Inside a Combat-Sports Report With No Data: Eight Dimensions and a Void

Inside a Combat-Sports Report With No Data: Eight Dimensions and a Void

2:14 a.m., at my desk in Incheon. The file opens with eight analysis tabs. Each tab has its own table, its own assessment column, its own conclusions section, and a line for risk level. I scroll from top to bottom, slowly, the way someone who has spent thirteen years reading data tables before reading prose tends to scroll.

Style matchup table: no data. Fighter condition table: no data. Revenue structure table: no data. Health risk table: no data.

Then the section everything else must hang from: the list of information points. Completely empty. Not a single numbered line.

A long report, fully framed, and hollow. Eight analytical dimensions, not one grain of data. Numbers do not lie; they simply wait for us to read them correctly. Here the number is zero, and it speaks very loudly.

That file cannot be published. It also cannot be deleted. It records the one thing combat-sports journalism rarely agrees to put in writing: its own limit.

To understand why such a file exists, you have to know what it passed through.

Vietnamese sports pages use one word for many things: martial arts. Under that word sit amateur and professional boxing, MMA, Muay Thai, kickboxing, sanda, and also vovinam and traditional Vietnamese martial arts — disciplines scored by a judging panel rather than by a win-loss record. One word, three incompatible logics, usually sharing a single section, a single keyword, a single breaking-news line.

The content classification system the newsroom uses runs into exactly the same problem. It has one label: martial arts. No subdivision. No distinction between professional combat and forms performance.

Then an article enters the pipeline. Two tiers: tier one breaks the article into numbered information points; tier two applies domain expertise to those points. Tier one returned a shell. No headline. No source. No date. No information points. No named entity. Article type: unclassified.

Tier two has eight dimensions to work with and nothing to work with.

The eight are: competition and technical-tactical analysis, fighter condition and career longevity, event and organisational landscape, business model and market, rules and governance-compliance, health and career risk, public narrative and market expectation, and finally industry transmission.

That is where the craft is tested. Every empty cell is an invitation to fill it in. And the best person at filling things in is always the writer.

The combat-sports news cycle in Vietnam during peak months has its own rhythm: bout announcements, weigh-in news, injury news, agent news, contract news. Each thread runs through dozens of outlets within hours. Most of those outlets cite each other, and if you trace far enough back, they usually resolve to one original sentence. That is why I never count articles to measure reliability. I count original sources.

Before believing a rumour, I count the phases. That habit formed in 2026, sitting in a review room during the World Cup in Russia. I noticed something odd: a defender repeatedly pushing high on the right flank between the 60th and 75th minutes. I spent three days rewinding the group stage, counted eight repetitions of the same position, and only then wrote. Three days of tape, and then a detail reveals itself. Had I written on the first night, I might have been right, but I could not have proven I was right.

Back to the empty file.

A blank field and a field marked no data are different things, and professional data readers know it. A blank field usually means: nobody asked. A field marked no data usually means: somebody asked, and there was no answer. In this file, both the headline and the source read no data rather than sitting blank. That is the trace of an extractor that ran and returned null values, not of a document that never had a headline. My confidence in that inference is only medium, because it rests on field formatting, not on content.

There are two hypotheses, and I keep them separate. First: the fetch stage broke — page load failed, parsing failed, and everything else is downstream of a single upstream break. Second: the source document was never bout news at all — possibly a cultural piece, a historical piece, a promotional piece — so the extractor found no competitive entities to pull. Confidence in the second is low, because it is a guess about the pipeline, not about the article.

Both hypotheses lead to the same operational conclusion: stop, return, re-extract. Neither branch leads to sitting down and writing.

The martial-arts label is too broad, and breadth here produces category errors — applying one discipline's yardstick to another.

Boxing and MMA run on win-loss logic: records, finish rates, style matchups, quality of opposition. Sanda is also contested and also scored, but permits throws, leg catches and ring-outs — the weighting diverges sharply from professional kickboxing. Vovinam and traditional Vietnamese martial arts include forms and performance, scored on difficulty plus execution quality, with no win-loss logic at all.

Those three systems share no yardstick. Applying a knockout rate to a forms athlete is a category error. Applying takedown defence to a sanda routine is a category error. And above all: writing that a fighter has a 62 per cent finish rate, when we have never established who that fighter is, is fabrication, not analysis.

On Vietnamese sports pages, these three tracks sit side by side every day. A piece on a professional boxing night in Ho Chi Minh City. Directly below, a national vovinam event. Below that, news of an MMA fighter signing a contract. Same section, same keyword, three sets of evaluation criteria that cannot be swapped.

