Swimming: when the numbers go silent, a report can still look dangerously perfect

6:40 a.m. in Brisbane. I open a fourteen-page swimming analysis file: nine analytical dimensions, forty-seven table rows, complete timestamps, formatting precise down to the last colon. Every data cell carries the same phrase: insufficient information to assess. The file is not broken. It passes every automated check in the system. No error, no red flag, no warning. Structurally correct, substantively empty. A month ago I sat through the replay of the men's 100m butterfly final at the Beijing 2026 Olympics, where Michael Phelps touched in 50.58 seconds and Milorad Čavić finished in 50.59. Serbia filed a protest; the organisers rejected it. Nearly two decades later, I still will not claim I know for certain what happened in that instant. The machinery was not lacking. The margin sat below the threshold that human eyes and most cameras can reliably read. This morning's file sits on the opposite side of the same problem. In Beijing, the data was so abundant that it generated its own controversy. In Brisbane, the data is so absent that nothing is left to argue about. In my trade, the second kind is far more dangerous. To explain why, I have to describe how we work. The swimming analysis process I use has two stages. Stage one reads the source article and breaks it into information points, core viewpoints, named entities and time sensitivity. Stage two takes that output and runs it through nine deep analytical dimensions: technique, performance and data, competition system, world landscape map, rules and anti-doping governance, athlete career pathway, risk profile, public narrative, and industry ripple effects. It sounds dry. But in swimming the structure has to be this tight, because the sport has a feature few others share: the same distance, the same motion, but entirely different rules depending on the stroke. Breaststroke allows a single dolphin kick after the start and after each turn. Freestyle and backstroke allow underwater travel of no more than fifteen metres before the swimmer must surface. Butterfly has no fifteen-metre rule but does bind the two arms to simultaneous recovery. If stage one does not establish the stroke, stage two does not know which rulebook applies. That is why the technique cell in this morning's file reads empty. Swimming is also a sport where physical context changes what a number means. A 50-metre long course and a 25-metre short course are fundamentally different frames of reference, because the short course has twice as many turns, which means twice as many chances to exploit wall push-off. A record set at the short-course world championships cannot be placed beside a long-course record without an annotation. If the input does not say long course or short course, every comparison downstream is meaningless. There is another variable outsiders rarely notice: the textile era. After 2026, World Aquatics banned polyurethane racing suits, ending the 2026-2026 period when world records fell in waves inside buoyancy-generating suits. A 2026 mark and a 2026 mark do not carry the same comparative value, even at the same distance. Any all-time ranking that ignores this variable is comparing two different sports. Then there is selection. The United States selects through Olympic Trials, taking the top two in each event, and that is often the cruellest mechanism in swimming: a third-place swimmer at Trials may hold a faster time than another continent's champion and still stay home. China uses a comprehensive evaluation mechanism. Australia runs a national qualifying system with A and B standards. An A standard grants direct entry. A B standard depends on quota allocation and may never be used. Three pathways, three completely different ways of telling the same athlete's story. Our stage one returned three empty fields: title, source and article type. No title means no subject. No source means no source-quality tier: specialist media, mainstream outlet, or self-published account. No article type means we do not know whether we are handling a results report, an interview, or an opinion piece. The nine dimensions downstream are therefore empty too. They are not blank because I was lazy. They are blank because the input was blank. And this is the point I want to leave exactly as it is, nothing added, nothing removed: silence in the data is not evidence of wrongdoing, and it is not evidence of compliance either. It is only silence. Now to the main part: what those nine dimensions actually require, and why emptiness is so dangerous. Technique first. To say anything about swimming technique I need at least five groups of numbers: start reaction time, underwater kick distance within the first fifteen metres, turn time at both walls, stroke rate and distance per stroke cycle, and touch time at the finish. These numbers are not decoration. They separate two athletes who finish with the same time. Two swimmers touching the wall together can have completely opposite technical structures: one wins at the start and underwater, the other wins in the closing sprint. The first tends to hold form better across multiple rounds, because underwater technique costs less energy than continuous arm turnover at maximum intensity. The second depends more on the aerobic base and risks collapse in the third swim of the day. Without splits, those two people are one identical line of results. Swimming also has a physical development feature few sports match at the same level: