The File Came Back Empty: When Cricket's Data Pipeline Mistakes 'N/A' for an Answer

It was raining in Liverpool on Sunday night when I reopened a file that had been shut for a fortnight. The screen held a table — entity, role, format, venue, ranking. In the box marked 'Entities Involved' there was a sentence: identify from the information points above. I scrolled up. The Information Points field was entirely empty. Zero items. Title: N/A. Source: N/A. Article type: Unclassified. The domain label read cricket_world, which meant the machine knew only that this belonged to the world of cricket, and nothing more.

What was supposed to be analysed had vanished. What stood in its place was intact: eight dimensions, every sub-heading in position, every cell carrying the same phrase — N/A, insufficient information. For thirty years I have searched scorecards for the number that refuses to fit. Last night the number that refused to fit was zero. In cricket, zero is never data. Zero is a record of where the route to the data broke.

This is not a story about a match. It is a story about the machinery on which every cricket story now rests.

March 2026, Eden Gardens. I was twenty-nine, filing for a London agency at the Wills World Cup semi-final. Sri Lanka made 251 for 8. India were 120 for 8 in 34.1 overs when the stands went up in flames and the match was awarded to Sri Lanka. The press box wrote riot colour. I sat with 37 completed scorecards and found what nobody printed: between overs 18 and 34, India's scoring rate had fallen from 5.1 to 3.4. The collapse was legible long before it was visible.

In April 2026 the Cronje story broke; in November Azharuddin was banned. By then I was compiling odds in Liverpool, still logging every one-day international I could watch. I ran my 2026–2026 file backwards and found eleven ODIs where the pre-toss price had moved more than three standard deviations from my model with no team-news explanation. I sent the file to two editors and a trader. Nobody printed it. The trader hired me.

That same file is open again today, in another form. Every cricket score passes through an API in fifteen seconds, every preview comes off a pipeline, every auction valuation is built in a spreadsheet. Broadcast rights, franchise valuations, central contracts, syndication — the whole commercial layer stands on a data layer nobody audits. And when that layer comes back empty, it does not say loudly that it knows nothing. It simply types N/A.

Now the actual audit.

The first error repeated in every hollow report is to drop 'unknown' and 'low risk' into the same box. This file said 'cannot assess' throughout, but it also carried the most honest sentence in the document: what is not known is not a low risk. In cricket that distinction carries price. A fast bowler whose injury record is unavailable is not fit; he is unverified. A team with no home-away split on file is not strong; it is undefined. An empty box is not a green box, and cricket's entire commercial economy is a bet on that single difference.

The second finding is where the fault actually sat. Inside the template, in two places, the text was not machine output at all but an instruction meant for a human — 'identify from the information points above', and 'judge from the source fields of the information points'. Where a result should have appeared, a prompt appeared instead. That is a clean fingerprint: retrieval probably ran, extraction did not. The article may well have existed; the layer reading it came back blank. When a file prints a question where an answer belongs, a human carelessness has entered the document, not a machine one.

The third is the format gate. This is not a technical nicety; it is the first question. Without knowing whether the case is Test, ODI, T20 or The Hundred, no figure in cricket is comparable to any other. A spinner's economy of 8 builds a career in T20 and means nothing in a Test. Test average, ODI strike rate and T20 economy are three separate languages. Converting between them is like mistaking ball speed for bat speed. In a report where the format field itself is empty, every remaining number is decoration. Format is the first question, not the last.

The fourth cut deepest. This entire file contained not one number. No transfer fee, no auction price, no ranking, not a single score. That is the good news, because the most dangerous thing in a hollow template is a figure poured into it. A report with no facts is useless. A report with invented figures standing in for facts is useless and harmful at once. Which is why the document was careful to record integrity status as 'unknown' rather than 'clear'. The ACU's language follows the same rule. My 2026 file taught me this: when you know nothing about a match, do not call it honest. Call it unknown.

The fifth is the most expensive one for any editor — time. Cricket narratives rot quickly. Bilateral series coverage is stale in days, auction-cycle coverage in weeks. If fixing an extraction fault takes 48 hours, those hours were not merely lost to the pipeline; they were taken from readers, traders and selectors. Football's transfer window is the clearest example: there, rumour is priced above fact, because rumour has itself become a data point. Cricket is drifting the same way.

Now the objection you are about to raise. This is just a bug. Fix the pipeline and move on.

I read it the other way. The pipeline behaved correctly this time — it invented nothing. That is fail-closed behaviour, and by design it is right. The failure sits in the human layer that will read this empty template and conclude all is well. We worry that AI will manufacture facts. In cricket media the more common and cheaper failure is manufactured silence. The checklist came back blank, therefore no red flag appeared, therefore everything was assumed green. The reader was told 'no problem' when the truth was 'no check was run'. Formatting took the place of information.

And here I must apply my own rule, because after thirty years I still distrust the obvious odds. This is n=1. One document, one pipeline, one night. I cannot call it systemic. I can only say the fingerprint exists — and where a fingerprint is found, you sample; you do not publish a verdict. Those who print the conclusion first and inspect the sample afterwards repeat the error next time, in a larger font.

The file will be re-run. When it comes back full — and it will — the first question will not be what we learned about the match. It will be how many hours of the news cycle the empty file consumed. Next time you read a preview, ask it three things: what is the format, how large is the sample, and where did that number come from. A report that answers none of the three has told you nothing; it has merely filled its template. The ledger never lies, but it often whispers. What it whispered this time was not a score. It was an empty box.