When the Result Sheet Came Back Empty: Auditing 'Information Not Available' in Athletics

11:47 PM, Miami. A nine-dimension athletics analysis template sat open on the laptop. Every field was filled. Every field said the same word: N/A. Title N/A. Source N/A. Athlete N/A. Mark N/A. Wind reading N/A. Time sensitivity N/A. Twenty-one rows, and every row answered the same way — insufficient information.

Since I was sixteen, I have had one habit: read the official result first, then re-derive it by hand. I hardened that habit in the summer of 2026 at Kalinga Stadium in Bhubaneswar, when the official split sheet for the men's 4x400m relay and the video footage disagreed by three-tenths of a second on the second leg. That night in Miami, for the first time, there was nothing left to re-derive. And in that moment one thing became clear: an empty result sheet is still a document, and it deserves the same rigour as a filled record.

The four-layer paper stack

In athletics, a performance never arrives in a single document. It comes through four layers. The first is the entry list — who runs, in which lane, in which heat. The second is the official result sheet: placing, time or distance, reaction time. The third is the split sheet and the wind reading — every relay leg, every 100m segment, and the wind speed in metres per second attached to every jump and throw. The fourth is the selection rulebook: qualifying standards, world ranking points, and the national federation's own selection document.

My method has always been the same. State the claim. Re-derive it from the source layer. Then publish the conclusion with its confidence level and my assumptions exposed. Working from a US timezone, those layers are often all I have — the federation PDF and the results feed. There is no colleague standing in the stadium to tell me whether the baton actually changed hands cleanly on the second leg.

So I keep a standing checklist of known failure points: hand-timed meets, where electronic and stopwatch times drift by tenths; missing wind-reading rows; late entry-list edits; and DNS or DNF reclassification after the fact. The checklist is not a hobby. It is insurance.

That checklist is why one thing about that night was immediately jarring. Any published athletics item — a one-line news brief or a ten-thousand-word profile — carries at least one of four things: an athlete's name, an event name, a mark, or a venue. All four going missing at once is abnormal. So abnormal that it stops being a story and becomes a signal.

N/A is not zero

The first error to clear is the most expensive one in any data report. N/A does not mean "no risk". N/A means "not assessed". The gap is not semantic. It decides whether you are treating an athlete as clean or as unknown.

At Tokyo 2026 I built a live javelin dashboard for a broadcaster. Every throw required three fields: approach speed, release angle, flight time. When a throw's wind reading did not appear on screen, the production team's first reflex was to leave the field blank. The second reflex was to carry the previous throw's reading forward. The second reflex is the dangerous one. Once you carry a reading forward, the viewer never learns they saw a wrong number.

That is why I added a differently coloured row to that dashboard — "unconfirmed". In the quick-reference guide I built for commentators, that row did the most work. The guide later became the team's standard, and the part-time contract followed. The lesson was simple: between speed and clarity, clarity comes first.

Athletics has five classes of unconfirmed data, and each needs its own protocol. First, hand-timed meets, where the electronic conversion rule must be applied and stated openly. Second, wind-assisted marks, where anything above +2.0 metres per second does not count for records. Third, altitude venues, where roughly 1000m and above helps sprints and jumps and penalises endurance events. Fourth, DNS and DNF reclassification. Fifth, the carbon-plate shoe and fast-track dividend, which makes any comparison false if it is not deducted.

Not one of those five classes appeared in the empty template. That does not mean none of those risks existed. It means the risk was never measured.

The record with a hole you could drive a truck through

In 2026, during the pandemic pause, I started digitising historical Indian athletics records. It was dull work, which is why nobody had done it. Three months, twelve sources and interviews with two officials later, an error surfaced in the official time of the women's 4x400m relay at the 2026 Commonwealth Games. That national record had a hole you could drive a truck through.

The authority's first reflex was denial. That was expected, and I had left room for it. The report carried two sections — "limitations" and "response to critics" — a habit that came directly out of that episode. A few months later, the record was quietly corrected.

That experience taught me something that applies to every audit: news of a correction is always smaller than the reason for it. When a time changes, the important question is not what the new number is. The important question is who the old number benefited. Which athlete got a medal off the wrong time, which one got a qualifying slot, and which one was denied both.

The answer is often deflating. Sometimes a reshuffled podium is purely clerical — nothing changes in the athlete's career. My own definition is blunt for that case: it is a note, not a story. And presenting a note as a story is my error, however clean the audit.

