The Blank Column on the Scoreboard: What Swimming Data Cannot Measure

The Blank Column on the Scoreboard: What Swimming Data Cannot Measure

A blank column

On the night of July 31, 2026, at La Défense Arena on the outskirts of Paris, I stood in the press area behind lane four, less than four metres from the water. The pool was so flat that the ceiling lights reflected off it as a single unbroken strip of silver. Eight swimmers stepped onto the blocks. Nobody said a word to anybody. Camera shutters rattled like rain on a tin roof, then cut out the instant the referee raised a hand.

The whistle. The water breaking. The breathing.

Forty-six seconds and forty hundredths.

When Pan Zhanle touched the wall, the scoreboard lit up a number that no prediction sheet I had ever read dared place in the “high probability” column. Three months earlier I had sat with the World Aquatics dataset: his progression curve, every fifteen-metre underwater split, his average stroke rate across the last two rounds. No column said “world record.” That column was empty.

And that empty column is what I brought home.

The measuring apparatus

Over the past fifteen years swimming has become the most heavily surveilled sport in the aquatic family. Every lane at a world-class meet is captured by at least four cameras, two of them underwater and tracking the swimmer along the longitudinal axis. Every touch is measured by a pressure pad with an error margin in hundredths of a second. Every stroke cycle is counted by computer-vision algorithms. World Aquatics publishes split data at the fifteen-, twenty-five- and fifty-metre marks for nearly every final.

At Paris 2026 the volume was far greater still. Real-time positional tracking. Force data off the starting blocks. Breathing-rate estimates derived from imagery. The depth and frequency of every underwater dolphin kick. A single men’s 100m freestyle final now generates more data rows than an entire week of competition did twenty years ago.

Here is the paradox: the more data there is, the fewer surprises there are. Prediction models are increasingly accurate at the medal tier. In some freestyle events, twelve of the sixteen finalists can be called before the heats even finish. But when a surprise does happen, it does not happen in the tail of the distribution. It happens in that empty column.

Four tenths of a second

The men’s 100m freestyle world record before Paris 2026 stood at 46.80 seconds, set by Pan Zhanle himself in February 2026 in Doha, on the lead-off leg of a relay. In the individual final in Paris he swam 46.40.

Six months. Four tenths of a second.

In an event where every hundredth is ground out over a full training cycle, four tenths in half a year is a leap that a standard progression curve simply cannot explain. For more than a decade before that, this record had crept forward by hundredths. Cesar Cielo swam 46.91 in 2026, in the high-tech swimsuit era. It took nearly thirteen years to erase that mark, and the man who erased it was David Popovici, a Romanian teenager, with 46.86 in Rome in 2026. Then, just two years later, people were swimming under 46.50.

What the data does not tell you is the stretch of time between Doha and Paris. After Doha, Pan Zhanle went almost silent with the international press. He raced little, appeared little, answered questions in very short sentences. The domestic press wrote about him in the language of compression.

And there was another variable no dataset can encode. In April 2026, a series of investigations by Western news outlets, in collaboration with a German broadcaster, reported that twenty-three Chinese swimmers had returned positive tests for a heart medication in late 2026 and had been cleared by the national anti-doping agency, while WADA chose not to appeal. The story spilled into every pre-Games press conference. The entire Chinese team walked into Paris under its shadow.

No column in any dataset reads “psychological pressure from a four-month investigation.”

Four minutes, two seconds, fifty hundredths

Léon Marchand arrived at Paris 2026 as the host nation’s most anticipated athlete and left it with four individual gold medals: the 400m individual medley, the 200m butterfly, the 200m breaststroke and the 200m individual medley.

The 400m individual medley is where the data has to bow. Marchand swam 4:02.50. The previous record was 4:03.84, set by Michael Phelps in Beijing in 2026, and it had stood for sixteen years — across four Olympic Games, across at least two generations of more systematically trained medley swimmers.

The last man to win four individual golds at a single Olympic Games was Mark Spitz, in Munich in 2026. I dislike comparisons of that kind because they belong to the language of grand pronouncements. But here it is simply a historical fact, and it is measurable.

The data says Marchand is the most complete medley swimmer of his generation. It says his butterfly speed, breaststroke speed, backstroke speed and freestyle speed all sit in the leading group. It does not say why he swam faster in his home water. No column reads “the roar of eight thousand people.” No unit measures the sensation of an entire arena pushing you forward.

I do not know what Marchand was thinking over the final twenty-five metres. I only know that after he touched the wall, he lay on his back on the surface and stared up at the ceiling for a long time — longer than he needed to catch his breath.

Fourteen minutes, thirty seconds, sixty-seven hundredths

Bobby Finke came to Paris as the reigning champion in both distance events. In the 1500m freestyle he swam 14:30.67.

