The Table Tennis Data File Came Back Empty: Notes on Verification Discipline

Opening

3:47 a.m., Nha Trang. I open the file stage1_deconstruction.json the way I would open a sealed test result. Inside there are eleven fields. Ten read N/A. One reads table_tennis.

The night before, I had staked my time on a four-stage chain: collect reports, extract facts, grade sources, build the analytical frame. At the second stage, everything stopped. No tournament name. No player name. No date anchor. Not a single claim to cross-check. Only the domain label survived, intact as the fingerprint of someone who left the room before I could turn on the light.

I read sports data for a living. I am used to a number betraying me. An empty file does not betray. It goes quiet. And in this trade, silence is the most dangerous substance there is, because it opens a gap that every writer has an instinct to fill.

Context

Vietnamese table tennis is at a stage where the data infrastructure has not caught up with expectations. Domestic tournaments are not scarce: the national table tennis championship runs annually, there is an age-group youth system, and there are open cups with international guests such as the Hùng Vương Cup in Ho Chi Minh City. Names are not scarce either: Đinh Quang Linh and Nguyễn Anh Hoàng on the men's side, Mai Hoàng Mỹ Trang on the women's side, along with others that domestic audiences know by face. Venues are not scarce: table tennis always sits in the group of sports Vietnamese fans expect medals from at the SEA Games.

What is scarce is a public layer of statistics.

A V-League football match now has at least three data providers running in parallel, each returning hundreds of metrics per game. A domestic table tennis match has a scoreline. And a few lines of description.

Fans read 4-2 or 4-3 and rebuild the story in their own heads. How did the winner take the fifth game, was it the serve, was it the opponent running out of gas, was it a lucky net ball at 10-10 — all of it sits beyond the reader's reach. The writer was in the room, but memory does not operate like an archive.

I have followed this industry for thirty-one years, starting as a fact-checker at a sports magazine where getting one number wrong cost you a client. That discipline stayed in how I write. It makes it impossible for me to write “this player dominated the final exchanges” if I have no way to count how many final exchanges that player won.

And that is precisely why my data file came back empty.

Analysis

What happens when one layer of data returns empty? In engineering, this is called null propagation. An empty field at the extraction stage empties the entire chain behind it. The source-grading layer has nothing to grade. The frame-building layer has nothing to build. The system still runs, still exports a correctly formatted file, still opens in the reading software. It simply contains nothing.

The fatal point is that the system raises no error. A malformed file gets caught in three seconds. A correctly formatted empty file passes every checkpoint, and only surfaces when someone actually opens it near four in the morning.

For a sports writer, the same problem carries a different name: the pressure to fill the blank. You have a deadline. You have a match that just ended. You have a template with nine cells to complete. And your data source returns zero.

The most obvious escape is to write in pure description. Retell the flow, describe the arena atmosphere, quote the coach. It is safe and it is honest, and it adds no value to the reader.

Another route is to lower the standard. Turn “I have no data” into “from observation, this player looks slightly sharper.” The phrase “looks slightly” works like a licence to practise.

The most dangerous route, though, is invention. Not outright fabrication. Just selecting a few round numbers that sound plausible, attaching a vague source, and believing them long enough for them to become true inside the article. I have walked all three routes, and this is the only one that leaves no trace.

In table tennis there is a principle anyone who has trained seriously knows: you do not read spin with your eyes. A topspin ball and a backspin ball can travel the same trajectory for the first two thirds of the flight. What you read is the opponent's racket response after the ball lands. Spin lives in the consequence, not in the shape.

Data behaves the same way. The spin of a number lives in how it was produced. A service-points-won rate of 68% may signal an exceptional server, or it may signal a scorekeeper who only counted short rallies. There is no way to tell the two apart if you look only at the final figure.

Table tennis has a scoring structure so tidy it seems almost designed for measurement. Eleven points per game. Serves alternate every two points. Deuce means both players have passed 10-10, and from there every point is a deciding point.

Which means certain things are measurable that football needed years to standardise. Win rate at deciding points. Win rate when leading 9-7. Loss rate after leading 9-7. Points won directly on the third ball. Average rally length. Share of points ending in a forehand loop versus a backhand block.

In football, you need a complex metric such as expected goals to convert chance quality into a number. In table tennis, one person sitting courtside with a notebook and a spreadsheet already has the raw material.

So why do we still not have it? Because manual counting costs people. Because nobody pays a writer to sit and count two hundred points over four hours. Because the revenue of a domestic table tennis report cannot support a full-time analyst.

