Nine Analytical Dimensions of Table Tennis and the Value of an Empty Report

11:47 p.m., August 13. A data file landed in my inbox, its name a single token: table_tennis. I opened it. Nine sections. Section one, technique, tactics and equipment: N/A – insufficient information. Section two, player data and head-to-head records: N/A. Section three, event system and points rules: N/A. Section four, competitive landscape: N/A. Section five, rules and governance: N/A. Section six, coaching staff and talent pipeline: N/A. Section seven, risk surface: N/A. Section eight, public narrative and expectations: N/A. Section nine, industry transmission: N/A.

Nine out of nine. Not one name. Not one line of figures. Not one date. The only field holding content was the domain label: table tennis.

I sat still for four minutes. Not out of surprise. I have taken enough midnight calls to know that a midnight call is never just a message — and an empty file is never just an empty file. It is the trace of a break somewhere upstream, at a point where nobody had yet read a single line.

That night I did not write. I went looking for the break.

The nine-dimension framework I work with is not a scorecard of feelings. Each dimension is a door that opens only with the right key, and the key is always a concrete fact, never an opinion.

The technique dimension demands a description of the movement, the tactical deployment, and the equipment: rubber, sponge, blade. A change of rubber can alter the speed and spin of an entire run of matches; without that detail there is nothing to analyse. The player dimension demands a name, an age, a ranking, and above all a win rate against associations outside the domestic one — the metric I still call the place where world ranking and true strength come apart. The event dimension demands a tier and a calendar anchor. The landscape dimension demands an event line, because men's singles is far more open than women's singles, while mixed doubles carries independent Olympic weight and a one-pair-per-association cap. The rules dimension demands a selection mechanism. The coaching dimension demands a roster age structure. The risk dimension demands at least one event capable of doing harm. The narrative dimension demands a source and a source tier. The industry dimension demands at least one equipment brand, one sponsorship, or one policy signal.

A table tennis article, however short, almost always touches at least three of those nine doors. A piece about a player touches doors two and six. A piece about a tournament touches doors three and four. All nine doors shut at once is close to impossible for a real document, however thin that document is.

We are in the middle of a transfer window, and that makes this harder to believe. Every day the desk takes in dozens of fragments: a European club asking a price, an agent sending a photo of a contract, an anonymous account posting a midnight move. Most of it is noise. So the professional reflex becomes: filter. Filter with three sources, filter with timestamps, filter with the question of when a sale is most profitable. That reflex carries a trap: it makes us believe there is always something to filter. Some nights the only thing in the inbox is an empty file. Filtering an empty file yields no story; it only says something about the pipeline.

What stands out is not the nine N/A lines. It is the fields that a working extractor fills in automatically, even when an article is content-poor. Article type: unclassified. Article source: absent. The system did not get the content. It also failed to establish what it was reading.

Three hypotheses, ranked by descending probability. One: the fetch and extraction step returned an empty document — a paywall, a bot block, a JavaScript-rendered page, or a PDF or image the parser could not read. Two: a field-mapping error between the analysis layer and the extraction layer. Three: the article was routed to the wrong domain label.

All three lead to the same place: the break sits upstream of the comprehension step. And this is the point I want to hold longest. The table tennis label is the only field with content, so it is precisely the field that must be treated as unverified. When every source is silent, the label on the envelope is the least trustworthy thing on the envelope. A label may be a genuine routing decision; it may equally be the system default after everything else failed.

So I walked through each dimension and asked myself: if there were a name, what would this dimension demand? An unknown name is still a name. Here there is no name at all.

The second dimension hurts most. The ITTF ranking system runs on a rolling 52-week window: a tournament's points count only within that span, and when they expire the ranking falls even if the player has lost nothing. That is points-defence pressure, the force that never appears in a match report yet governs a professional's entire schedule. To analyse it you need a name, an event calendar, and an expiry date. Without a name, all three are zero.

The third dimension has the same problem in another form. The event tier determines the whole points gradient: under the WTT points table, a Grand Smash awards 2,000 points to the champion, a Champions event 1,000, a Star Contender 600, a Contender 400. A player can change an entire season with one good week at the top tier, and lose an entire season with one early exit at the bottom. Without the tier, every sentence about form is meaningless, because form only means something once you know where it was measured.

