The Empty Report in Busan: The Cost of a Table Tennis Data Skeleton With No Body

At 6:40 in the morning, on the second floor of a table tennis academy in Haeundae, Busan, frost still clings to the window glass while inside the ball keeps a metronome rhythm. A tablet lies tilted on the umpire's desk. On the screen is a scouting report that is formally complete: a title, a trainee's name, nine numbered sections, a table, a date line. The young coach picks it up, skims it, nods, and gets ready to forward it to a parent.

Then he stops.

Nine sections. None of them has a body. The title field reads N/A. The source field reads N/A. The article type reads "unclassified". The information points field is an unnumbered blank. The one-sentence core viewpoint: empty. The author stance: empty. The article purpose: empty. The entities field carries an instruction to "identify from the information points above" while there are no information points above. Time sensitivity: not assessed. Source quality: "judge from the source fields" while the source fields are N/A.

One field is populated. Domain: table tennis.

That report did not lie. It was simply hollow. And because it looked finished, it was more dangerous than a sheet that says plainly "insufficient data" — the plain sheet gets rejected at the door, while this one gets signed.

I have seen many reports like that in forty-five years of watching this industry. They do not come from lazy people. They come from machines that ran the process correctly but lost an input layer, and from reviewers who were never taught that an empty frame can look like a finished report.

Beneath the glass

The tablet in Haeundae is the endpoint of a two-stage pipeline. Stage one extracts: it reads match footage, academy training logs, tournament records, and returns a numbered list of information points — each one a piece of evidence that can be traced back to a source. Stage two analyses: it takes that list, runs it across nine dimensions, and every conclusion must cite a specific stage-one point, by number or by content.

That constraint is what makes the report worth reading. It is also what killed the report that morning.

Stage one returned the skeleton with no body. Stage two refused to continue.

Such refusal is rare. Most scouting systems on the market fail open: when data is missing they fill the gap with inference, label it "estimated", and publish. The word "estimated" is a fig leaf. It does not tell the reader which parts were measured and which were guessed; it only says the writer anticipated being wrong. Inside an academy, such a label goes straight into a fourteen-year-old's file and stays there for four years.

The system in Busan fails closed: no evidence, no conclusion. It audits the input layer before doing anything else, and when that layer is broken it records what is missing instead of guessing what is missing. All nine dimensions were marked N/A — insufficient information. No technique breakdown, no head-to-head table, no risk matrix was constructed.

The stage-two operator wrote a line I would frame: a confidence label on an invented fact is worse than no analysis at all.

That is why I am writing this. Not to tell the story of a broken machine, but to talk about our trade — a trade so used to filling blanks that it has forgotten that blanks have a grammar of their own.

Table tennis is the sport where the distance between what a clip shows and what a player actually is stretches widest among the small-ball games. Forty seconds of a rally can make a fourteen-year-old look like a national-team berth. The clip does not show recovery time between points, does not show the footwork in the backhand corner when the ball is placed short, does not show the third-ball decision at 9-9. And a clip with no scoreboard, no commentary and no metadata shows nothing at all — including that fact.

Audit first, analyse second

Table tennis scouting has three input streams. First, match footage: camera angle, frame rate, whether a scoreboard is visible. Second, load logs: sessions, volume, intensity, recovery between games. Third, competition records: opponents, scores, rounds, ball type.

When one stream drops, the other two do not fill the hole. They lie. They lie persuasively, because each stream is correct in its own terms.

Suppose the footage is complete and the load log is gone. You watch three games, see tidy footwork, even lateral movement, a forehand loop that still carries spin. You write: peak form. The missing log would have said something else: repeated lateral accelerations down twenty-three percent over six weeks, recovery between points up 1.4 seconds, four consecutive heavy sessions last month — one too many.

I have lived through exactly that mechanism. In 2026 I held a twenty-three percent signal in my hands and sat on it, waiting to cross-check medical records. I held it long enough for a young athlete to leave the court in the first game of a quarterfinal with a torn hamstring. That lesson did not teach me to guess when data is missing. It taught me that data sufficient to warn must be published, and data insufficient to conclude must be labelled as such.

