Zero Information Points: The Silence in Cricket's Data Supply Chain That Nobody Files a Report On

A record landed on my desk last week. Its entire payload was this: Domain Label: cricket_asia. No title. No source. No date. An empty list of information points. The part that should bother you is not the emptiness. The part that should bother you is that the system raised nothing — no error, no red flag, not even a two-word note saying the copy never arrived.

Eighteen years ago in Johannesburg the opposite happened to me. March 2026, the World Cup final. Australia made 359/2, India folded for 234, the margin was 125 runs. I spent eleven nights logging all 538 deliveries by hand onto a borrowed laptop, because nobody in India sold ball-by-ball data then. I logged 538 deliveries by hand before I learned to distrust totals. The ledger produced one thing: between overs 12 and 30, India's middle order played 71 dot balls. My editor ran it as a sidebar. The sidebar out-read the main report.

Last week's empty record is the other end of the same story. Then the scorecard had numbers and no truth. Now there is no scorecard, and nobody has filed a complaint.

Cricket's data supply chain runs in three layers, and that matters because the empty-record question applies to all three at once. At the bottom sits the official scorer, entering every delivery — runs, batter, bowler, field position, the reason for an extra. Above him is the broadcast feed, converting those entries into graphics, wagon wheels and commentary. At the top sit the aggregators and newsrooms, converting that feed into scorecards, net run rates, strike-rate splits and 'pressure indexes'.

Everything above is derived from the layer below. The delivery is the atomic unit, and no calculation survives without it. Hand someone a label like cricket_asia and nothing else, and you have handed them a compression artifact of a compression artifact — not a total, but a rumour of a total. The scorecard is a compression artifact, not a fact. That is my oldest line, and it gets truer every season.

The problem is not new in this market. The board releases what it chooses to release; domestic knockout games often have no ball-by-ball question to ask, because what arrives is a photograph of a scoresheet and every model downstream launches from it. Try computing spin economy from a Ranji knockout and you will not know that three overs were scored as 'byes' when they were 'leg-byes', because that distinction was never filed publicly. We do not lack data. We lack the data about the data that is missing — and the alarm that should ring when it goes missing.

So let us open the empty record. Four causes are plausible, and three of them are harmless because three of them get caught.

One: the crawler never reached the page — JavaScript rendering, a consent wall, a paywall, a redirect. Two: a language or encoding mismatch — the extractor reads only English while the original is Urdu, Hindi or Sinhala, or broken encoding kills sentence segmentation. Three: after a site redesign the selector drifted, the parser hit zero nodes, and it silently emitted an empty structure instead of throwing. Four: the input was hollow to begin with — a title or a URL handed in with no body attached.

The third is the dangerous one, because the third is the only one that fails without a sound. The empty list threw no exception. Which tells you the pipeline carries no assertion that zero information points means the output has failed rather than succeeded. It is precisely a scorer submitting a blank sheet while nobody checks whether the over-by-over sums reconcile with the total. Before the number there is a ball, and someone has to count it. On the day that chair is empty, the scorecard is not produced — trust is produced instead, and trust never gets audited.

I have a cheap rule for this, and the rule came out of that 2026 ledger. Every information point carries four mandatory fields: claim (what is being asserted), evidence_type (what the claim rests on — a delivery, a quote, a scoresheet, video), source (who said it, and where), and timestamp (when). Lose any one of the four and you are holding seasoning, not a claim.

Take an ordinary sentence: 'the fast bowler proved expensive in the death overs.' How do the four fields fill? Claim — economy of 10.4 across the last five overs; evidence type — ball-by-ball log, runs per delivery, with the match ID; source — the match's official scorer; timestamp — the match date. The labour of filling four fields versus the ease of saying 'proved expensive' is the gap between reporting and guessing.

Put the 2026 ledger back on the table. It held 538 rows, each with over, ball, batter, bowler, runs and dismissal reason. Had someone accidentally dropped an eleven-over block from that file and the system said nothing, my 71-dot-ball finding would have been wrong, my editor would have printed a wrong sidebar, and no reader would ever have learned. Every scorecard is a summary of decisions made by people you never met — and we must now also accept that some decisions were never made at all; the system simply did not make them.

I learned this once at considerable cost, with my own pressure index. The model said five, I said five, the panel said no. I spent a week deciding that people did not trust the model; it actually took me a week to understand that the people who did not trust it were right about the input. The fault was not in the broadcaster's model, it was in my empty input. Treating a model's defeat as a public-relations problem is the easiest and most dishonest route available; a model's defeat is an input's defeat, and it should be printed first and fixed second.

The real story here is not the empty list. The real story is that one label survived alongside it: cricket_asia. That label is geography, not subject. It tells you where the publisher sat and nothing about what the piece was — a match report, a board appointment, a television-rights deal, an Asian Cricket Council meeting. A system that assigns subject from publisher location routes sport into geopolitics, and the misrouting stays invisible because the label looks like a fact. Everything downstream then gets analysed under a geopolitics-and-business frame instead of a sports-reporting frame, and that wrong frame also travels silently.

The second inversion: more automation has not increased trust, it has reduced it. The old desk had a sub-editor whose entire job was to see whether there was ink on the page; his two words, 'no copy', were a real assertion. Today a function sits in that chair, returns an empty object, and smiles while returning it. I want to credit the print and regional-language desks that were doing verification work decades before the spreadsheets arrived — the sub-editor on the Hindi, Tamil or Bangla desk can still glance at a stat and tell you it is broken. Calling him obsolete is easy. The expensive truth is that once you replaced him, nobody was left to catch it.

The fix is one line, and I will write it without decoration: if the count of information points is zero, the system must make noise — it must not return a result. That is the same assertion that belongs on a scorer's sheet: if the over-by-over sums do not reconcile with the total, the sheet is incomplete, however clean it looks.

The question that follows is who audits the pipeline that audits the cricket. When ball-by-ball data becomes the default for the women's game and Ranji knockouts as well — and it should — an empty record will stop being a silent bug and become an editorial scandal. Until then my question stands: if the scorecard is blank, say so — and if nobody is sitting there to say so, whose scorecard is it?