And because nobody demands it.

This is where the empty file connects. A market with no data consumers will not produce data producers. When no reader verifies whether the number in the article actually exists, the incentive to produce real numbers weakens and the incentive to produce plausible-sounding numbers strengthens.

I once sat in the press room of a domestic tournament and heard a coach describe his student as “controlling the match well.” I asked what control meant in that case. He smiled and said: “It means the ball rarely goes where he doesn't want it.”

That answer is better than any metric. But it cannot be measured. And because it cannot be measured, it will be replaced by a louder-sounding phrase in tomorrow's report.

There is another variable anyone doing table tennis data in Vietnam must account for: playing conditions. Humidity in a northern arena in February differs sharply from humidity in a southern arena in August. The 40+ plastic ball absorbs moisture, gets heavier, and loses flight speed. A player who relies on pace loses an edge in humid conditions, while a player who relies on spin and placement barely changes. If I pool an entire season into a single metric without splitting by venue, I am mixing two different sports into one table.

That is the lesson of stratification. The market administrator does not manage money flows. They manage expectations. In my trade, expectation means readers believing that every number in the piece has a root. When that belief goes, no metric rescues you.

There was a period when observation conditions collapsed in a different way: matches with no spectators. When the stands emptied, I found the transfer pattern. Without crowd noise, players lose a layer of feedback and a layer of pressure. Young players, who lean on external energy, fall back; older players, who already play on internal rhythm, barely change. A single metric measuring both groups in one table produces a meaningless average.

Contrarian angle

The conventional response to an empty data file is to fill it. Improve the collection tools, add a fallback layer, increase the number of input sources. I did exactly that for years, and I suspect most colleagues in the field are doing the same.

I now think differently.

An empty file does not need to be hidden. It is the most accurate measurement of the distance between infrastructure and expectation. My extraction layer did not break because of poor code. It returned empty because the input material contained no verifiable fact. That is a problem of source, not of system.

If I patch it by letting a model infer the missing parts, I get a complete file. And I lose the ability to know how weak my source was. I blind myself with my own tool, and the complete file will look exactly like a real one.

This runs against the instinct of most people doing this work. We are taught that silence is failure, that an empty table is an unfinished table. But in sports analysis, some tables must be empty. The deuce statistics table of a match with no deuce must be empty. The expected-goals table of a team that took no shots must be empty. And an article with no data must say that it has no data.

In football, I learned this late. In 2026, I staked my position on expected goals. The V-League answered with a shock, when a team won a match with a metric nearly half that of its opponent. That club then dropped points repeatedly, and I was credited with being right. But what stayed with me was not winning the argument. It was the realisation that a correct number can still lead to a wrong conclusion if I forget it describes only one layer.

The 2026 World Cup taught me: data is never a single layer. I collapsed the group stage and the knockout stage into one metric, then was surprised that the champion was not the team with the prettiest numbers. That was a stratification error. The error in today's empty file is the mirror image: I let a later layer invent material for an earlier one.

There is another trap that is rarely discussed, and it is spreading through both football and table tennis: the heat map. In modern coverage, the heat map has become a familiar decorative device. It looks good, it conveys professionalism, and it hides the player's real role in the tactical system.

A player who specialises in short serves to open up the third-ball attack will have a very faint heat map, because most of the time that player stands near the table and moves within a small area. Their real value lies in the quality of the serve and in the decision about which direction to attack next, things a heat map cannot draw.

The heat map is a new form of divination. It gives us an image to believe in, rather than a question to answer.

In table tennis, the equivalent trap is the score-progression chart. A handsome rising line across the first three games may simply reflect an opponent testing a new serve pattern before unleashing their real game in the fourth. That line is right on the numbers and wrong on the meaning.

Takeaway

I still keep stage1_deconstruction.json in my root folder. Not to fix it. To be reminded.

After seven years, I trust the silence between two numbers. That is where I have learned the most about how a sport operates: not in the final scoreline, but in its capacity, or its incapacity, to generate the next number.

The domestic table tennis season ahead is long. There will be more SEA Games, more youth tournaments that surface new names, more matches ending 11-9 where nobody can recall how the ninth point was won. If Vietnamese readers start demanding a verifiable number in every report, the market will have to produce it. If not, we will keep getting fluent articles about matches nobody measured.

What I leave for myself, and for anyone else holding a pen: if every table tennis report this season had a blank cell where the technical data belongs, who would be the first to dare leave it blank?