The model I built at seventeen, and the grade I gave myself

At Gold Coast in 2026, Neeraj Chopra won Commonwealth gold with 86.47m. I was seventeen, working from frame-by-frame video and open-source biomechanics. I built a model of his throw: release angle around 34 degrees, approach speed around 8.2 metres per second. Comparing him to global elites, I wrote that he would cross 90 metres within three years.

At seventeen, I modelled a javelin throw and saw the future — I said that then, not now. An athletics correspondent at The Hindu cited the model in a feature.

Now the grade. Tokyo 2026 gold came at 87.58m. My point estimate missed. The lifetime best of 89.94m arrived in 2026 — six centimetres short of the threshold. Paris 2026 silver came at 89.45m.

The lesson here is not about the model. It is about how the model was published. I printed a point. I should have printed an interval — something like "88 to 91 metres by 2026, midpoint 89.5, and this estimate cannot see the three things outside the model: injury history, a gust in the final, one fouled attempt."

That is model confidence bleed — a clean spreadsheet convinces you a probability has become a forecast. A javelin final is exactly where biomechanics, nerves and weather break the fit together. My rule is now strict: publish the interval, not the point estimate. And state what the model cannot see before the meet, not after.

The same rule extends to that empty template. When you have not a single information point, the only honest output is N/A across every dimension. And it should stay that way. It should not be compressed into a summary reading "no risk identified".

Absence of evidence is not evidence of compliance

Here is the biggest ethical trap. If a report carries no anti-doping red flag, the reader's brain automatically reads "clean". The report actually says no check was performed.

In athletics this distinction carries extra weight, because the sport's record system runs on long memory. Samples are stored for ten years, and retrospective testing can redistribute medals. The Athlete Biological Passport tracks blood and steroid markers across seasons. The whereabouts rule says three missed tests in twelve months become a violation. The clearance that looks clean today can become tomorrow's document.

There is a direct consequence for my reporting. When I write that no anti-doping signal exists around an athlete, I also write which signal I was looking for and why it was not found. If I cannot write that sentence, I delete the sentence.

Where the pipeline broke

Now the forensic read on that empty output. Nine dimensions, twenty-one rows, and only one field populated: domain label, athletics.

This matters because it locates the break. The title and source layer comes first. Entity extraction — pulling out athletes, coaches, competitions — comes after. When both title and source are empty, and the entity list is empty too, the most probable reading is that the failure happened at the title layer, before entity extraction. The document never made it inside.

The second possibility is that the document arrived and was genuinely content-free. I rate that lower, because no athletics item clears editorial review without a name, an event, a mark and a venue. All four vanishing together points at a technical failure.

A third thing falls out of the empty output, and it concerns the domain label itself. When "athletics" is populated but the document contains not a single athletics entity, the question becomes where that label came from. Inferred, or assigned. The distinction looks small and is not, because the entire nine-dimension framework rests on that label. If the label is wrong, the framework will work hard in the wrong direction.

The contrarian angle: a filled template is more dangerous than an empty one

Here is the most useful part of the whole episode, and it runs the other way.

An empty template makes noise. It looks so uncomfortable that any editor stops it immediately. Twenty-one rows of N/A are a kind of alarm.

The dangerous object is the filled template. When all twenty-one fields are populated, when the numbers are clean and the tables are tidy, nobody feels the need to ask which layer those numbers came from and what assumptions they stand on. And the real problem in athletics media is not a shortage of data; it is confidence without stated method.

I have seen season-best lists with no note anywhere about which marks were wind-legal. The list looked clean. The conclusion was wrong. I have seen graphs drawing a smooth junior-to-senior progression with the injury year silently absent from the line. The line was smooth. The career was not.

So my recommendation runs backwards. If forced to choose between an empty sheet marked N/A in every field and a filled sheet with no methodology recorded, I take the empty one. The first tells you what is unknown. The second gives you false confidence, and false confidence costs more than a gap.

What to watch next

The most solid result of this episode is that the empty output is itself a diagnosis. It told us exactly where the supply chain broke — at the title and source layer, before entity extraction. That points to a repair path before the next batch run, and the path is cheap.

I will track four things. First, source-document recoverability — whether the original item can be re-ingested, and whether a title and at least one information point come back with it. Second, extraction integrity — information-point count against document length. Third, domain-label reliability. Fourth, date restoration, because without a date neither heat cycle nor peaking window can be fixed.

One question for the reader, and it applies to my own work too. When a record comes back empty in front of you, do you close it with "information not available", or do you read it with the same seriousness you reserve for a filled record? For me the answer is fixed — audit, verify, then speak. And when verification is impossible, the most important thing to say is that verification did not happen.