative shares a tiny movement changes the ranking, and a changed ranking changes the headline.

From more than twenty years of watching matches, one thing I know: what we measure in a stadium and what actually happens are never the same thing. When I called France 4-3 Argentina in the 2026 World Cup Round of Sixteen, the scoreboard shouted a seven-goal festival. The Argentine section of the 42,873 crowd made no sound at all, and no box on the scoreboard recorded that. Since then I have kept one rule: one image per half, never the scoreline alone.

The wall standing between the poll and the decision

The poll does not decide anything. TV Azteca's own voting system decides. The poll was conducted entirely outside that system, and the report itself concedes this—the poll does not represent the official result.

When a proxy signal and the actual decision mechanism are structurally separate, the gap is not marginal. It can be enormous. Anyone assuming the outside poll will match the inside vote is assuming a correspondence nobody has demonstrated. Poll respondents select themselves in; the most digitally active are over-represented. The official vote draws a different demographic. That distortion has a name—self-selection bias.

Then the second matter, the one that matters most. Thirteen against fifteen. A two-point gap. The report itself calls it minimal.

So where is the sample size? The fieldwork date? The geography? The weighting? No answer anywhere in the twenty-two points.

Where the sample size is undisclosed, a two-point gap is not a ranking. It is a tie. Two candidates standing for a photograph, with the lens sitting precisely between them.

This is not an accusation against the report. The opposite. Its most honest sentence is the one admitting the result can change before and during the gala. The headline asks who will be the third eliminated. The body does not answer; it places two names side by side and hands the imagining back to us.

A three-tier structure nobody flagged

Above the noise, the data forms three tiers.

González at 31% leads the next nominee by a full ten points—the only figure with real separation, though the sample-size question returns even here. La Bebeshita at 21% and Trejo at 20% sit one point apart, effectively level, yet seven to eight points above the floor: safe this round, exposed next round. Camejo at 15% and Avilés at 13% sit eight points behind, and two points apart from each other.

The three tiers are distinct, but the bottom tier is not internally differentiated. That is the most important and least discussed observation in this dataset.

Two different kinds of conclusion hide inside that structure. The upper conclusion—who is safe—is comparatively reliable. The lower conclusion—who leaves—is not. A report presenting both with the same confidence is placing a measured truth and a guess on the same scale.

We have known this disease for a long time. When I left civil engineering for Ajker Kagoj in 2026, I learned early that a number is the shadow of a truth, and a shadow's length depends on where the light stands. Here the light source is undisclosed.

The crack between headline and body, and who profits from it

The headline asks a determinate question. Determinate questions expect determinate answers. The body supplies caveats instead: the poll is unofficial, it sits outside TV Azteca's system, the result can change.

The headline claims more certainty than the body will defend, and that empty space between them is where the value sits. If the result matches, the headline wins. If it does not, the body is safe. The piece has arranged to stand in both places.

In March 2026, at thirty-seven, I left a sub-editor's chair at a Rajshahi daily to call Sheikh Russel KC against Abahani Limited Dhaka on a Facebook page. Abahani won 2-1 with an 89th-minute goal, and I described it as a city holding its breath. The stream drew 180,000 views, but I had hesitated two weeks before accepting, fearing online commentary lacked prestige. That hesitation taught me the lesson I still use: do not hide uncertainty; show where it lives. I write pre-match scripts in three acts now—silence, rupture, memory—and I record stadium ambience separately so my writing can return to the crowd's breathing.

Provenance, and the promise blockchain cannot keep

The core problem here is not data quality. It is provenance. Who ran the poll, when, how many respondents, how they were weighted. If those four answers sat in an immutable, timestamped, publicly visible ledger, we would not be estimating. We would be verifying.

That is blockchain's most usable promise: provenance of method. A fingerprint, a time-stamp, a record nobody can later renegotiate.

But the harder point belongs here.

A blockchain can prove where a question came from. It can never prove the question was a good one. An unbreakable lock says nothing about what is inside the box. If the sample is terrible, the wording biased, and the respondents only the most online slice of the audience, then a permanently fixed record of a flawed poll is still a flawed poll. Certification does not rescue truth; it only removes the option of denying what was claimed.

Almost every blockchain-voting conversation centres on whether results can be rigged. This file poses the opposite question. There is no evidence of rigging here. The problem is labelling and prediction, not counting.

And that brings the sharpest detail: the word on top of the file—football—is wrong. That is not a machine's error. It is a human haste error. Somebody attached a tag and nobody checked it. Then downstream, a database recorded that a football item contained no club, no player, no transfer.

A wrong category label works like data's door-to-door product: nobody inspects it, and it quietly contaminates everything inside. Imagine a future system training on this file as football analysis. It learns that a reality-show voting mechanism is football-adjacent. Once the label is wrong, every downstream task is silently poisoned, and the poisoning never shouts.

The contrarian angle

The most professional feature of this report is not an error—it is the language of its own disclaimer. In a marketplace of confident previews, a media outlet questioning its own headline is rare. The tag is indefensible; the caveat is civilised.

Drop the poll and we lose nothing—we discard something. The poll cannot state an outcome, but it can map where attention sits. In that sense it is not intelligence; it is cartography. And its map carries a useful warning: the safest competitor is not the least discussed one.

There is also my own reluctance to name power on air. I have noted it in my notebook and then avoided it live more than once. This report, in its small way, is more honest than I have been: it did not challenge TV Azteca's system, but it stated clearly that this was not that system, and this was not its result. That one line matters. Yet the boundary has to be drawn somewhere. When authority's weakness becomes injustice, the writer's duty is to name it. And the moment a broadcaster implies that success is measured only by votes arriving in its own house, a question surfaces: five adults laughing—on whose account? Nobody asked it. The person voting is the one who most deserves it asked.

The most valuable asset of this cycle is an unanswered question

The report contains a line about mounting tension. That tension may be real, may be the writer's own adrenaline, may be sales. We have no instrument to separate them: no ticket data, no follower-growth figures, no fan-campaign evidence.

Where I cannot measure, I imagine—and then I serve the imagination as information. That is the era's most disciplined form of deception.

What would help is not another prediction. It is a plain sheet with five lines: the fieldwork date; the number of respondents; where they live; the exact question asked; how votes were counted. Where an answer is absent, write simply: not provided. Without those five lines, any percentage is decoration. It is not wrong. It is simply not evidence.

From the Rajshahi print desk I learned that silence also has a deadline. The silence that outlasts the deadline is no longer silence. It is testimony.

Sunday, 27 September, 20:00. One chair empties. If the 13% name leaves, the poll survives partially. If the 15% name leaves, likewise. If a third name leaves, the poll's entire credibility collapses. I do not know which will happen. I do know the largest truth: the poll was never a forecast—its own terms forbade that. Our error was reading it as one. And in making that error we lost the quieter half of the story: the safe competitor, whose silence gets buried under the noise of probable defeat.

In my notebook these days there are two kinds of pages, one for football and one for information. The silence between them has become part of the material. What matters now is what we choose to store, and whether we store the questions alongside the numbers.