The Empty Data File and the 4,000-Word Analysis: The Failure Sits at Ingestion

2:47 in the morning in Chiang Mai, and a 4,000-word document landed in my inbox. The header read "Tier-2 Deep Analysis". Inside were eight sections, each split into tables, each table with columns for "Assessment", "Benchmark", "Notes". Every cell carried the same phrase: "N/A — insufficient information". No fighter's name. No event's name. No date. A document nearly as long as a book chapter, and it was about no one.

The sender was a young editor I know from a data project at the 700th Anniversary Stadium. He wrote: "Could you take a look? We need to publish tonight." I read it through. Then I read it again. I read it to see whether, anywhere in those pages, someone had dared to invent a name.

No one had. At least the writer kept discipline.

To understand why such a document exists, look at the pipeline that now produces sports coverage. An analysis piece passes through three tiers: raw ingestion, event extraction, deep analysis. The middle tier is the fragile one. Feed it an empty payload — blank headline, blank source, blank event list — and every tier downstream has two choices: stop and report the error, or keep running and fill the gaps with inference.

That document chose the first. It printed the entire analytical frame: technical matchup, condition, organisational landscape, revenue model, rules and governance, health risk, media narrative. Eight dimensions. At every position it wrote "insufficient information to assess". This is unusual behaviour in newsrooms. We are used to desks inventing more. Few are used to a desk refusing to invent.

But the story does not end there. The ingestion stage had died hours earlier, and nobody noticed.

An empty file is a missing document, and a missing document is more dangerous than a bad one.

Here is the difference. When a source is weak, an editor can cross-check, doubt, pick up the phone. When a source is empty, there is nothing to cross-check — and professional instinct switches instantly into fill-the-gap mode. That is why empty files tend to produce the longest articles. Data does not lie, but the people who read it do.

I have seen this at a smaller scale. In 2026, analysing 40 Thai 400m hurdlers, I found the group was running at 78% of international benchmarks because they skipped the acceleration phase over the first three steps. That figure came from video data you could scrub frame by frame. But if the cameras had failed that day, and I still had to file a report, what would I have written? I would have written about "fighting spirit". I would have written about the things nobody can measure, because nobody can refute them.

In the summer of 2026, I watched Jamaica's 4x100m relay team go out in the heats with 38.83 seconds. Colleagues blamed Usain Bolt's retirement. I went looking for baton-change sessions: Jamaica trained twice a week, Britain five times. The conclusion sat with no individual at all. We used to think speed belonged to individuals, until the system collapsed.

Now apply that logic to a fight. Suppose ingestion fails the week of a major card. No fighter names, no weight classes, no head-to-head record. The analysis tier can still produce 4,000 words describing "finishing ability" and "weight-cut risk". It sounds expert. Not one fact can be verified.

This is the point I want to pin down. In a transfer window, the noise does not come from obvious lies. It comes from stories that are formally correct and substantively empty. A line reading "sources close to the deal say talks are ongoing" has perfect grammar. It is missing one thing: evidence.

Professional instinct says an empty file is harmless. No information means no errors. I think the reverse is true: an empty file is the most dangerous object in a newsroom, because it does not incriminate itself. A wrong number gets caught. A wrong name gets caught. A blank space does not. Blank space invites.

At 59, I have written enough to know the biggest temptation is not inventing events. The biggest temptation is inventing the explanation for an event that never happened. A fighter never falls from lack of strength, but from the structure around them cracking first.

There is a signal I learned over the years. When an analysis reads too smoothly, with no line marked "unverified", no line marked "source unknown" — that is when to stop. Honest analysis always has a crack in it. The document from 2:47 had eighty such cells. It was ugly. It was technically correct.

That night I replied with one line: "Don't publish. Re-run ingestion first." The next morning the editor called back: the source article had never been loaded into the system — a fetch failure. Within two hours a real document appeared, with names, dates and numbers.

When the stands are empty, you hear the breathing of the contest more clearly. An empty file is the same. It says nothing, and precisely because of that it forces the reader to decide who they are: someone hunting data, or someone hunting a story good enough to replace it. Data draws the map, but memory is the terrain.

Sports will keep adding pipelines, tiers and automated documents. The writer's job is not to fill the blanks, but to mark them.