When the Empty Cell Is the Evidence: Audit Trail of a Null Result

2:17 a.m. A nine-column table on screen, every cell carrying the same sentence: insufficient information, assessment impossible. In the next window a blank report template waited, headline and summary fields empty. Filling it would have taken four minutes. A name, an event, a time — the model would have arranged the rest. I did not fill it. What those four minutes would have produced would have looked like a report and sounded like analysis, with not one verifiable information point inside it. This piece is the case for that refusal, and a confession, because a null result has to be published the same way a miss rate is published. For several years I have run sports analysis in two stages. The first is extraction: pull from the source only those atomic facts that can be checked independently — who, when, at which event, with what result, from which source, on what publication date. Those are the information points, the particles of the analysis. The second stage hangs a nine-dimension frame on them: event and performance, athlete condition, qualification structure, national competitive landscape, rules and anti-doping, team and training system, risk matrix, public expectation, and industry transmission. The rule is one line long: every conclusion carries the address of its evidence, and where there is no address the cell stays empty. Athletics makes that rule unusually hard, because the sport's memory infrastructure is weak. For about seven years, at the meets I have attended in person, my notebook has sat open beside the scoreboard, and the first thing I write is the timing method. In 2026, with stadiums silent, I began digitising hand-timed national sprint records from federations that never kept electronic backups. The lesson from that archive: a record that is not in a ledger is not a record, it is folklore. The ledger you cannot go back and fill is the only real ledger. Now the actual event. The extraction report that was supposed to anchor the analysis came back empty. No title. No source. Article type unclassified. Core viewpoints blank. Information points: zero. Entities not identified. Time sensitivity unassessed. Source quality unjudged. Each of those eight empty cells closes a specific door. Without a title the subject is unknown. Without a source, bias cannot be measured — a federation press release and an independent reporter's copy do not weigh the same. Without a known article type, the analytical lens cannot be chosen; a results report and a process report are not read the same way. Zero information points means the foundation has been pulled from under all nine dimensions. No identified entities means no age curve, no peak window, no injury history can be positioned. No time sensitivity means the claim cannot be placed on the competitive calendar, so there is no way to know whether it was made before or after a specific meet. Two hypotheses remain open, and telling them apart is the most urgent task. One: the source text never existed, or existed and could not be read — the file was truncated at ingestion. Two: the source existed but the extractor could not parse it — unfamiliar layout, mixed language, broken tables. In the first case the fault is in the handoff; in the second, in the machine. The test is simple: open the raw file. An empty file points to ingestion; a full file with an empty output points to extraction. Skipping that test and moving downstream is knife-work in the dark. Insufficient information, assessment impossible — that sentence sounds like an admission of weakness. It is actually the discipline of the frame. Putting it in every one of nine cells means the analyst is quietly admitting there is no foundation. The alternative was worse: inference. Had I written that the athlete's age curve is peaking, someone would ask at what age and on what basis. Had I written that the wind was favourable, the question would be which anemometer, at which meet. Every inference would have bred another, until a whole structure stood on a foundation nobody could ever revisit. The real risk here is not that an article would be bad. The real risk is that it would be believable. An invented name, an invented time, an invented event — once published, those three things enter the record. Next month someone cites it. The year after, it is reconciled against an old federation file. Athletics has no shortage of near-true records: the hand-timed marks that look impossible in the electronic era, with no way to prove whether they ever happened. Four SAF Games 100m titles between 2026 and 2026 are a measurable asset in this region's sporting ledger, but without the timing method, the wind and the meet conditions of that era, comparing them to electronic marks is meaningless. A comparison that hides its own conditions is not a comparison, it is publicity. The same logic runs in reverse. One 60m indoor gold, one universality place at a global championship, does not prove a domestic training pipeline exists, because that result was born outside the system, not inside it. A single live data point does not fill an empty table; it only shows how empty the table is. Three warnings belong here, stacked by level. Highest: the source pipeline returned empty, so downstream analysis is impossible — the only fix is to re-run extraction and verify the raw text survived. Second highest: imagination entering downstream. If any system fills the empty cells to look presentable, fabricated athletes, marks and competitions enter the record; publication should stay closed until real information points exist. Third, medium: silent data loss. An empty result does not always mean an empty source; often it means a broken handoff between stages. That possibility has to stay open, or the actual fault stays invisible forever. Which forces an uncomfortable question back on me. Is refusing to fill the template really honesty, or another form of avoidance? For a perfectionist analyst, saying I will not write is easy — the risk of error drops to zero, and so does the accountability. The reader gets nothing from it. Someone opening the paper in the morning wants to know what happened at the meet. Telling them there is no data is not an answer to their question; it is a confession of our failure. A null-result report earns its value only when it ships with a repair path — who fills the gap, when, and how. Otherwise it is accountability avoidance wearing the mask of humility. A second discomfort: the gap may be a signal rather than an accident. The part of athletics I cover has a defining feature — often there is no source text at all. A meet happens, results are announced, and nothing reaches paper; or one line reaches paper, with no time, no wind, no source. The extractor returns empty, and we blame the machine and stop. The gap is a photograph of a structure with no accountable record-keeping at all. Dropping a name into it erases the real address of the problem. The next step is therefore single. Go back to stage one. Obtain the source text, re-run extraction, and see whether the information-point cell stays empty. I will watch three signals. First, at least one concrete information point in the regenerated output — if it arrives, this entire piece becomes obsolete, and I will say so out loud. Second, the readability of the raw file — that decides where the fault sits. Third, entity extraction — one named entity brings dimension two back to life. If none of the three appears, the verdict is clean: the problem is infrastructure, not journalism. And infrastructure failure cannot be answered with fiction — it is answered with a repair order. That night's template is still empty. I have not deleted it either. An empty cell carries information: it says exactly where our knowing stopped. A sporting culture that cannot keep its own records owes an accounting of that empty cell before anyone tells stories about its stars. The first condition of accounting is to write down what is absent.