The Empty Cell Problem: The Cricket Data Fault That Gets Read as 'No Risk'

A record landed on my desk last week with six of seven columns empty. Title — N/A. Source — N/A. Article type — Unclassified. And at the bottom, where five or six hard facts should have sat, an empty list. Only one field in the whole record was populated: cricket_asia — a tag that says the coverage area is the Asian market, and nothing beyond that.

My first thought was that something had broken. My second thought was worse. If that sheet stays on a dashboard as it is, the next reader will file it as "a low-content article," open the risk matrix, see N/A in every row, and conclude — "no risk found." In cricket data right now, that is the most expensive mistake available, and it does not happen on the field. It happens in a spreadsheet.

Cricket's data work runs in two stages. The first is raw extraction: read the article, pull the facts out, convert each one into an "information point." The definition is strict — an atomic fact that can be checked on its own, carrying a date and a format tag. The second stage is analysis, and it has exactly one condition: every conclusion must trace back to an information point.

Here the second stage came back with nothing, because the first stage gave it nothing. The emptiness was not local. Title, source, type, summary, purpose, information points, entities — all blank at once. The entities column said, in so many words, "identify from the information points above." The list those names were supposed to come from did not exist.

Data scarcity is not new in cricket, which is probably why the blank sheet looked familiar. The IPL sits at the top, with ball-by-ball records going back to 2026 — in the lockdown summer of 2026 I cleaned roughly 1.1 million deliveries across eleven seasons, and only then got a sense of how dense the upper pyramid is. Below it, the Ranji Trophy, Vijay Hazare and Syed Mushtaq Ali sit in half-light. But that is not what happened here. Here the article was never captured at all.

When title, source, type and information points all come back empty together, that is not a thin article — it is an extraction failure. The claim rests on logic, not instinct. Had the source been cricket-adjacent but hollow — a sponsorship release, a dry newsletter — the pipeline would still have returned a title and a source. Their joint disappearance places the failure upstream of classification, at ingestion. A paywall, a JavaScript-rendered page, a truncated fetch, a non-Latin encoding: all four leave that signature.

The emptiness does not stop there. An empty field is never alone. The entities column depended on information points, so it went blank; the risk analysis depended on entities, so it went blank; and now every dimension carries the same sentence — insufficient information. This is contagion from zero, and it cannot be repaired downstream, because the facts needed for the repair never arrived. A model is a story with its seams showing; I map the seams, and here the seam snapped at the first knot.

Time sensitivity also returned as "not assessed," which means we cannot even establish whether the event under discussion has already passed. No analysis can claim freshness without a staleness check.

There is a quieter consequence. The argument Indian cricket keeps having under the label "bench strength" is really an argument about data collection wearing selection's clothes. When the tier below is only partially logged, every model that claims to measure performance there is also quietly setting its own ceiling. When the first stage returns empty, that ceiling is zero.

The sharpest point is this: an empty risk matrix does not mean "no risk" — it means "risk unknown." "No risk detected" and "risk could not be measured" read identically and land in opposite places. Before I trust a table, I ask who assembled it and why. Here the answer is that nobody assembled anything; all of it is still to be assembled.

One more suspicion is worth recording. If this record was produced in an automated batch, its siblings likely carry the same signature. Four fields failing at once can be an accident; the same pattern twice is a pattern. The batch needs auditing before any downstream decision leans on it. What looks like a trend is often a missing column in disguise.

Then comes the temptation that caught me most often early on: filling the blanks from memory. An average, a strike rate, an economy rate — any of them could have been written in. A number without a source is not analysis, it is decoration. Cricket adds a specific hazard here: cross-format contamination. A T20 strike rate is not evidence of Test batting capability, and one format's economy rate cannot speak for another's bowling. So the hard rule holds — no metric enters the table without a source field and an explicit format tag.

Which is why the correct label for this record is "analysis void — input failure." It cannot be archived as a completed piece of work. The archive is the comeback, but only when it records what was missing and why.

The instinct is to blame the source — "there was nothing in it." The evidence points the other way. The more uncomfortable point is that the industry treats speed as competence. The analyst who stops a blank sheet by calling it blank looks slow; the one who fills it and ships looks fast. That asymmetry is what keeps these errors alive.

Building a narrative out of a tag like cricket_asia is easy too. India–Pakistan, the Asia Cup, board politics — three sentences assemble themselves instantly, and all three sound familiar. But a tag describes a coverage area, not a fact. An inference resting on a tag is not analysis; it is a guess waiting for a sponsor. The same family of error is modelling a small sample: a 40-ball phase split is technically computable, but when the foundation is zero, the exercise is decoration.

For the next run, the minimum is this: a title and a named source; the article type; the format stated explicitly, never inferred; at least five information points, each with a date and a format tag; entities — teams, players, boards, competitions; and a publication date so staleness can be measured. Without those, all eight dimensions stay blank no matter how tidy the report looks.

The question is this: of the "complete" analyses sitting in our archives, how many were written on an empty sheet — and who never reopened them?