Auditing the input layer is not paperwork. It is the only step in the whole process where you can still discover that you have nothing.

A blank cell and a zero are different things

This is where most youth scouting reports lose their credibility, and also where an empty report is more honest than a full one.

A cell reading zero means measured and found to be zero. A cell left blank means not measured. The two get mixed in the same spreadsheet, and once mixed, the blank is always read as zero, because zero is easier to read.

A concrete case from Korean youth table tennis. A fifteen-year-old girl has a blank head-to-head cell against defensive players. The coach reads: has not beaten a defender. The harsher reality is different: the domestic youth circuit now has almost no defenders to play. The away-from-table blocker and the chopper have thinned markedly since the eleven-point rule compressed rally length, and academies optimise for the rule. That blank describes the circuit, not the girl.

Reading a blank as zero is how you discard an athlete who has never been tested, rather than one who has failed. In a scouting economy with limited time, that happens weekly and nobody records it.

For the same reason I always separate two kinds of conclusion in my reports: source-based, and inference. A source-based conclusion must point to which information point, at which layer. An inference must declare itself as inference, and must survive being removed from the report without collapsing it.

Three hypotheses, and their order

The stage-two operator did not stop at marking N/A. He ranked three possible causes by confidence, and the order matters more than the list, because it decides where the next forty-eight hours go.

First, high confidence: upstream extraction returned null. It received a source it could not classify and, instead of raising an error, emitted the template with empty fields. In table tennis terms: the source was a video with no commentary, no scoreboard, no metadata. The extractor had nothing to key on, so it keyed on nothing, and it did not say that it had keyed on nothing. This failure mode is the most common because it is the quietest.

Second, medium confidence: the pipeline dropped the body. The source was fetched empty, or sat behind a paywall, or was blocked. In table tennis terms: an academy's internal report behind a login page, or a federation database returning access errors to every automated request. You have the address, you do not have the content, and the system logs the absence of content as a successful run.

Third, low confidence: schema mismatch. The information points exist under different keys than the ones stage two expects. In table tennis terms: one system logs points won in rallies, another logs points won in the first five shots. Same match, same player, two definitions, no warning. This is the failure people discover only after the decision is made.

Three hypotheses, three completely different responses. If it is the first, you re-extract. If the second, you buy access. If the third, you rewrite the mapper. Chasing the wrong one costs a tournament cycle, and for a fifteen-year-old a tournament cycle is a quarter of the growth curve.

Nine dimensions, nine different prices

A report with nine empty cells is not a report that is missing nine times. Each blank has its own price, and that price depends on which decision is waiting for it.

An empty technique, tactics and equipment cell means you cannot even build a baseline. No blade, no sponge thickness, no shakehand or penhold, no service pattern, no third-ball efficiency. A left-handed penholder travels a completely different path from a right-handed shakehander, and every comparison after that is wrong if this cell is empty. This is the most expensive blank in the technical group.

An empty player data and head-to-head cell means you cannot price a record. A seventy percent win rate on the regional youth circuit says nothing without knowing whom, in which round, under which format.

An empty event system and scoring cell means every cross-era comparison is meaningless. Table tennis moved to eleven-point games in September 2026, to a 40mm ball in October 2026, to a plastic ball in July 2026, and from January 2026 the world ranking counts only a player's best eight results in twelve months. Every rule change is a broken data series, and people still splice numbers across the break.

An empty competitive landscape cell means you read an individual phenomenon while the problem is structural. The gap between China and the rest in table tennis is not produced by a handful of outstanding players; it is produced by the density of a training system in which finishing tenth domestically can be harder than finishing first elsewhere.

An empty rules and governance cell means you do not know what is permitted. The hidden-serve ban took effect in 2026, along with equipment approval rules and limits on coaching intervention — each of which changes how a child is taught.