the puberty barrier. For teenage female swimmers, this is the single most important filter and also the most ignored. A fourteen-year-old breaking an age-group record offers no guarantee she will still be improving at eighteen, because changes in body structure can shift the ratio of propulsive force to drag in the water. Articles that call these girls the next phenomenon usually have no tool at all to assess this variable. Performance and data require a coordinate system, not just a single value. The same 200m freestyle time for a woman, placed against the world record, against the all-time list, and against the current-season ranking, produces three completely different stories. How a swimmer distributes speed across the two halves of a race, accelerating late or fading late, is a signal the leaderboard never displays, because the leaderboard only records the final result. I still remember the first time I saw a negative split structure, meaning the second half faster than the first, in a middle-distance final. That day I stayed in the stands long after the crowd had gone, rewinding the footage, writing every 50-metre segment into my notebook. What I learned was not that negative splitting wins races. What I learned was that a swimmer with a negative split structure usually sits in the group with a better fitness reserve, and that only becomes visible when you break the data down. For events of 200m and above, energy distribution across three rounds matters even more than the final result. A swimmer can win the heats by emptying the tank, pay for it in the semi, then fade in the final. The final results sheet records one line for that person. No sheet records that the title was lost in the morning heats. The competition system has to establish which tier the meet occupies: Olympic Games, long-course world championships, short-course world championships, World Cup, continental, or national. The tier determines the discount applied when reading results. In a year when an athlete is training heavily toward a major meet two years away, swimming slower than a personal best at a World Cup event says nothing about real form. The data table has no column for training heavy or peaking. The reader has to add it. This dimension also carries officiating risk, which the numbers never display. One false start removes a swimmer from the event they are strongest in. One relay take-off error costs four people a medal. Those events do not appear in any long data series, and no model predicts them. The world landscape needs a map drawn stroke by stroke. Swimming has no single ruler. Every distance and every stroke has its own power structure: some events are dominated so completely that the rest of the field is racing for silver, some are a four- or five-way melee separated by half a second, and some are quietly changing hands because the previous champion group has passed its peak. The men's 100m breaststroke is an example of the first type. Adam Peaty set the world record of 56.88 seconds in Gwangju in 2026, and for years he was the only man under 57 seconds. Reading an event like that, the runner-up's time says more about the winner than about the runner-up. The women's 400m freestyle is an example of the second type, where Katie Ledecky once held the 800m freestyle world record of 8:04.79 set in Rio in 2026, and then Ariarne Titmus beat her over 400m in Tokyo 2026 in 3:56.69. The same pair of rivals, two events, two opposite readings. Each type of map demands a different reading. In a melee, finishing order says far less than the gaps between touches. And in a transition period, the most common error is taking one season's results and labelling an entire generation. Part of that map sits in the talent supply chain. The American collegiate system produces swimmers at scale, with enough depth that athletes replace each other internally. A centrally funded national system produces fewer swimmers but invests deeply in each one. A club system depends on private money and tends to concentrate in a few large cities. These three models differ in how fast they identify talent and in how much risk families have to carry. Speaking of families, it has to be said plainly: scouting networks in developing countries do two things at once. They find talent, and they create lottery tickets. A family pouring money into a big training centre is not buying a medal; they are buying a probability, usually undisclosed, and the price is paid in the child's childhood by the family itself. I have seen enough cases not to write about young talent in an excited tone. Rules and anti-doping governance demand their own discipline. When related content appears, I am required to separate four tiers: a confirmed positive test, a contamination dispute, a procedural violation, and an allegation that exists only on social media. These four tiers carry completely different consequences, and merging them is the fastest way to destroy a career with prose. An empty file in this dimension does not mean clean, and it does not mean problematic. It means there is not yet anything to tier. The career pathway needs biographical data. A position on the age-performance curve, whether early bloomer, rising, peak, veteran maintaining, or returning, determines how every subsequent result is read. The same result means something entirely different for a seventeen-year-old and a twenty-nine-year-old. This dimension is also where I have to speak with what cannot be measured. The shoulder-injury history of a freestyle