A Report That Looks Finished and Says Nothing: Cricket's Invisible Data Labour and the Disease of the Hollow Skeleton

Hook: The File That Looked Finished

Last month a document landed on my desk. Fourteen subheads, six tables, thirty-one numbers. I read it once and thought the work was done. I read it again and understood the document answered no question at all, because it contained one empty chamber, and that chamber was where everything should have begun. No title. No source. No core argument. No list of information points. Only a label survived: cricket. A label is not content.

That evening I thought of the press box at Chinnaswamy. Monsoon outside, a man in the second row with two laptops and a pencil. The thirty-fourth ball of the over went toward cover, the fielder half-stopped it, two runs. He typed: two runs, cover, fielding error yes. That single line produced the wagon wheel at midnight, the pitch map, the fantasy points, and by morning an analysis claiming the bowler could not turn the ball from mid-off. Nobody prints his name. The beat is not in the drum; it is in the water carrier. I learned that in 2026, riding Bengaluru FC's team bus for nine months and watching more than a hundred training sessions — the visible contest and the invisible scaffold are two different games on the same field.

Context: Where Cricket's Data Chain Begins

The chain starts at the ground. A scorer makes the first mark, an encoder sitting beside him converts it into structured fields, the vendor pipes it to the broadcaster's graphics package, an API pushes it to fantasy platforms, auction models, team analytics desks, and finally to notebooks like mine. Each step's output is the next step's input. When the first step is empty, the second step does not fail — it fills defaults. And defaults look exactly like data.

That is the disease of this moment. An IPL season is 74 matches; a Test day is up to 90 overs, 540 deliveries; five days is roughly 2,700 balls, each carrying eight to twenty fields. Domestic cricket is worse, because cameras and encoders are fewer. Volume pushes toward automation. Automation is not the risk. A pipeline that never stops is the risk. A system that shouts when it fails is a good system. A system that inserts defaults into an empty chamber and reports success is a liar.

The document on my desk had that illness. Nothing at the root, sentences arranged above it. In the regular-season content economy this has become ordinary, because nobody is held to account.

Anatomy of a Hollow Report

I opened the file and checked field by field. No title, so there is no source to verify. No source, so the type cannot be classified — match report, selection critique, contract news? No type, so the core argument is blank. And the information-point list is empty. That is where the chain breaks. Entities were supposed to be identified from the information points; time sensitivity and source quality were supposed to be assessed from the same place. With the root empty, all three collapse. This is not shallow analysis. There is no object to analyse.

In scorecard language: imagine a sheet with extras filled, total filled, and no batter credited with a run. You would suspect the arithmetic instantly. Cricket makes that check easy because the sums reconcile. Writing does not. Fourteen subheads with zero information underneath cannot be added up. That is why the hollow skeleton is the safest crime in cricket journalism — no scorecard testifies against it.

I have kept two columns in my notebook for years. Tactical on the left, human on the right. When the left fills and the right stays empty, I know the piece will be decoration. The reverse is also true. In Russia in 2026, standing in the mixed zone at three in the morning, I learned that asking a player how he feels produces a quote, while asking whose work his role served produces information. Only the second survives editing. Hollow reports never run that test: they record roles, not the relationships that make roles mean anything.

The Root Node: Second Row of the Press Box

I have counted, in scoring rooms at several grounds, how many fields go into a single ball. In domestic games it runs between eight and twenty — bowler, batter, runs, zone, delivery type, line, length, shot, fielding, field setting, physical note. Multiply by 540 deliveries and a Test day holds five to ten thousand fields, plus review timelines, DRS decisions (three per innings in Tests, two in limited overs), field changes, the new ball, the light meter. The person carrying that is paid a match fee, not an hourly rate, and the contract never says a match can run ten hours.

I refuse to write these people as selfless, silent, fated. The truth is more ordinary and harder. They calculate. The man in the second row with two laptops wants a chair in a team's analytics department. Scoring is his staircase, not his destination. That gives him bargaining power and he knows it. No vendor can replace him mid-tournament, because he knows this ground's third-man boundary is short, the scoreboard is manual, and the sight-screen shadow falls into the batter's eye at a particular over. That spatial knowledge is in no database, which makes him close to indispensable.