Seasoned readers tell the three tracks apart by instinct. Classification systems do not, unless someone applies a sub-label at the entry point. A coarse label does not make an article wrong. It makes the article impossible to get right.

Inside an empty file, every sentence can be written and every sentence sounds plausible. That is precisely the danger.

The first assumption anyone can make: the article is about a fight. It feels natural, because most combat content is about fights. The file does not confirm it. The second: the subject is a professional fighter. Also natural, also unconfirmed — the file names nobody. The third: there are records to quote. The fourth: there is money in the story — contracts, purses, broadcast fees. All four assumptions are reasonable, and none is confirmed by a single line.

The fabrication ladder works like this. Lowest rung: a sentence with no number. Above it: a sentence with a number that has no source. Above that: a sentence with a number and a source — but only one source. Top rung: a sentence with a number, a source, and cross-checking from at least two independent directions. The craft does not forbid writing on the lower rungs. It forbids calling a lower rung the top one.

From my own experience watching matches, a good analytical piece rarely needs many numbers. It needs the right number, in the right place, with its origin stated. Three sourced numbers are worth more than thirty unsourced ones.

There is one dimension I always leave for last, because it is the easiest to misread: health and career risk.

In this file, the health risk table is entirely empty. Six rows: cumulative brain injury, weight-cut incidents, orthopaedic injury, post-retirement security, psychological safety, systemic risk. All six show no data.

What I want to state plainly: a blank cell in a risk table is not a certificate of safety. It is a gap. Weight cutting is the highest-frequency acute hazard in combat sports, and it can only be screened with walk-around weight, weigh-in weight, and a history of missed cuts. Without those three, there is no conclusion. Cumulative brain injury can only be screened with bout counts, round counts, and knockout-against history. Without them, there is none.

When the stands are empty, the pitch starts telling the truth. In 2026, during the no-spectator period, I stayed for training sessions nobody bothered to attend. Young players were afraid of contact, and that fear showed in a very small detail: they pushed the ball wide far more often instead of driving through the middle. No cameras, no crowd, and the fear was still there, still leaving traces. But if I had not been in the building that day, I would have had no right to write a line about it.

An empty data file has no traces to read. No presence at the scene, and no tape either.

I choose to trust the tape, because tape has no emotions. But when there is no tape, I choose silence.

So what does an article minimally need for the analysis tier to run without fabricating?

It needs a headline, an outlet, and a publication date — those three establish source quality and freshness. It needs three to five numbered information points: the specific factual claims the article makes. It needs at least one named entity: a fighter, a coach, a promotion, a federation. It needs a sub-domain label, so nobody grades a forms routine with knockout metrics. And if it is bout news, it needs the bout or event named, along with any record, ranking or result quoted inside.

Five minimum inputs. Miss one and the chain breaks.

Here I want to push against the current, because the industry's normal reflex is the opposite.

Readers trust a report with eight sections more than a wire line with one sentence. Section count becomes a proxy for rigour, and it is a poor proxy. Eight empty sections are still eight empty sections. Format completeness can mask a content void, and it masks it very well, because format is what we see first.

My profession is usually measured by output. But in a dense rumour cycle, the highest-information act is sometimes an unpublished piece. A newsroom willing to publish the line cannot be analysed for lack of data is doing something harder than guessing: it is holding the baseline where it is. Lowering it once is fine. Lowering it weekly and the floor disappears, and nobody remembers it was ever there.

The null result here is not a result about a fighter, an event or an organisation. It is a result about the machine. The only real finding in that file is this: the extraction stage failed or was never populated, and the report must be returned for re-extraction before anyone analyses further. That is a diagnostic conclusion, not a domain conclusion. But it is a conclusion, and it is correct.

There are four signals I will track over the coming days, written down so anyone can verify them rather than trust me.

First: whether the information-point list gets re-extracted from the source document. The trigger is simple — at least one numbered information point appears. Then all eight dimensions open up.

Second: whether the system can apply a sub-label separating professional combat from sanda and from forms. With a sub-label, we know whether the competition, rules and health dimensions apply or must be swapped for a different criteria set.

Third: whether source provenance is recovered — outlet name, URL, date. With a source, source quality and author stance become assessable.

Fourth: the entity list. One name appears, and the first four dimensions immediately have something to grip.

I am leaving that file in the folder. Not deleting it, not publishing it. It is a specimen. It reminds me that whenever I sit down in front of a combat-sports section packed with breaking news, the first question is not what happened, but who is speaking, and how many times they counted.

Football does not rush; neither does the writer. Combat sports rush even less. A strike takes two-tenths of a second, but to know whether it was real, I need three days.

Tonight I will open the file again. If it is still empty, I will close it and wait. If a single numbered line appears, I will start counting from that line.