The Empty Data Table and the Classification Trap of Vietnamese Martial Arts

One January evening, I sat between two screens. On the left was a recording of a sanda bout at the national championship. On the right was a recording of a professional boxing bout at the Nguyen Du arena. The same roundhouse kick, nearly the same contact angle, but two referees recorded two different scores. I flipped back through my notes and noticed something more troubling: I had no way to compare these two bouts using a single yardstick.

The problem was not the fighters. The problem was that I had called both of them "martial arts" without asking one simple question first: which kind of martial art is this?

In Vietnam, the word "vo" covers at least three different competition systems. Traditional martial arts compete through forms and sparring, scored on a technical scale. Sanda competes in full-contact bouts on a platform, scored by the number of valid strikes and the impact of contact. Professional boxing scores round by round, with a mandatory 10-point system. Three systems, three logics, three ways of reading an identical movement.

I once tried to merge them all into a single data table, labelled simply "martial arts". The result was a mess. A sanda fighter with seven straight wins looked identical to a boxer with seven straight wins, until I looked at how they won. One won by controlling distance and landing on the third beat. The other won by absorbing punishment in the opening rounds and exploding in the fourth. The same "7-0" column told two completely different stories.

That was when I understood why an empty data table is more dangerous than I had assumed.

Last March, I re-ran the eight-dimension analysis I use for combat sports: style, condition, organisation, business, rules, health, narrative, and transmission chain. All eight dimensions returned a single line: insufficient information. Not because I lacked raw data. I had thousands of records. I lacked the thing that comes before raw data, a subject-classification layer.

This lesson is not new. Based on my experience covering bouts across all three systems, I learned that you cannot compare a 1,500m runner with a marathon runner using time alone. When I wrote about Nguyen Thi Oanh accelerating in the final lap of the 1,500m, I could write it because I knew how that distance operates. Different distances mean different pacing logic, different heart rates, different tactics. Placing two athletes side by side in one ranking table is a methodological error, not a numerical one.

Vietnamese martial arts sit at exactly that point. We have three, four, sometimes five competition systems running in parallel, but often only one shared data label. That shared label creates the impression that everything has been standardised. It creates the impression that a 60kg fighter at a national championship and a 60kg fighter on an international stage are two points on the same line.

They are not.

Data never shouts, but it will repeat itself until you are willing to listen. What I heard from my messy data table was not the voice of fighters. It was the voice of a system error repeating itself every time I tried to compare two things that did not belong to the same category.

I call it the classification trap.

When an authority groups traditional martial arts, sanda, and boxing under a single "martial arts" heading, it is not wrong administratively. But when a journalist like me takes that heading and builds analysis on it, I am running the logic of one discipline over the data of another. For MMA or boxing, I need finish rates, knockout counts, and takedown-defense efficiency. For traditional martial arts, I need the difficulty score of a routine, the stability of a stance, the margin of error in each movement. Applying one discipline's toolkit to another is not a minor discrepancy. It is a systematic bias, and it is stable over time, meaning it will not disappear on its own.

This is where I want to state plainly the opposite of a common belief among analysts.

The common belief is that more data is better. I believed it for years. But a large dataset that is not classified at the root layer is more dangerous than a small dataset correctly labelled. A large dataset creates a false sense of certainty. It allows a writer to build a fluent, number-heavy conclusion running on the wrong axis. And because it reads fluently, nobody questions it.

Sports journalists in Vietnam, especially in martial arts, are often placed in situations where they must write quickly. I understand that pressure. But precisely for that reason, the classification layer must come first. Three layers of verification are not meant to find the truth, but to calculate how many times the truth can survive being distorted. If the first layer assigns the wrong subject class, the next two are merely confirming a mistake more carefully.

The fix is not complicated. When facing a martial arts dataset, the first task is not to run the numbers. It is to ask: how does this discipline score, how does it win, how does it lose. Only when I can answer those three questions do I touch the keyboard. For traditional martial arts, I read the technical scoring sheet first. For sanda, I read the valid-strike count and control time. For professional boxing, I read the three judges' scorecards before looking at any other metric.

The difference between a data label that is empty and a data label that is correct is the difference between knowing and not knowing. An empty table honestly labelled empty will remind me to go find a source. An empty table labelled "martial arts" will make me believe I already have the answer.

A tank's road wheel never stands out in any photograph, but it decides which mud pit the vehicle can cross. The subject-classification layer is that wheel. It sits out of sight, nobody praises it, but every sports conclusion passes over it.

I still keep those two screens on every evening. Now, before I record scores, I record category. Before I compare, I classify. The work is slower, but at least I know what I am comparing with what.