A Football Label on a Funeral-Service Story in Islamabad: Labeling Systems and the Limits of Sports Data

A record sat in the folder labelled "football." I opened it close to midnight, after finishing a review of three Asian World Cup qualifying matches and reorganising my own tracking database. The first line of the record stated the subject plainly: Islamabad is receiving two mobile funeral units, providing free washing and shrouding services for the deceased. The second line noted a sum of 250 million Pakistani rupees. The third named a religious organisation and its general secretary. No team. No player. No tactics. No league table. The label on top still read "football."

I sat still in front of the screen for about two minutes, not out of confusion but because I recognised something very familiar: the labelling system was lying, and it was lying confidently. Forty years in this trade have taught me that the most dangerous error in analysis is never the large one. It is the small error that travels without anyone checking it. A mislabelled record is a small thing. A database that swallows a mislabelled record without a sound is a large thing.

What made me write this piece was not the technical incident. What made me write was the question the incident poses to my own profession: if a story about funeral vehicles in Islamabad can put on a football costume and pass through three processing layers unchallenged, how many of the figures we cite every day are travelling down that exact same road?

Context: a civic story, a sports label, and an unchecked pipeline

The original article was headlined "Capital gets mobile funeral units" and published by The Express Tribune. That is Pakistan's English-language daily, launched in 2026 in Karachi, which at one point carried content partnerships with The New York Times group. The content was purely civic: a religious organisation in the capital, Islamabad, was preparing to put two mobile funeral units into service, provided free of charge, at a stated cost of 250 million Pakistani rupees, roughly 0.9 million US dollars at the exchange rate of the period. The organisation's general secretary, Abdul Wajid, was quoted saying the work was carried out under the directives of the centre's president and was intended to serve humanity.

Not one word of that belongs to football. No team, no player, no coach, no competition, no transfer, no tactics, no federation governance. The only point of contact with sport was a Reuters photo caption placed above the story, referring to peace talks and mentioning the Faisal Mosque in Islamabad. The Faisal Mosque was designed by the Turkish architect Vedat Dalokay and completed in 2026, once the largest mosque in the world. A caption about a landmark, sitting beside a story about funeral vehicles, was enough for an automated classifier to file the whole record under sport.

I work in Tokyo, reporting basketball for the Japanese market, but in recent years I have also tracked Asian football because editorial work forces me to read data from many aggregation sources. And I noticed something: the sports data supply chain we rely on has three layers, and all three have their own incentives not to check too carefully.

The upstream layer is newsrooms, wire services, federations, clubs and official statements. Its incentive is speed. Publish first, correct later. The midstream layer is aggregators, scrapers, labellers and language models. Its incentive is volume. More records is better, and a handful of wrong labels produces no measurable damage. The downstream layer is index providers, betting platforms, recruitment departments and fan media. Its incentive is certainty. They need a number to say out loud, and they rarely ask where the number came from.

The gap sits between those three incentives. Nobody owns a wrong label. And so a story about funeral vehicles in Islamabad passes through all three layers.

The core: how a wrong label crossing three layers becomes a fact

I want to separate two very different things that people habitually merge. The first is a classification error. The second is the propagation of a classification error. The first is cheap and fixable in thirty seconds. The second is expensive, and it can live inside a system for years.

Start with the money. The record states 250 million Pakistani rupees for two vehicles. That is a figure that has nothing to do with football, but it is exactly the kind of figure sports data systems love, because it has the shape of "a large outlay." If an extraction routine strips currency suffixes during normalisation, 250 million can be read as 250 million Indian rupees, and the converted value instantly triples. If the routine maps every large outlay to a "transfer" field, then that 250 million becomes the third-largest transfer outlay of some imagined South Asian league. Nobody needs to act in bad faith. Three lines of carelessly written code will do it.

I am no prophet; I only read the evidence before the current has time to change direction. And the evidence here is this: three of the most consequential facts in the original story — the existence of the two vehicles, the 250 million cost, and the units' design — carry no attributed source in the extract I have. For a local civic item, that may be acceptable. For a record about to enter a database used for analysis, it is a transparency gap that must be flagged, not waved through.

Then there is entity resolution. The organisation's name is a religious centre. If a fuzzy entity matcher handles strings approximately, a religious body can be matched to a similarly named club in another country. The same applies to people. Abdul Wajid, general secretary of a religious centre, can be assigned the role of "club official" simply because that field exists in the schema. A data schema always has an empty box to fill. The problem is that people fill the box because it is empty, not because the information fits.

The heart of the matter is this: a wrong label does not turn itself into false news. It becomes false news when it passes through a system with no feedback loop, where every layer assumes the layer before it already checked.

