Seven in the morning on a Tuesday in Brisbane. On the desk sits a fourteen-page PDF titled "Stage-2 Deep Professional Analysis — Swimming Domain". I open the first page, turn to the second, and set the coffee cup down. Nine sections. Nine tables. And almost every cell in those nine tables reads the same phrase: "N/A — insufficient information".

No athlete name. No event. No finishing time. No competition date. No federation. The note at the top of the file explains why: the upstream extraction step returned an empty list, so the entire analytical system behind it had no raw material to run on.

I read all fourteen pages. By the last one I realised something that made me sit there longer than I had planned. This was the most honest analysis I had received in months. It did not invent. It did not fill the gaps with lines like "the young athlete is in excellent form". It said plainly: no data, no conclusion.

Then I thought about the hundreds of swimming articles I have read this year. The ones where every cell was filled, generously, smoothly, and not a single line could be verified.

What happened to that file

A professional analysis system runs in two passes. The first reads the source article and extracts information points: who, which event, what time, which meet, what date, what source. The second compares those points against the sport's reference frame — world record, all-time list, season ranking, qualification standard — and only then issues a judgement.

In this case, the first pass did not run. Not because the source was hard to read. The residue left in the file shows the model echoed its own instruction instead of executing it: in the field "entities involved", the content is "identify from the information points above" — while the list of information points above is empty. That is the signature of a pipeline that stopped before inference.

To an outsider this is a technical fault. To someone who does this for a living, it is a story worth writing, because it exposes exactly the boundary the sports analysis trade is standing on: we have plenty of tools for producing the form of understanding, and very little discipline for refusing when the raw material has not arrived.

In a newsroom, the pressure always runs the other way. The editor needs copy. The reader needs content. A fourteen-page file of "insufficient information" cannot go to press. So someone opens another document, picks a few familiar numbers, and assembles a smooth article. I did that a decade ago. I now know what it costs.

In June 2026, in Kazan, I published a conclusion built on a complete and correct dataset. Germany had 74 per cent possession, only eleven passes into the box, an xG of 0.7 — lower than South Korea's. A week later FIFA's official numbers confirmed every figure. But I have also seen the opposite, many times: analyses with every cell filled, printed, broadcast, and afterwards not one line traceable. Kazan is the day I learned that a 99 per cent probability can still die on the betting table. But an article that looks perfect while having no provenance dies the moment it is published — the reader simply does not know yet.

The anatomy of a swimming analysis

The smallest unit of raw material is the split — the time for each 50 metres within a race. A final result tells you who touched first. The split sequence tells you why. A winner who goes 25, 26, 27, 28 is distributing effort in a completely different way from a winner who goes 27, 27, 27, 25. Same total time, two different tactical stories, two different forecasts for the next meeting. Without splits, every tactical claim is an inference from feeling.

The second layer sits in the start and the underwater. Reaction time is measured in hundredths. The underwater segment after the start, usually measured to 15 metres, is where the leading programmes have won most of their advantage over the past four years. A swimmer can lose the race in the first fifteen metres and never catch up, even though the eye sees two people leaving the blocks level.

The third layer is turns and finish. In a 50-metre pool, a 200-metre race has three turns. In a 25-metre pool, the same event has seven. Each turn is an opportunity to gain or lose two to four tenths. That is why a swimmer with excellent turns can dominate indoors and look ordinary outdoors.

The last layer is efficiency: stroke rate and distance per stroke. These two trade against each other. When a swimmer raises stroke rate without losing distance per stroke over the final 50, that is a body with fuel left. When distance per stroke drops eight per cent in the last 50 while rate rises five per cent, that is a body paying back an oxygen debt — usually visible two to three weeks before the result shows it.

I have spent many mornings at Chandler Aquatic Centre in Brisbane recording stroke rates in the final set. The coaching staff watch the stopwatch. I watch whether the rhythm holds from metre 35 onward. Two different views, and the second one usually announces what the first has not yet seen.

Two pools, two record systems

This is the most common error in amateur swimming comparison, and it appears in professional coverage too.

A 50-metre pool is the Olympic and World Championship standard. A 25-metre pool has more turns and is generally faster. World Aquatics ratifies records separately for each. A short-course record does not replace a long-course record, and the two datasets cannot sit in the same table without a label.