An empty coaching and talent pipeline cell means you do not know who is responsible for progress. Six months of improvement may come from the coach, from a move to a different training group, or from simply playing more matches. Without this cell you credit whoever is standing closest.

An empty risk surface cell means you cannot see what is accumulating. Repetitive-load injury, burnout, early specialisation, and the psychological cost of a fourteen-year-old being ranked on an adult list.

An empty public narrative cell means you cannot measure pressure. A viral clip creates an expectation level, and that expectation becomes a variable in coaching even though nobody logs it.

An empty industry transmission cell means you do not know where this report will travel. It may be resold, reused for another selection cycle, quoted in a transfer file the athlete never sees.

A teenager entering an Olympic Games at seventeen, as Shin Yu-bin did in Tokyo, is the output of a chain of decisions in which every link rests on a report. How many empty cells can that chain absorb before it collapses?

Degraded-profile analysis

There is a way to work when only a fragment of the input survives, and it deserves to be written as method rather than treated as a trick.

First principle: run every populated dimension at full depth. If the technical cell has data, the technical analysis must be as deep as ever, not levelled down to match the rest.

Second principle: mark empty dimensions with the same symbol, in the same position, in every report. Consistency of the marker matters more than its content, because it lets a reader compare two reports without guessing.

Third principle: every conclusion carries its evidentiary base. Source-based, or inference. There is no third category.

Fourth principle, and the one I paid for: do not manufacture the missing eighty percent to make the report look complete.

I moved to publishing in the form of a data diary — raw figures plus a prediction, even when incomplete, updated continuously. It builds a readership that follows the reasoning process rather than the result, and it turns scarcity into part of the product instead of a defect to be hidden.

Perfection delays the warning, and time does not wait for anyone to finish the data.

The other side of the mirror

Here the story turns to what I consider the most important part, and it does not flatter our trade.

The empty report in Busan is more honest than most of the full reports in circulation. The empty one fails visibly. The full one fails invisibly, and fails more often.

Picture a report with forty-seven parameters. Six are measured. Forty-one are inferred from those six, or from the writer's memory, or from an old report on a different athlete. Not one of those forty-one lines carries any mark distinguishing it from the six. The reader sees a uniform block, and that uniform block becomes the basis for pricing.

I have to be direct about a structure I have watched for years: the satellite club system lets large clubs sidestep domestic training-compensation rules, and prodigies from small circuits become satellite assets. The scouting report is the instrument of that transfer. An empty report blocks a transfer. A full report prices one. Between those two failure modes, which is more common?

I am not indicting anyone here. I am pointing out that the same system which produced the empty report also taught people how to fill blanks, and that the second is many times more dangerous than the first.

There is a point I cannot leave out, though, and it makes the picture less clean.

Refusing to publish is also a form of completeness worship. The machine in Busan did not invent. It also said nothing. The young coach stood in front of an empty frame and nearly sent it to a parent, and had he done so, that parent would not have read "insufficient data". They would have read something else: my child is not good enough.

A blank cell carries a social meaning entirely different from its data meaning. That is the trap. Honest in the file, cruel in the delivery. And the fix is not to fill the file. The fix is to change what the file says when it is empty — adding a line explaining that this blank is due to a missing source, that one due to a blocked source, that one because nobody has measured it yet.

The first brick is not on the blueprint; it lies under the dust of Busan.

What is worth demanding

After that morning in Haeundae I sat alone in the hall once the balls had stopped bouncing. A chalk trace of a service drill was still on the board. Outside the window, the frost had lifted.

Every generation is a geological layer; the patient read the sediment, the shallow look at the cross-section.

We are used to demanding that scouting reports be complete. Perhaps we should demand that they be traceable. Every line must be able to say where it came from. A blank line is traceable — it points straight at the absence. An invented line is not, and precisely because it is not, it is never questioned.

Value is not listed on a price sheet; it is excavated from forgotten matches.

And if a line in a report about a fourteen-year-old cannot say where it came from, who benefits from that line staying exactly where it is?