I have seen this in basketball. In 2026, while writing a six-part series on what I called the new geometry of modern basketball, I built a private database of something I called the silent influence index. I rewatched footage and logged, by hand, every off-ball movement by Jayson Tatum, the Boston Celtics' number three pick in the 2026 draft. On my own measurement, he generated an average of 6.2 points per game through off-ball movement — points that never appear in a personal scoring column. My editor rejected the series as too dry.

But what I learned from that series was not that I was right. What I learned was that my database is only as good as my labelling. I hand-labelled every possession. That is precisely why, years later, when a mislabelled record surfaced in my own database, I spotted it immediately. Someone who has never hand-labelled anything will never see the error.

Around the same period, the Houston Rockets under Mike D'Antoni played five-out basketball with James Harden, pulling all five players beyond the three-point line and turning the space under the rim into private property. They won 65 games in the 2026-18 season. The positionless revolution does not abolish positions; it makes them obsolete. That lesson maps directly onto football: a position is only where you start, the system decides where you arrive.

Pakistan: the team nobody bothers to label

This is the part I find most regrettable about reading that mislabelled record. Because while the labelling system was filing an Islamabad story into the football basket, in that very city Pakistani football was staging one of the most analytically worthwhile stories in South Asian football — and almost nobody was analysing it.

The Pakistan Football Federation has had a turbulent decade. It has been suspended from FIFA membership on several occasions, including October 2026 and April 2026, alongside interventions through normalisation committees. The direct consequence is that the national league has not run steadily, domestic players have lost their stage, and the national team has had almost no continuous competitive cycle.

Then, in October 2026, in the first round of Asian qualifying for the 2026 World Cup, Pakistan beat Cambodia over both legs, securing the first World Cup qualifying win in the country's history and a first-ever place in the second round. That is a genuine, verifiable milestone against the official AFC fixture list.

But in the second round, Pakistan were drawn with Saudi Arabia, Jordan and Tajikistan. They lost every match and finished bottom of the group. The gap was not that Pakistan had no players. Pakistan had players. The problem was that they had no system.

Take Easah Suliman. He is a centre-back developed at Aston Villa's academy, capped at youth level by England, who later chose to represent Pakistan. Otis Khan and Adil Nabi also belonged to the group of diaspora players brought in. Reading the squad list, one might assume Pakistan had just solved its personnel problem. Watching the footage, one sees the opposite: a back three or back five sitting deep, vertical distances between lines stretched wide, and turnovers in transition zones with nobody covering.

This is the classic signature of what I call positionless football without a system. Players still stand in the right spots on the diagram, but nobody knows how to shift when the ball changes direction. A position is only where you start, the system decides where you arrive. Pakistan had positions and lacked a system.

If Suliman plays inside a defensive block organised on the principle of synchronised shifting, his reading of situations — honed at Aston Villa's academy — becomes an asset. If he plays inside a block where the midfield line does not drop in rhythm, every individual quality is neutralised. This is the point most scouting reports skip: they assess the player, not the distance between the player and the system around him.

I once covered the Japan national team at the 2026 World Cup in Russia. In the round-of-16 match against Belgium, Japan led by two goals and lost 3-2. The media called it a tragedy. I wrote a piece headlined "Why this defeat is a cultural victory," citing the fact that Japan ran 12.4 kilometres more than Belgium per match and dominated in high pressing. My editor wanted me to add something emotional. I refused.

My reason was simple: emotion is not a substitute for structure, but it is also not excluded from structure. I treat it as a variable inside the system. Crowd pressure, dressing-room morale, fear of losing — all of them act on the space on the pitch. But they act through mechanisms, not through miracles.

That discipline, I call soft discipline. The Japanese are not strong because they are disciplined; they are strong because they understand the reason discipline is necessary. And that is exactly what a team with only surface-level discipline, without a foundation of systemic awareness, collapses without when it is pinned back. Pakistan in the second round are a clear case: they did not collapse for lack of effort, they collapsed because effort was never organised into geometry.

PPDA, the low block, and reading metrics without reading space

Here I have to be blunt with people who cite data without watching footage.

PPDA, passes allowed per defensive action, measures how many opposition passes a team permits before each of its own defensive actions. Lower values mean more aggressive pressing. Higher values mean a deeper block and more ceded territory.

But PPDA only means something alongside spatial context. A deliberately low block produces high PPDA, and that is not cowardice — it is a design choice. Read the PPDA of a weaker team in World Cup qualifying without knowing they are intentionally funnelling the opponent wide, and you will draw an entirely wrong conclusion about their defensive quality.

In Pakistan's case, a high PPDA does not say they defend badly. It says they defend in a way that accepts being pinned. The problem is that when pinned this way, they lacked a fast enough counter-attack mechanism to turn the opponent's pressure into space behind. A low block is only valuable if it is wired to a counter-attack. Without that, it is a trap that closes on itself.