The practical consequence: if someone tells you a swimmer "just broke the national record", the first question is which pool. If they cannot answer, the rest of the conversation has no technical value.

In that empty file, the field "pool conditions" reads "insufficient information". It sounds trivial. It is in fact the single most important field in the entire performance section. Without it, even if someone supplies a specific time, I still cannot place that time on the performance axis.

The fabric era and the 2026 line

No comparison across swimming eras can ignore material. From 2026, polyurethane suits pushed the performance baseline to a level the world federation itself had to intervene on. The 2026 World Championships in Rome produced 43 world records — a figure widely recorded in international swimming media and the direct trigger for the textile-only rule that took effect on 1 January 2026.

From that line, every performance comparison must answer one question: does this result sit before or after 2026. A time from 2026 cannot share a table with a time from 2026 without a note. Official record books still carry rows from the old era, which is why analysts must use separate reference frames for each material period.

The governing body's name has also changed. In December 2026, at the World Short Course Championships in Melbourne, FINA announced its renaming as World Aquatics. The detail matters for sourcing: a document published before December 2026 carries the name FINA, and citing the wrong issuing body is a source error, not a typo.

A-cuts, B-cuts and the price of a place

At the selection layer, swimming has the clearest standards system of any Olympic sport, and is also the most easily misread.

An A-cut grants direct entry. A B-cut depends on a federation's quota allocation. But holding a standard is not enough. Each country has its own selection mechanism, and that mechanism determines the real value of a result.

The United States system is the harshest example: only the top two on the day at Trials are selected, regardless of the third finisher's world ranking. A swimmer can be the third-fastest person on the planet and not go. For an analyst this means a national trials result cannot be read like a World Championship result, even at the same distance in the same pool.

Vietnam sits in a different model: a limited quota, standards set by the international federation, and entry lists often finalised after weighing one athlete across multiple events. For a swimmer such as Nguyen Huy Hoang — who has raced the 800m and 1500m freestyle at recent Olympic Games — the value of a metric is not the metric itself, but whether it opens an additional slot elsewhere.

In the empty file, this entire section is marked unassessable. Correct. With no meet name, no date and no federation, there is no way to know where the result sits in the Olympic cycle — Olympic year, adjustment year, build-up year or sprint year. That is the most important discount factor when reading any swimming result, and its absence alone voids every valuation.

The puberty line

This is the part I want read most closely, because it is where women's swimming is most often misread and where few people speak plainly.

In women's swimming, puberty is the most frequent cliff on the career path. Physical changes between 13 and 16 alter body proportions, propulsion, buoyancy and feel for the water. Many swimmers who broke age-group records stall for two or three years before finding a technique that fits the new body. Some never find it again.

So when reading a teenage female swimmer's result, the mandatory question is: where does this sit against age-matched swimmers historically, and which phase of development is she in. Without age, sex and a longitudinal series, that cannot be answered.

In the file, the "age-performance position" field reads insufficient information. The "puberty barrier risk" field reads the same, with a notable note: it cannot even be determined whether the subject is a teenage female swimmer, because there is no swimmer in the data at all.

I read that line three times. It is the most honest sentence in fourteen pages, and it is a sentence most swimming coverage will never write. Instead it attaches the word "prodigy" to a fourteen-year-old girl, compares her with a 24-year-old Olympic champion, and calls that analysis.

Where I refuse to put the pen down

The rules and anti-doping section is left entirely void, with a note worth paraphrasing: no doping element exists in the input, and the analyst declines to impute one.

That is the right call, and harder than it looks.

Swimming carries a complicated testing history across several distinct levels: confirmed violations, contamination disputes, procedural problems in sample collection, and allegations that exist only on social media. These four have entirely different consequences for a human being, and collapsing them into one word is the most damaging error a sports writer can make.

So I keep one rule: I write about doping only when there is an athlete name, a deciding body, a document and a date. Without those four, I do not write. That rule once cost me a well-paid commission. I kept it.

In the file, risk is assessed at two levels. Subject level: unassessable, because there is no subject. Process level: high, and that is the only genuinely assessable item in the whole report. The stated reason: an analyst who fills nine sections with plausible-sounding swimming content will produce a document indistinguishable in form from real intelligence and entirely worthless in substance.

