A Football Label, and Not One Football Name Inside
**मुख्य उत्तर** एक “फ़ुटबॉल” लेबल वाली विश्लेषण फ़ाइल के 25 सूचना बिंदुओं में एक भी फ़ुटबॉल इकाई नहीं थी — कोई क्लब, खिलाड़ी, कोच या प्रतियोगिता नहीं। 13 बिंदुओं पर स्रोत दर्ज ही नहीं था, यानी स्रोत-घनत्व 48 प्रतिशत रहा, जो डेस्क के 60 प्रतिशत न्यूनतम मानक से नीचे है। **मुख्य तथ्य** - 25 सूचना बिंदुओं में शून्य फ़ुटबॉल इकाई; फ़ाइल पर लेबल फिर भी “फ़ुटबॉल” था। - 25 में से 13 बिंदुओं पर स्रोत “कोई नहीं” — स्रोत-घनत्व 48%, मानक 60%। - नौ विश्लेषणात्मक आयामों में सात N/A लौटे; केवल दो आयामों में सामग्री थी। - फ़ाइल की तारीख़ें 2026 की थीं और किसी प्राथमिक दस्तावेज़ से पुष्टि नहीं हुई; फ़ाइल क्वारंटीन में रखी गई। - सूचना-अधिकार अनुरोध से प्राप्त रिकॉर्ड श्रृंखला इस फ़ाइल का सबसे मज़बूत हिस्सा थी। **स्रोत** Stage-2 गहन विश्लेषण रिपोर्ट, आंतरिक डेटा-गुणवत्ता ऑडिट, 2026। **संबंधित प्रश्न-उत्तर** प्रश्न: स्रोत-घनत्व क्या है? उत्तर: स्रोत-घनत्व किसी फ़ाइल के उन सूचना बिंदुओं का प्रतिशत है जिन पर नामित स्रोत दर्ज हो; यहाँ यह 48 प्रतिशत रहा। प्रश्न: इस फ़ाइल से फ़ुटबॉल डेस्क को क्या सीख मिली? उत्तर: फ़ुटबॉल की कोई सामग्री न होने पर भी उसकी प्रक्रिया-श्रृंखला अधिकांश घरेलू फ़ुटबॉल फ़ाइलों से साफ़ थी, इसलिए स्रोत-टैगिंग का मानक ऊपर उठाना होगा। प्रश्न: क्या यह एक व्यवस्थागत पैटर्न है? उत्तर: नहीं, नमूना एक फ़ाइल का है; तीन सीज़न से कम के आधार पर कोई संरचनात्मक दावा नहीं बनाया जा सकता।
Twenty-five information points. One file. And a label stuck on top of it: football.
I counted three times. The first time in a hurry, because the Friday draft of the "Kick Off" newsletter was open on my screen. The second time with my glasses on, because I assumed age was doing something to my eyes. The third time when I was certain the error in the count was not mine but the file's.
Not one club among the twenty-five points. Not one player. Not one coach, not one competition, not one transfer fee, not one governing body. The names that did appear belonged to a US actress, a county coroner's office, two American news networks and a state authority. Not a single football entity.
That file reached my desk described as football. In nine years, it was the first time the game itself had gone missing.

My desk runs in two stages. The first breaks raw text into information points — who, when, where, who said it, and from which source. The second runs nine analytical dimensions across those points: tactics and technique, club finance and the transfer market, results and the public-opinion cycle, league landscape, rules and compliance, management and dressing-room, risk profile, media narrative, and industry transmission.
I built that structure in 2026, when the Under-17 World Cup was running in India. I watched nineteen of the fifty-two matches from the stands, coded the rest off broadcast, and published an expected-goals and pressing dataset for the whole tournament free the same week. By December the blog had drawn forty thousand reads. That dataset gave me a habit I have never reversed: method first, conclusion second. No table goes out without a method note, a sample size, and a paragraph on what it cannot see.
My source tiers are four — official, reported, estimated, reconstructed. And there is a fifth I added later because I had to: "Source: None". This file became the cleanest specimen of that fifth tier I have handled.
So, the count, because a count is the only honest place to begin. Thirteen of the twenty-five points carried a blank source field. That is an attribution density of forty-eight per cent. My desk's floor is sixty per cent — the line below which nothing is called verified. Forty-eight means fewer than half the points could tell you where they came from. When more than half the rows in a report cannot state their own address, what you are holding is not a report. It is the scaffolding of a rumour.
The twelve points that did name their source were the strongest and dullest part of the entire file: the procedure. A coroner's office formally classified the death as accidental, with a date attached. One network obtained the emergency-service record through an access-to-information request — meaning it asked and was given, rather than being handed something. A second network was running the same record in parallel. The chain has four steps — coroner, classification, public-records request, newsroom publication — and every step has a document sitting behind it.
That is why seven of nine dimensions came back N/A. Tactics N/A, finance N/A, results cycle N/A, league landscape N/A, management N/A, sporting risk N/A, industry transmission N/A. Seven empty dimensions were not an analytical failure; they were proof of an identification failure. I walked into a room that was supposed to contain football and the room was bare.
And yet one thing in it belongs to my trade — and I will say plainly that this is source-process analysis, not football analysis. From years of watching matches I have learned that the value of any number lies not in its digits but in the document behind them. The record derby of 2026 — one lakh thirty-one thousand seven hundred and eighty-one spectators — will stay tagged "reported" in my archive and never "official", because no turnstile count from that era survives in any register. I hold four hundred raw match files up to 2026, and how many of them can I honestly tag? Fewer than half.
The file carried dates in 2026. I could not match them against any primary document I hold, so it stayed in quarantine. I will not go further into the personal or medical detail of that matter, and that is deliberate. Repeating unverified detail about a sensitive event involving a real public figure is not data discipline; it is carelessness, and it is not my desk's work.
The easy lesson is to fix the classifier and move on. I do not believe it, because the classifier is the symptom, not the cause. The real cause is that we reward volume. A desk that gets forty thousand readers off a free dataset slowly learns to rank throughput above provenance. I wrote the Germany File two weeks early and hoped to be wrong — that is the most dangerous habit of all, because a finished draft creates pressure to be right rather than accurate. Had I written a tactical piece on this file — and the words existed, sentences could have been built — it would have been my cleanest and most dishonest copy.
The second point is less comfortable. The file I quarantined had a cleaner procedural chain than half my football files. I have ISL attendance rows tagged "estimated" that I never interrogated as hard as this non-football file interrogated me. That is the real audit.
And one file is an event, not a trend. The pattern was real and one file long, which is not real. So I sample-audited the batch instead of building a structural claim on a single item. I do not write a structural sentence on fewer than three seasons; here the sample was one.
Three rules follow, and they are boring on purpose. Attribution density prints with every table, inside the table and not in a footnote. No file enters the analysis queue without clearing a domain-confidence check — if not one entity in twenty-five matches, the file goes back. And a file that turned out wrong goes into the public ledger, the same ledger where I grade my own bad predictions. This file is boring on purpose. Boring survives a season.
Fifty-two matches fit in one spreadsheet. The fifty-third breaks the model. This time the model did not break — the wrong sport was placed in front of it, and the model answered N/A seven times, politely.

The question is not what was in that file. The question is: if we cannot tag a source on a public-records story, on what basis do we trust that next season's expected-goals input is clean?
