The Empty Data Sheet and the Decision Not to Write

A morning in Incheon, I opened the analysis file passed over from the input-processing stage and found a blank sheet. No fighter's name. No event name. No ruleset. Not a single figure. All that stayed in the corner was a domain label reading "martial_arts", an unfilled classification box, and an information list exactly zero lines long. My editor asked whether I could still file a weekend preview. I read the file a fourth time in three days, then answered: not yet.

Three days of rewatching tape, and then one detail reveals itself. This time the detail that surfaced was emptiness. The lines below record why I refused to write, and what an analytical workflow must hold before a writer is permitted to open the document.

In my trade, a martial arts analysis comes to life through two layers. The first layer reads the source article and pulls out discrete information points: fighter names, events, organisations, dates, specific figures, together with article classification, author stance, article purpose and time sensitivity. The second layer takes those information points and builds eight analytical dimensions: tactical matchup, athlete condition and career age, organisational and event landscape, business model and market, ruleset and compliance, health and career-lifecycle risk, media narrative and market expectation, and the transmission path of the whole industry.

The second layer is only worth anything when the first layer leaves traces. This time, the first layer returned zero. The information-point list was completely empty. Not one entity was extracted. The article type was left unclassified. Author stance was blank. Article purpose was blank. The source title and source publication were both absent from the file. Time sensitivity had never been assessed, and source quality had never been graded.

My own experience of watching matches teaches one simple thing: a pattern only reveals itself when it repeats. During the 2026 World Cup finals, tracking the national team, I noticed a defender repeatedly pushing forward on the right flank between the 60th and 75th minutes. It took me three days of rewatching group-stage matches, counting eight repetitions at the same position, before I dared to write. One appearance is coincidence. Eight appearances are a designed drill.

For a data person, a blank file is the most comfortable situation in which to refuse. For a person who has to file, it is the most uncomfortable situation to explain. I choose to trust the tape, because tape has no emotions. Here there was no tape at all.

An information point is the smallest unit of truth a news article leaves behind after being deconstructed: a name, a date, a number, a conclusion. All eight analytical dimensions in the second layer are designed to cling to those points. When the count is zero, every analytical sentence produced is nothing more than speculation dressed in technical vocabulary. Without information points, there is no floor on which to place any judgement.

Entities also vanished from the file. The system extracted no fighter, no coach, no gym, no event, no organisation and no governing body. The transmission chain of the martial arts industry normally has three links: upstream gyms and talent supply, midstream organisations and events, downstream broadcast, data and consumer markets. A transmission chain needs at least one actor to serve as the origin of the shock. No actor, no chain.

The most worrying part sits at the subject-classification step. That step must be taken before anything else, because the three subject classes in martial arts operate on three entirely different logics.

The modern competitive class includes MMA, professional boxing, kickboxing, Muay Thai and grappling. The correct framework here is win-loss record, finish rate, takedown defence efficiency, striking accuracy, submission frequency, quality of opposition and weight-cut condition.

The traditional and performance class, taolu for instance, demands a different framework: difficulty score, performance score, judging practice and consistency across competitions. Win count says nothing in this class.

Sanda is the third class, where combat and performance coexist under a ruleset of its own.

Choosing the wrong framework produces systematically wrong conclusions. Apply win-loss logic to a taolu article and the analysis sounds very confident and is wrong at the root. Apply difficulty-score logic to an MMA article and the same happens, only in the other direction. The domain label in the file reads only "martial_arts", and it does not answer which class the subject belongs to.

The applicable ruleset is equally undetermined: the Unified Rules of MMA, professional boxing rules, K-1 or Glory rules, Muay Thai rules, IBJJF or ADCC rules, sanda rules, or a taolu scoring scale. Each ruleset carries a different way of reading numbers. Without a ruleset, every comparison is meaningless.

The source-verification gap follows immediately. The source title and the source publication are both missing, so source quality cannot be graded and no claim can be traced back to its origin. When nothing can be traced, the writer is forced to downgrade every sentence to unverified.

The risk principle is suspended along with it. Screening weight-cut risk requires a weight class, a weigh-in result and rehydration data. Screening brain-injury risk requires cumulative strike counts, knockout history and concussion markers. Assessing post-career security requires an athlete's name and an earnings profile. None of those boxes exists. One point needs stating plainly: the absence of a risk rating does not mean the absence of risk. The correct status of this file is "unknown", not "neutral".

Before believing a rumour, I count every passage of play. This time there was no passage to count, so there was no rumour worth rebutting either.

There is a very human temptation in this trade: fill the empty box with the voice of an expert. A preview can still be produced. A headline can still promise an outcome. The body can still quote a couple of coaches, add a few lines about fighting spirit, and close with a prediction that sounds very reasonable. That approach has no shortage of followers, and it always sells.

My reasons for refusing are verifiable. Without information points, all analysis is fabrication, and fabrication on a sports page is a systematic distortion rather than a style. A null result here is a workflow diagnosis, not a verdict that this story is trivial. Confusing "unimportant" with "undetermined" is the most expensive error a data writer can make. And if I published an analysis built on a blank file, I would be borrowing seven years of ringside credibility to vouch for something with no basis. That credibility can be spent once, and I am not spending it here.

There is a counter-reading worth considering. Many newsrooms treat a data gap as a small matter, because readers only need a story. But combat sports readers look things up very fast. They know how many fights a fighter has had, the weight class, the most recent weigh-in. A piece built on fiction gets caught within hours, and what is lost is not the article but the right to be believed in the ones that follow.

Frequency is what audiences overlook, but coaches do not. Gyms are the same: they log every session, every round, every cut. A writer has no right to be lazier than the athletes.

What needs fixing is not the draft. It sits in the subject-classification box still empty in the first layer. To re-run the workflow, the deconstruction stage must declare the subject class up front, must leave at least five information points on one coherent topic, and must attach the source title and publication with an absolute date. Without any one of those three conditions, the deep-analysis layer should not be allowed to start.

Over the coming days I will watch three signals. Whether the reloaded information-point list reaches five items or more. Whether the domain label migrates from "martial_arts" to a resolved subject class. And whether the source appears with a publication name and a publication date. When all three light up, I will sit down at the desk.

A good analysis begins with one correctly filled box. That box is still empty, and I am leaving it that way.