swimmer and the knee-injury history of a breaststroker are medical data, but the feeling of an athlete stepping onto the blocks for the third swim of the day is not in any table I own. I do not trust emotion. I trust a data series longer than your emotion. But I also know a long data series may still never reach the one thing that decides everything on the blocks: the fear of being left behind in someone who has just lost their place. Public narrative is a little easier to measure. The successor label is applied very quickly and verified very slowly. I once spent time counting how often labels like phenomenon of the new generation appeared in the international press over a four-year cycle, then cross-checked how many of those swimmers actually won an Olympic medal in the labelled event. The ratio was not pretty. But that figure does not prove the media wrong. It proves the media has no forecasting duty, and readers should not use journalism as a prediction table. Now the risk profile, and this is where this morning's file exposes itself. In my risk matrix there is one line that belongs neither to an athlete nor to a meet, but to the process itself: the risk that an empty input is used to produce a document called deep analysis. Level: high. Probability: high. Impact: high. Mitigation: block it at stage one before anything moves forward. I rate the overall risk file high, and I want to be explicit that the high rating belongs to the process, not to anyone in the water. Those nine dimensions have not been assessed. Not assessed is not the same as confirmed safe. The two are very far apart, and in my trade that distance is usually erased with a few adjectives. Industry ripple effects also need a concrete event before anything can be said. A medal in a popular event can lift swimming-lesson sales in that country the following summer, push pool rental prices up in large cities, and send sponsorship budgets toward the federation. Emma McKeon won seven medals at Tokyo 2026, and back home in Australia that effect was measurable in junior enrolment numbers over the next two seasons. Kaylee McKeown won both the 100m and 200m backstroke at the same Olympics, and training centres in Queensland recorded enrolment increases in the teenage age groups. A world record in a low-viewership event hardly ripples at all. Swimming is an attention economy driven by events, and an event that does not exist has nothing to ripple. The betting analysis industry, where I earn my living, sits inside that ripple too. I have been in it long enough to know what happens when a model meets empty data: it does not go quiet. It interpolates. And that interpolation becomes a very confident-looking figure in a price table. Kazan was the day I learned that a 99 percent probability can still die on the betting board. But it took several more years to recognise the second lesson, the more uncomfortable one: a 99 percent probability dying from bad data hurts, while dying from empty data that was still counted is shameful. The reflex of the sports media when it meets a gap is to fill it with adjectives. Missing numbers, add explosive. Missing splits, add character. Missing context, add under enormous pressure. Those sentences are not grammatically wrong, and they are why readers increasingly do not trust sports journalism. There is a notable contrast here, and I will say plainly that it is a contrast of process, not of culture. A system that errors loudly stops everything and forces a fix. A system that returns an empty result in the correct format passes every validation gate, and only stops when a real person sits down, reads it, and asks why the data cell on the twelfth row is empty. The paradox sits right there: loud failures are safe, silent failures are dangerous. And in sport we are building a great many systems capable of producing silent failures, because we reward formal completeness. I also want to place two things side by side that our industry often merges: correlation is not causation, and missing data is not exoneration. Everyone knows the first. Few are willing to say the second, because it permits no conclusion in either direction. It permits only one action: go back to stage one, find the original article, get the title, the source, the publication date, and run it again. If that cannot be done, the only thing left is a long document, structured, annotated, and without content. That kind of document does no harm sitting still on a hard drive. It only does harm when it is handed to someone else to read as a conclusion. Numbers have no gender, but the people who read them do. And a reader of an empty table usually does not realise they are reading an empty table, if that table is presented beautifully enough. The work required is not mysterious: add a gate at stage one that rejects any result with zero information points or no named entity; require the source URL and retrieval timestamp; audit the whole batch for other articles that came back empty for the same technical reason. Then run it again. Not to rescue the old file, but to learn what is missing: a swimmer, a distance, a stroke, a long course or short course, a meet tier, a timestamp. With those six, the nine dimensions downstream finally have ground to stand on. And the question I leave for myself, and for anyone who read this far: if a swimming report looks perfect enough that nobody checks it again, what happens to the swimmer in the next round, who has no idea their name was just read out inside an empty document?