I make a habit of rewatching every phase and logging the shifts by hand rather than reading a pre-built stats table. That habit costs ten times the time, and it is the reason I catch discrepancies that a stats table will never reveal.

Take Japan against Belgium in 2026. Read the stats table alone and you see a defeat. Watch the footage and measure the running, and you see a team covering 12.4 kilometres more than its opponent. Keep digging into structure, and you see that Akira Nishino's system converted collective spirit into an active defensive machine rather than a passive wall. Those three levels of reading produce three different conclusions, and only the third is usable for the next match.

That is also how I read a database. The table is level one. The provenance of each field is level two. The mechanism that produced that field is level three. Anyone who reads only level one will be deceived in silence, forever.

The verification protocol I impose on myself

I have nothing special except habits. But a habit, repeated long enough, becomes a system.

First, every cited fact needs at least two independent sources. With only one, it goes into the pending column, not the events column. In the original funeral-units article, the three main facts each had a single source, and in the extract they had no source recorded at all. If I were to place that record into any aggregate table, I would have to mark it unverified.

Second, I separate inference from confirmation. Inference is written in if-then form. If the tournament organiser keeps the qualifying format unchanged, then Pakistan will again sit among the lowest seeds. If domestic federation instability continues, then milestones like October 2026 remain isolated points rather than a trend line. A reader is entitled to know what is certain and what is assumed.

Third, I publish my data limitations. When I wrote the 2026 Tatum series, I stated clearly that the 6.2 points per game figure was my own footage-based measurement, not an official league metric. That led some editors to call the piece lightweight. It also made the piece honest, and honesty is the only thing left standing after a season ends.

Fourth, I treat emotion as a variable. Not excluded, only controlled. A fractured dressing room affects the distance between lines on the pitch. Crowd pressure affects whether a team dares to hold a high line. An analyst who refuses to look at those variables has covered one eye and declared himself three-dimensional.

Fifth, and most importantly: when I push back, I must identify a specific flaw in the opposing argument. Dissent for the sake of looking different is not dissent, it is performance. I do not need performance. I need to be right.

The counterintuitive angle: the mislabelled record is not the scandal

Here I want to push the argument one step further, to a place most sports readers will not enjoy.

That mislabelled record is not a scandal. It is an error. A Pakistani daily reported on funeral vehicles, an automated classifier saw the word Islamabad and a photo caption about the Faisal Mosque, and assigned the wrong tag. Small matter. Thirty seconds to fix.

The scandal lies elsewhere. The sports industry has built an entire data supply chain trained to accept labels, not to interrogate them. And that reflex does not live only inside automated systems. It lives inside practitioners like me, and inside fans.

We are perfectly willing to interrogate a coach's substitution. We will argue about a penalty. We will write thousands of words on whether a forward should play left or right. But when we see a spreadsheet with a club name and a sum of money, we believe it by default. Nobody asks who labelled the spreadsheet.

Meanwhile, world football spends almost all of its analytical bandwidth on the five major European leagues. There are roughly two hundred FIFA member associations, and most of them have almost no verified data at all. If a match in Islamabad has nobody recording indices, then Pakistan's football story will forever be told through belief rather than through geometry. That is the biggest blind spot in global sports analytics, and upgrading a model will not fix it.

But I want to go one step further still, and this step will irritate some colleagues. Judged by its own standards, the Islamabad funeral-units article is a more carefully reported civic story than most transfer items we read daily. It names a specific organisation. It names a specific person with a specific title. It carries direct quotes. It states a cost. The transfer item we consume every day often carries two words, "reportedly," and an unverified name.

That contrast is worth sitting with. We distrust a civic story because it falls outside our topic of interest, yet we trust a transfer story only because it falls inside it. That is a standard of judgement based on preference, not on evidence.

What to watch

Over the coming months, three things will hold my attention.

First, whether that record's label is corrected. If the database keeps a "football" tag on a story about funeral vehicles, the problem is not the classifier — it is the person operating the database. An error that cannot be corrected is the signature of a system with no feedback loop.

Second, Pakistan's next qualifying cycle. The question is not how many more diaspora players they can add, but whether they can build a synchronised shifting system. Players alone will reproduce the last second round: no wins, bottom of the group. A system would turn the Cambodia win of October 2026 into the start of a trend line rather than a statistical anomaly.

Third, whether sports data providers begin publishing their own error rates. A provider willing to say that three in a hundred of its records carry a wrong label will be far more credible than one claiming universal accuracy. Transparency about error is the mark of a mature industry.

I am no prophet; I only read the evidence before the current has time to change direction. And the evidence I read from this story is not that a machine got something wrong. It is that nobody in that chain treats being wrong as something that must be reported.

If a story about funeral vehicles can pass through three processing layers without anyone stopping it, then the next time you open a data table and see a number printed in bold, who will you ask about it?