His resentment is real and correctly aimed. The graphics package takes the credit, the insight piece takes the byline, the correction never gets a line. Yet if a single field is wrong — caught or not caught — that field decides the bowler's wagon wheel and the batter's control percentage. James Anderson's 704 Test wickets and Ravichandran Ashwin's 537 are the big numbers, but one bad entry underneath can invert the sentence built on them.

On the Bangladesh–India corridor the problem sharpens. In Mirpur the press box writes in Bangla shorthand — cover for two, misfield. In Chattogram the board is still manual, every run changed by hand. That shorthand does not map cleanly onto a global schema. The translation layer quietly drops fields. The pipeline reports success while an over's story disappears. Data that cannot read Bangla will not read Bangla cricket.

Data Measures Outcomes, Not Decisions

My objection is not to data. It is to the claim. Expected runs, control percentage, pitch maps, wagon wheels — all describe distributions of outcomes. None explains a decision. Picture the sixty-first over, the ball old and reversing. The captain brings third man up and posts him at slip. The log records: field change, over 61. Why is in no field. Data will say the scoring rate fell that over. The cause was the bowler's wrist position and a one-minute conversation with the wicketkeeper.

Data measures the outcome of a decision; it does not measure the reason. Filling that gap is the writer's job, not the scorer's. A writer at the ground sees the point fielder take three steps back, and that appears in no field. If it is not in the notebook, only numbers remain, and numbers cannot write a reason. A Cheteshwar Pujara-type innings proves it from the other side: forty-four strike rate, two hundred balls. The log says slow. Inside those two hundred balls, seventy runs came at the other end. Control percentage counts the balls missed, not the plan to absorb them.

Error Is Less Dangerous Than False Confidence

A wrong report can be corrected. A report that admits no gap inside itself has no door for correction. I began filing minute-by-minute tactical annotations beside my match reports for exactly this reason: if I cannot write the minute, I was not there, I only watched the scoreboard. The annotation is a fail-safe. The hollow skeleton removes it.

A report that admits no gap inside itself is the one most likely to be wrong, because its confidence comes from structure, not substance. The check is a single question: where did each number in this table come from? If the answer stalls, the table is furniture. Readers rarely ask, so the duty sits with editors — and in my experience the failure happens precisely there, where captions are polished but the provenance of the numbers inside them is never requested.

Money Sits in the Visible Layer

Broadcast rights money flows to the visible layer: graphics packages, AI-powered insight modules, sponsor boards, pre-show segments. At the root — the match fee of the man with two laptops — the figure barely moves. To my knowledge many domestic tournaments run encoding on contract, where the commercial target is more matches with fewer people. The sponsor who paints a name on a ground board does not want to paint it on a scorer's pay slip.

This is the shirt-sponsor problem in cricket-data form. When a global brand erases a local name from a club's chest, the club becomes a logo and stops representing a community. The same happens with data: a global measurement standard severs the number from the ground, because it has no metric for belonging, only for exposure ROI. Graphics get prettier, roots dry out. Domestic cricket loses most, since the gaps are covered with tables rather than people.

Fewer Fields, Verified

Not all information is equal. Forty fields checked beat two hundred fields collected. A document should carry its own provenance layer: who entered it, at which ground, in which over, how many corrections, how many fields missing. Then the reader knows which sentence is heavy. And an editorial pipeline needs a hard stop: if the information-point list at the root is empty, the process halts rather than assembling defaults. A pipeline that never stops never tells the truth.

Contrarian Read

The outside view is that more technology, more models, more layers will fix this. More layers on an empty root only decorate emptiness. A second misreading holds that a failing pipeline is a bad one. It is the opposite: loud failure is health, silent success is the risk. A third assumes structure implies reliability. An article without subheads is visibly unfinished; an article with fourteen subheads is believed first and read second.

Someone will argue that humans are slow and expensive. True. But the expense is at the root, and it is the only expense that buys accuracy. Every other rupee buys visibility. Confusing the two is how hollow skeletons get born.

Looking Ahead

If a board publishes the root log beside the dashboard next domestic season — who entered, how many corrections, which fields missing — will readers behave differently? Will we pay for provenance? The next time you read a statistic, ask who typed it, at which ground, in which over. If you get no answer, the number is not a witness.