Empty Cells, Nine Dimensions: The Silent Failure of Sports Data Pipelines and the Audit That Could Catch It

Two in the morning in a Liverpool flat, rain against the glass, and a report glowing on the laptop: nine dimensions, each with a table beneath it, each table with rows, each row with a verdict and a confidence level — high, medium, low. On first glance it looks like finished work. Someone has touched a subject from nine angles, measured it, and ruled on it.

Then I scrolled.

First cell of the first table: insufficient information. Second cell: not applicable. Third, fourth, fifth — the same. No team, no coach, no formation, no xG in the tactical section. No fee, no contract length, no wage benchmark in the finance section. No charge and no sanction precedent in the governance section. Nine dimensions. Zero information points.

The bigger the tables grew, the emptier the inside became. I counted the words until the terrace started speaking — except there was no terrace here. There was a flawless frame, and inside it a very carefully arranged silence. The loudest silence was not empty; it was full of everything absent.

What happens between two layers

Modern sports analysis runs on two layers. The first holds raw material: headline, source, publication date, article type, one-line summary, author stance, purpose, a list of information points, core viewpoints, named entities, time sensitivity, source quality. The second distributes that material across nine dimensions — tactics, finance, results and public opinion, league landscape, governance, management and dressing room, risk, media narrative, industry transmission.

The problem sat in the first layer. No headline, no source, an empty list of information points, entities underivable, time sensitivity unassessed. The raw material was zero. What does the second layer do then? There is only one honest answer: nothing.

That is where the trap opens. A frame can still be erected on zero material — nine headings, twenty tables, thirty cells, each politely marked not applicable. The frame looks exactly as heavy as a complete report. A report's weight is not measured by how many tables it carries but by the density of its verdicts.

Why this silence is familiar

May 2026, Anfield, final day of the Premier League season. Liverpool 3-0 Middlesbrough, Wijnaldum scoring in first-half stoppage time, Coutinho and Lallana sealing it. I watched from the Kop on my sister's season ticket and wrote nine hundred words on the phone on the 26 bus home — about the noise, not the goals.

What I learned that night came back to me looking at this report: put the scoreline in the first paragraph and the story dies. The real information lives inside the sound, not the scoreboard.

Nine years of watching matches has taught me this above all — when a crowd goes quiet, one of two things is happening. Something has occurred, or nothing has. Both are information. They cannot be written the same way. In June 2026, when football returned without its choir, I stood outside Anfield with roughly two hundred supporters who could not go in, phones held up, singing to nobody, and refused to file a match report. The phone buzzed; the crowd answered in ninety-minute paragraphs. That night there was no crowd, no buzz, no answer — only an empty ground and a question: how do you report what did not happen?

That question now sits in front of a software pipeline. And software has a habit of walking around the question.

Nine dimensions, nine kinds of absence

The tactical dimension depends most heavily on what the source explicitly states. With no team, coach, formation or fixture referenced, tactical discussion has no anchor. Without a single quantitative field — xG, xA, PPDA, possession, pass completion — even a reconstructed tactical claim could not be assigned a defensible confidence level. Where there are no numbers, writing a confidence level turns the number into decoration.

Finance needs three minimum inputs: a fee, a contract length, a wage benchmark. None exists. The panic-premium test — overpaying under bidding pressure or public noise — cannot be applied when there is no transaction to test.

Results and public opinion need a table position, a form string, a fixture list. The data-results divergence test needs at least one process metric and one results sequence. Managerial pressure needs a manager's name. There is none.

League landscape needs a league. Without one, the map from title contenders to relegation zone cannot be drawn, and tier positioning without club identity and table context is meaningless.

Governance delivers the most useful lesson here, and not about itself. If the article type is an ordinary match report, this dimension is not 'insufficient information' but 'not applicable' — most match reports carry zero regulatory content. Conflating 'we have no data' with 'there is no question' is the most common error in modern analysis.

Management and dressing room would be the highest-value dimension had the source been a player profile or interview. The triad of age curve, contract year and availability is the spine of the contract-year-breakout and ageing-core risk lenses. With no named individual, the triad has no subject.

Risk has six categories — sporting, financial, personnel, rules, public opinion, systemic — and all six lie dormant because none has a subject. But one risk is unambiguously alive: analytical-integrity risk. Invented football content presented against an empty input is the single most damaging failure mode available.

The media narrative dimension holds a small tell. A blank source field is itself diagnostic: the article entered the pipeline without a URL or publication metadata, and that gap typically breaks headline extraction and entity recognition downstream. One empty cell can empty ten more. This is not a series of failures but a single chain.