The fourth dimension needs a first discriminator: the event line. Men's singles has been markedly more open than women's singles across many cycles, and that changes how a result is read. A men's loss to an opponent outside the leading group is routine; the same result in women's singles is a signal. Mixed doubles is harder still: the one-pair-per-association cap turns pair selection into a resource problem rather than a purely technical one. Who plays with whom is an allocation decision, and it is usually made earlier than outsiders assume.

The seventh dimension, the risk surface, is the easiest to fake. People list generic risks: injury, overload, being decoded. A risk table untethered to names, ages, calendars and tiers is just a keyword list. It manufactures the feeling of analysis without any. Such a table does more damage than a blank page, because a blank page persuades nobody.

The eighth dimension demands the one thing I never trade away: source tier. Narrative analysis is the only dimension where an unrated source is more dangerous than no source, because conclusions about public sentiment flow straight into public expectation. When the article source is blank, every statement about the story being told has nowhere to stand — including statements that sound harmless, such as this piece is positive.

A report with the right shape is not a report with the right substance. And in this trade, the most dangerous artefact is a report that looks finished. Nine sections, nine headings, correct format, correct order, correct terminology. An automated system skimming it will find it valid. Only a slow reader sees that it is hollow.

The counterintuitive part sits here. Table tennis fans, and practitioners too, carry a background set that is right at an average level: men's singles is more open than women's singles; the world's leading group concentrates in a few associations; mixed doubles allows one pair per association; the WTT calendar thickens every year. That is enough to write a piece that reads smoothly. And that is exactly the temptation: fill the gap with background knowledge, then present it as a conclusion about a specific document.

I turned the three-sources reflex back on itself. The right question is whether the third source is genuinely independent of the first two, or merely a copy of the same information passed through three mouths. Here, all three sources are silent. Three independent silences are still silence.

One small detail in that report is the most instructive part: all six professional risk categories were marked N/A, none marked low. Because marking something low asserts that a situation was examined and found benign. It is a difference in language, but it is the difference between a writer and a template-filling machine. A hidden clause only surfaces when everything collapses — but this time what surfaced was not a clause, it was the break in the pipeline. And that break surfaced only because someone was willing to write do not know instead of writing a plausible-sounding answer.

From my experience following matches, there is something at the table that teaches exactly this lesson. Reading a serve requires the ball to actually cross the net. You watch the wrist, the contact point, the direction of spin — but none of it means anything until the ball has left the opponent's racket. With no ball there is no read; there is only a person staring at the table thinking they are analysing. An edge ball works the same way. It wins the point. It teaches you nothing about the opponent's serve. An empty file is an edge ball: a signal, no information.

That report left four warnings, and the first deserves to be printed on a wall. The risk of analytical fabrication: producing domain-flavoured conclusions from an empty payload, addressed by treating every N/A in the report as binding rather than as a gap to be filled.

The second is silent propagation: a correctly formatted nine-dimension report can be read by downstream automation as a completed analysis. This is the highest process risk of all, because it breaks nothing anywhere — it merely creates a false belief.

The third is mis-routing: the domain label is the only populated field, so it must be verified independently once content returns. The fourth is recurrence: if the root cause is a paywall or an anti-scraping block rather than a transient error, that source will fail again. The cheapest fix is to log the source identifier and failure mode at the fetch step, so the next failure is seen rather than re-diagnosed.

There is one metric I want on the desk: the null rate. Count records with an empty fact list against total records in a batch. If a batch crosses a few percent, it is a systemic fault; if it is a single record, it is an isolated one. Sports data has learned to measure model error, but it has barely learned to measure the silence rate of its pipelines. A system returning nothing more often than it should is a system lying through silence.

The self-assessment table in that report was blunt too: competitive value one star out of five, industry value one star, timeliness undeterminable because no date anchor was supplied, reference value two stars. All of those two stars come from a single thing: it documents a reproducible null case, together with the correct handling of a null case. No player, no match, no tournament inside it.

The next step is clear and needs no meeting. Recover the raw document. Re-run extraction, enable OCR if the file is an image or PDF, enable ASR if it is video. Log the source identifier and failure mode at the fetch step. And once content is back, the most valuable cut remains the second dimension: separating world ranking from true win rate against non-domestic associations. That is where media narrative and competitive reality diverge most.

One question I keep for myself after closing the file: if a nine-section report with not a single fact can pass through a system unchallenged, then across the sports copy published every day, how much of what is labelled analysed is in truth merely correctly filled in?

I do not believe in luck. I believe in how I wait for the phone. That night the phone did not ring. And that was information.