Prodigies and the label trap

Most of the media heat in swimming does not come from records. It comes from labels.

A fifteen-year-old who finishes fourth in an off-event is called a phenomenon. Six months later, when she finishes sixth in her main event, the same outlet calls it a step back. No data changed. Only expectations changed, and those expectations were never recorded as a number to check against.

Measuring the gap between market expectation and underlying reality requires two sides. The first is expectation: a label, a headline, a density of coverage. The second is fundamentals: age, sex, results series, season ranking. In the file, both sides are blank. No source headline, no outlet, no date. Even the article type — news, profile, business or opinion — is recorded as "unclassified".

Genre is the coarsest predictor of a story's heat. A profile of a sixteen-year-old has a completely different expectation structure from a news item about a selection slot. Without genre, every judgement about media heat is a guess.

Why the empty file is more trustworthy than the full one

The natural reflex on receiving an empty file is to go find another file. That is the reflex of a content producer. But that reflex produces something with the shape of understanding and no interior. And in a market where readers find it ever harder to verify sources, the shape is the only part that still gets trusted.

An analysis that declares itself empty is worth more than one that fills every cell with inference, because the first points out where it cannot stand and the second does not.

This is also why I distrust the increasingly beautiful heat maps in sports analysis. The heat map has become a new form of fortune-telling: it gives the reader the feeling of having seen the whole race, when in fact it only encodes what the person drawing it chose to count. In swimming, a heat map of split distribution can completely conceal that the swimmer turns badly over the last two walls — something that does not show up in a picture but does show up in a column of numbers.

A calculation with no reader is meaningless. The same split sequence is read by a coach to adjust a training block, by a bookmaker to adjust a line, by a fan to find a reason to believe or stop believing, and by a swimmer to learn where the seconds went. Four readers, four conclusions, from one attitude-free series. Numbers have no gender, but the people reading them do.

I once sat in a Brisbane meeting room and heard a performance director say I "looked at people like machines". Six months later he called back to ask why my table had been right. I said: because that table never claimed to know everything. Data has no gender. The way people use it does.

The six-field minimum

There is one practical finding in that file I want to pass straight to people working in Vietnam.

To unlock almost the entire analytical system, six fields are enough: athlete name, sex, age, stroke and distance, time with pool length, and meet with date. Six fields. No giant database. No contract with an international data vendor.

Those six fields unlock four sections at once: technique, performance, competition system and athlete career. Only then can media heat and commercial impact be discussed. If I had to choose one thing to do before next season, I would record those six fields for every domestic swimming result rather than write ten more opinion pieces.

For Vietnamese swimming, the gap has concrete consequences. Names such as Nguyen Thi Anh Vien carried Vietnamese swimming at the SEA Games for years, yet the split sequences from those championships sit scattered in meet reports and were never standardised into a reusable database. The result is that every new generation starts from zero, with no domestic reference frame to tell them whether they are fast or slow compared with their predecessors at the same age.

The next-cycle signal

One detail in the file is the signal I would track above all others: the prompt residue left in the output fields shows the fault lies at extraction, not in the source. Which means the source article is very likely still there, readable and analysable. Re-run the first pass with a stricter prompt and most of those fourteen pages move from void to substance.

For anyone following swimming, this is a reminder about order of priority. Before arguing over who wins, check whether anyone recorded that swimmer's reaction time in their last race. Before calling a fifteen-year-old a phenomenon, check which phase of physical development she is in. Before quoting a record, ask which pool and which year.

Those three questions cost far less than one wrong analysis, and they are the only thing I have carried with me from Kazan to now.

Map of limits: What this piece can assert — the structure of a swimming analysis across splits, reaction time, underwater distance, turns and efficiency; the non-convertibility of 50-metre and 25-metre pools; the textile rule effective 1 January 2026; FINA's renaming as World Aquatics announced in December 2026 in Melbourne. What this piece only infers — the specific reason the extraction pipeline halted, and the genre of the source article. What this piece rests on professional instinct — that a declared-empty analysis earns more long-term trust than a fully populated one, a belief I cannot prove with numbers and will not pretend I can.