Industry transmission requires an identifiable event — a transfer, a broadcast deal, a governance change. Multi-club ownership, agent networks and solidarity mechanisms are all event-driven. With no event, any arrow drawn is a fabricated causal chain.

Failure, or genuine emptiness?

Two possibilities, with entirely different treatments.

One: the article genuinely had no tactical content — a four-line transfer brief, a social post. Then the defect is not in the material but in the frame: nine dimensions were never appropriate.

Two: the pipeline failed. Paywall, JavaScript-rendered page, broken encoding, truncated hand-off. A missing source field alongside an empty information list points strongly here, because even a content-free article usually leaves a headline and a URL behind.

The only way to tell is to inspect the raw fetch output and parser logs. And if more than one empty payload appears in the same batch, the defect is systemic. The null-rate audit matters more than any single report, because one empty report is visible and ten are not.

The real danger is silent loss. A pipeline under pressure to look complete has no incentive to say what it lacks — the opposite, in fact.

Technology that will not let silence hide

Machine verification is not new to football. Goal-line technology arrived in the Premier League in 2026-14, VAR in 2026-20, semi-automated offside later still. Whether the ball crossed the line, whether a shoulder was offside, is no longer left to memory.

One thing remains entirely memory-based: the record of how a decision was reached. Why the penalty, who said what, which data arrived when. No continuous, tamper-evident account exists.

For analysis pipelines the gap is sharper. An empty payload written into a hash-chained, append-only log — fetch, parse, extract, analyse, each step's output hashed against the previous — cannot hide. Delete a step and the chain breaks, and a broken chain is visible.

This is the same principle football's own economy already runs on. FIFA's solidarity mechanism distributes five per cent of transfer compensation to clubs that trained a player between the ages of 12 and 23, per the FIFA Regulations on the Status and Transfer of Players. That accounting is possible because transfer events are documented. No such document exists for where an analysis report came from, which source it stood on, or who verified its verdicts.

Goal-line technology verified the ball. The next step is verifying the words — which claim came from which source, who checked it, and what was missing. Anyone who works transfer rumours knows the truth of it: a transfer is not a transaction; it is a rumour learning to breathe. The only way to still a rumour's breathing is a chain of evidence. The Premier League's Profit and Sustainability Rules have produced points deductions as sanction precedent — enforceable because the accounts were recorded. Analysis has no such record, so analysis has no such enforcement.

Effort numbers and analysis numbers

Distance covered and high-intensity sprints get packaged as effort metrics. A player who runs ten kilometres in all the wrong places produces a beautiful number and no work. Data analysis has the same trap. Nine dimensions, twenty-five tables, ninety-nine cells looks like effort. If every cell reads insufficient information, the number is beautiful and the work is zero. The two deceptions are the same deception: show proof of labour, never proof of result.

The reversal: completeness is camouflage

We assume more dimensions means more reliability. It is the other way round. A two-dimension report containing one real verdict is more dangerously honest than a nine-dimension report containing none — and the nine-dimension empty report is more dangerously credible. The size of the frame stands in for the truth that is missing. A reader who sees nine tables does not ask what is inside them.

The second reversal is less comfortable. We assume an empty output means failure. Here there were two failures, not one: a probable pipeline defect, and the insistence of a frame that forces every article into nine dimensions. A four-line transfer brief does not want nine dimensions. Force it, and you build a machine that goes looking for what is absent and counts itself broken when it finds nothing.

The third reversal is least comfortable of all. These pipelines were built to remove human judgement, so that verdicts would arrive without bias. What kept this run honest was precisely the thing we tried to remove: a person who read the empty cells and refused to fill them.

12 June 2026, Parken, Denmark versus Finland. In the 43rd minute Christian Eriksen collapsed. I was in a Liverpool pub; two hundred strangers went silent, then began checking on one another, sharing numbers, nobody leaving. My piece the next morning was about the Danish players' shield, not the result. That night I set myself a rule that has only hardened since: what is not known cannot be written as inference. The piece I wrote after standing outside Anfield with two hundred locked-out supporters — The Loudest Silence — carried the same lesson. To write a crowd that could not go in, you must first admit what you did not see.

Outlook

Every street has a pulse; mine learned to hold its breath. This report is that held breath — a system admitting ignorance while stacking nine tables beside the admission so nobody notices.

The most valuable measurement in sports media next season will not be how much a system produced. It will be how honestly it reported what it did not have. An empty cell written into an append-only, hash-chained log, with a clear status flag — insufficient input — stops being an embarrassment and becomes evidence.

Which leaves one question. If a system can build a complete nine-dimension frame from zero material, how will we tell its zero-per-cent report from its best one?

For ninety minutes, the whole neighbourhood becomes a single lung. Tonight the neighbourhood is quiet, and we are about to file that as analysis.