Zero Evidence, Complete Template: The Danger in Cricket Analysis Isn't the Result, It's the Scaffold

Last week a document landed on my desk. Eight sections. Every heading in place, every sub-heading filled, every table ruled and rowed. Format and match analysis, player technique, team landscape, league economics, governance, a risk matrix, narrative, industry transmission. Four thousand words, give or take. And not one fact in any of them.

Every cell was occupied — by the phrase "N/A — insufficient information." One thing had survived the wreck: a tag, cricket_asia. Two words. That was the entire payload.

In March 2026, in the press box at Eden Gardens, I had at least thirteen wickets, a 281, a 180 and a 376-run stand to work with. Those numbers existed because somebody logged them. That night I read the space rather than the score. But reading space requires a score to read it against. Here there was no score at all.

Cricket coverage now runs on a three-storey supply chain. The ground floor fetches the source, parses it, classifies it and pulls out facts. The middle floor hangs an eight-dimension framework on those facts. On the top floor sits the writer, who turns the scaffold into sentences. The chain works only if something survives the ground floor.

The failure visible in this document was manufactured before anything reached the top floor — a paywall, a JavaScript-rendered page, a PDF container, a blocked request. The pipeline did not read the article. It looked at it and moved on.

There is a detail here that lifts the whole episode above "technical glitch" territory. The cricket_asia tag survived. Which means the raw text existed at some point: the classification layer saw it and filed it under Asian cricket. The question is not whether the article existed. The question is that it was seen and not captured.

The first of the eight dimensions is format. That is not decoration, it is a precondition. A Test average, an ODI economy rate and a T20 strike rate are not substitutes for one another. Without format, the other seven cannot open — powerplay structure, the two-new-ball phases, session attrition, all of it is format-relative. Without a single element, the scaffold becomes a locked room whose key has been lost.

In February 2026, sitting inside Delhi Daredevils' support staff, I learned the same lesson from the other side. A five-million-dollar purse, fourteen league matches, and no historical data for pricing a role. We did not have the information — but we knew we did not have it. That is the difference. An empty template does not know it is empty.

Analytical failure comes in three kinds, and each is sold at a different price.

The first, and noisiest, is the empty scaffold — this document. Honest, transparent, useless. It does no harm, provided nobody forwards it as "analysis."

The second, and the real danger, is the filled scaffold. A model that could not find format, players or teams does not fill its blanks with estimates. It fills them with plausibility. In this document's risk matrix, one row sits at the top: "Downstream hallucination risk from an empty evidence base — High." The system named its own worst defect, and I think that is the only genuine finding in the whole file.

The third failure belongs to the consumer — mistaking the scaffold for findings, and it is the costliest. Faced with eight sections, tables and a risk rating, the mind does not go looking for the missing facts. It accepts the structure as a substitute for evidence. I made exactly this error in Ahmedabad in November 2026, from the opposite direction. My model rated India's middle-overs control the most robust structure of the tournament across ten matches; one night, Travis Head's 137 and a 192-run stand with Marnus Labuschagne ignored that structure entirely. I rebuilt the variance layer and filed three hours late, thinner than I wanted. The lesson was to ship an error band instead of perfection — "India control 68% of middle overs," not "India will win."

But an error band has a component too: a centre. If the centre is zero, the band is decoration. You cannot wrap uncertainty around something that does not exist.

There is a second thing in this document that grates professionally — the circular dependency in source grading. The ground floor deferred source quality to "the source fields of the information points," and supplied no information points. A door whose key is locked behind it. This is not a bug to me. My vertical runs on the reader being able to check me: author, outlet, date, and the number behind the claim. The day those four go missing, the right to write goes with them.

Which is where the auction room earns its keep. In 2026, with no data, we did not price reputation. We priced role scarcity — death overs and powerplay enforcer held in separate ledgers from batting average. That was legitimate because the sheet declared its own assumptions. "In the auction room, I stopped counting bids and started counting incentives." An empty template declares nothing — no assumption, no error band, no centre. Only cells.

And a third point no pipeline report ever writes: the second clock. "Every auction room has a second clock running." Here that clock is cache expiry. The raw text sat in a buffer somewhere, classified, and is now moving out of reach. Refetch the same URL and the article may not be there — the page updated, the paywall closed, the CMS redirected. "Reject it and redo it later" is not good enough. This is a now job.

I should say plainly that the cricket risks I cannot see here — injury, governance, fixing, financial fragility — are not thereby clean. No assurance of any kind can be extended to any entity, because no entity was present. Absence is not evidence of cleanliness. This is the most basic analytical discipline, and the most frequently broken.

Now the part that makes me uncomfortable, and the part least likely to be said.

The industry's great fear is AI — that a machine will invent something that never happened. My fear runs the other way. The more common and more expensive failure is a system producing the shape of rigour without the rigour, and an editorial process accepting it because the shape looks like work. A forty-page risk matrix in which every cell reads N/A appears far more credible than a clean five-hundred-word "we don't know" — and is far less useful.

Here is an uncomfortable truth I apply to my own trade: length is the oldest counterfeit evidence in the business. The piece you are counting lines on was commissioned to a fixed word limit. Behind that demand sits a market — readers who watch every match and want the signal before the headline. I read that demand as data, while recording that demand tells you what the audience wants, not what happened on the field. "The ledger said one thing; the stands told another." Here the ledger was blank and the stands still wanted copy — which is itself a story, provided it is filed under opinion and not under news.

One more inversion. When source grading depends on information points and there are none, the correct editorial call is not "we'll look at it later." It is: label this BLOCKED and do not pass it on. That feels like shirking. It isn't. An unlabelled empty scaffold becomes a filled scaffold the moment it changes hands.

I will pre-commit in print to what changes my call: if re-extraction returns at least three substantive information points and one explicit format statement, the entire scaffold becomes executable immediately, and this article's premise — that the facts were never there — collapses on the spot. I will put that correction on the record, because I promised myself one public reversal.

So what comes next. One deadline and five fields.

Re-extraction from the raw source before the cache expires. The minimum contract: article title, outlet, publication date and author; format and competition name; teams, players and venue; match state or the specific dispute; and at least one quantitative anchor. Five things. Without them, not eight dimensions — not one.

"Tactics are patterns that survive contact with money, weather, and ego." I would add a fourth contact: the blank page. A framework that starts calling itself fact on contact with a blank page was never tactics. It was typography.