A document landed on my desk on a Wednesday afternoon. Eight sections. Each with subheadings, comparison tables, a risk matrix, confidence tags, and a closing 'Comprehensive Judgment'. The format was complete; not a single cell left blank. And every filled cell carried the same sentence — 'N/A – insufficient information'.

Cricket produces such documents every day. The only difference is that there they are called match reports, and nobody thinks to doubt them. What I was holding was a Stage-2 cricket analysis: eight dimensions, format-based tables, player technique and data, team landscape and rankings, the league's commercial ecosystem, a governance checklist, a risk matrix, public narrative, and an industry transmission map. Its Stage-1 input had come back empty. No title. No information points. No entities. No assessment of time sensitivity. The thing that was supposed to be analysed did not exist.

And yet the document did what the third umpire does on the field — it declined. It wrote 'NOT-GRADED' and asked for Stage-1 to be re-run. That is the real story here, and it is bigger than the story nobody printed that day.

Context: Four layers, and one without a door

Cricket decisions are built in three layers. The first is law — the MCC's Laws, which carry a phrase like 'benefit of the doubt' precisely because the umpire does not always know. The second is Playing Conditions — the board and ICC regulations that change before every series and that nobody reads more closely than the players fail to. The third is technology — reviews, ball-tracking, UltraEdge, and now machine-learning models.

All three share one thing: each layer has a door that says 'input checked'. An umpire confirms the ball was live before hearing an appeal. A DRS review begins only if a review is still available. A match referee's report is written only after the match has ended. These doors look boring, which is why nobody watches them.

The fourth layer — analytics — was born after these doors, and no door was built into it.

I have my own example. In March 2026 in Sydney I watched what happens when a rule is applied to the wrong input. England versus South Africa, World Cup semi-final, SCG. Graham Hick's 83 had carried England to 252; South Africa were 22 to get from 13 balls when the rain came, and when it went the target was 22 from one — because the 'most productive overs' rule then in force was drafted for a format with 60 overs, not 50. Forty thousand people sat in the drizzle and the rule asked for no review. I wrote two thousand words that night at 2 a.m. off three scorecards and a pocket calculator; the sports editor buried it on page 19, because readers wanted a match report, not a clause-by-clause demolition of a regulation.

That night my card index took its real shape. In Sydney, the rain rule was rebuilt on three scorecards. The same habit is now a four-hundred-row spreadsheet: every ICC playing-condition change since 2026, cross-referenced.

Core: 'N/A' is not a blank, it is a verdict

Translate what the Stage-2 document wrote across eight sections into cricket's language and something familiar emerges. Writing 'insufficient information' against a player's technique means exactly what 'not out' means in a scorebook — not a claim that the batsman is innocent, but an admission that the evidence is not good enough.

Cricket has institutionalised this language. When the review system was first trialled in a Test in Colombo in July 2026, India versus Sri Lanka, its most disputed component was 'umpire's call' — the situation where ball-tracking shows the ball so close that it does not clear the threshold to overturn the on-field decision. Critics called it a shield for umpires. It was in fact cricket's first clean, declared 'insufficient information', given a place in the law instead of being hidden.

Duckworth-Lewis followed the same logic. After that night in 2026, boards recognised that a rain rule counts resources, not overs. By the 2026 World Cup, DLS was in place. The new rule was not better because it stopped rain; it was better because it defined its own input — wickets remaining, balls remaining.

That definition was missing from the Stage-2 document, and its absence is the document's most valuable finding.

The interesting part is that the door existed. Section Zero was titled 'Upstream Integrity Check' — a check on the source. Across six points it said plainly that the input was empty and therefore no conclusion could be drawn. But it sat outside the framework, as a preamble rather than a field inside it. A preamble can be skipped next time. A field inside the structure cannot.

That is the genuine structural gap in cricket analytics. We built the models and never placed the input-integrity checkpoint inside the model.

The consequence shows up in three places.

First, the match referee's pitch report. Grading is now a form — Poor, Below Average, Average, Good, Very Good. The form has no field asking what the curator sowed, what the weather was, when the water was given. The form is filled, the verdict is issued, and at the end of the season that verdict affects the venue's next fixture. The man writing the document knows what he saw; the man reading it knows what was written. Nothing sits in between.

Second, auction valuation. A franchise model prices a player off forty balls and releases him off the same forty balls. The quality of the input — how many matches, which format, on what pitch, against which bowler — appears in no table. When everyone chased the money, I stayed with the footnotes. The footnote holds what the headline never does.

Third, broadcast. A win-probability graphic never goes blank. It never says, 'the data for this match is insufficient.' Because the graphic's job is not to inform but to display certainty. And the cheapest way to display certainty is never to express doubt.

The document carried thirteen risk flags: mixing formats, small-sample conclusions, home-ground bias, failing to strip out luck, DRS controversy. Each was marked 'cannot evaluate'. The framework recognised its internal risks, yet had no single row asking, 'does the layer I am analysing actually exist?'

Contrarian: The honesty is in the rulebook, not the media

Here is an uncomfortable point. Cricket's rulebook is more honest than its media. The Laws contain 'not out', 'inconclusive', 'benefit of the doubt', 'dead ball' — a whole vocabulary admitting that the umpire does not know. Analytics has not one such word. Everything there is a percentage.

The argument is made the other way: that writing 'we don't know' is weakness and the audience wants clarity. My experience says otherwise. On the second day of the 2026-83 Ranji Trophy final at the M. Chinnaswamy Stadium, I watched the umpire instead of the batsman. A low, dying catch came to short leg; the umpire said not out; Karnataka's fielders stood their ground; and nobody in the press box could name the clause that governs a ball touching grass while the fingers are still closing. The card index begins at short leg, and the rulebook follows. I went home, read the 2026 Code end to end, found the wording useless, and wrote nine hundred words on it. The desk ran two hundred.

I lost three routine match reports that month. I also became the one reporter in Bangalore who could settle an obstruction argument without leaving his chair.

The genuinely counter-intuitive fact is this: the capacity to express doubt falls with the budget, not with the evidence. The more expensive the production, the fewer 'we don't know' it can afford. Conspiracy-hunting is wasted effort; this is an incentive structure. And it lands on the player — a bowler released on forty balls of data turns up at another franchise next season, and no form anywhere records that the assessment was made on an incomplete input.

Takeaway

What the Stage-2 document did is a template for cricket, and it is easy to dismiss as a process failure. I read it as a model. It demonstrated that a framework can still function when the input is zero — provided it is willing to write 'N/A'.

The question is whether cricket's analytics layer will build that field inside itself. The proposal is small: every analytical product should ship with a mandatory source-quality tag, exactly the way a tournament regulation declares its reserve day before the first ball. Since 2026 I have read every tournament's regulations before the first ball, which usually makes me the only man in the room who knows the reserve day has already been burned.

The index card is small; the precedent it holds is not.