The previous world record was 14:31.02, set by Sun Yang in London in 2026. Twelve years. Throughout those twelve years, this event was regarded as the slowest-improving discipline in the pool, because it depends on pain tolerance more than on optimisable technique. Every time someone crept close to the record, people talked about “physiological limits.”

Finke broke it at twenty-four, in a final in which he swam his last two hundred metres faster than almost his entire first two hundred. That is a speed distribution the optimisation models typically advise against, because it demands a capacity to suffer at high lactate that cannot be held consistently across multiple rounds.

The decisive number is not in the split sheet. It is in the pain threshold, which no sensor measures directly.

Age as a faulty variable

Katie Ledecky won the 800m freestyle for the fourth consecutive time at the age of twenty-seven. It was the ninth Olympic gold medal of her career. Sarah Sjöström won the 100m freestyle at thirty, and then the 50m freestyle.

Both ran counter to a common assumption in forecasting models: that the peak of women’s swimming sits around twenty to twenty-two, followed by decline. The assumption is not statistically wrong, but it was built on samples that were too small and too heavily weighted toward the pre-2026 period, when average competitive lifespan was shorter mainly for economic and medical reasons, not physiological ones.

In Paris, two of the most significant individual gold medals went to two women at an age the models call “post-peak.” That does not prove the assumption wrong. It only proves the assumption is insufficient for prediction.

A streak of fifteen Games

In the men’s 4x100m medley relay, the United States had won every Olympic Games it contested since 2026. It is one of the longest dominance streaks in modern Olympic sport, and it was not built by any single individual. It was built by a system: high-school recruitment, university scholarships, a dense internal competition calendar, and a culture in which relay swimming is treated as the highest honour.

At Paris 2026, the streak ended. China took gold, with Pan Zhanle on the anchor freestyle leg.

There is a tempting way to read the data here. You look at the result, see that Pan Zhanle swam the fastest leg, and conclude that China won because of one individual. That reading ignores the first three legs, ignores the starting order, and ignores the fact that China spent a full four-year cycle building capacity in the other three disciplines. An individual can only swim the anchor leg if the three before him have kept the gap within reach.

The system always sits behind the number, but the system rarely shows up on the scoreboard.

Development systems and patience

Discussions of swimming performance tend to give almost all their airtime to athletes and almost none to the development system behind them.

In France, Marchand’s emergence was not a random event. He was trained in a high-standard domestic environment, then moved to work with Bob Bowman — the coach who guided Michael Phelps — in the United States. That is a hybrid model: European roots, North American technical and coaching culture, and a clear national objective.

In China, the rise of the men’s squad unfolded over roughly a decade, with a centralised training centre, a dedicated data-analysis unit, and a long-term strategy adjusted after each Olympic cycle.

What the data does not show is how long it takes a system to produce a world-medal swimmer. The figure usually cited is eight to twelve years. No split sheet measures those eight years. No heat map draws an investment decision.

Four hundred metres and three women

The women’s 400m freestyle at Paris 2026 was the final in which the data was almost entirely right, and precisely for that reason it is a lesson about limits.

Ariarne Titmus won in 3:57.49. Summer McIntosh took silver in 3:58.37. Katie Ledecky took bronze in 4:00.86.

Three athletes, three coaching systems, three continents if you count training origins. On the data sheet, the gap between first and third is 3.37 seconds, roughly seven metres at that speed. In the stands, that gap is imperceptible until the final twenty-five metres, when Ledecky begins to fade and McIntosh begins to accelerate.

What the data got right was the finishing order. What it does not convey is the feeling of that final twenty-five metres, when an athlete understands she is swimming exactly to plan and it is still not enough.

Heat maps and divination

In modern swimming analysis, no tool is presented more scientifically than the heat map. It displays the speed curve metre by metre, the distribution of underwater time, stroke rate by cycle, and the correlations between them. Looking at one, you feel you are reading a blueprint.

But the heat map has become a new form of divination.

It describes what happened. It does not explain it. A heat map showing Pan Zhanle going out fast will lead a reader to conclude he is “weak on the back half” — when in fact he still won, because everyone else’s back half was weaker. A heat map showing Marchand’s butterfly leg outclassing the rest of the medley field does not explain why he held that butterfly rhythm in an arena so loud that coaches had to signal with their hands.

More dangerously, the heat map conceals an athlete’s real role within a system. A swimmer going slower may be doing something else entirely: setting the rhythm for a relay squad, preserving a start for a later event, or swimming in an accumulated state to serve the plan for the whole week. No curve on a heat map annotates that.

And when analysis leans on the heat map, it begins to shape its own subject. Young coaches read heat maps to find the medal-winning pattern, then coach to that pattern. Stroke rates converge. Underwater time converges around the fifteen-metre mark. Technique gets faster and more uniform. That is a form of homogenisation I have seen in another sport, and I do not think it produces a better competition.

A contaminated dataset

The historical continuity of swimming data has been broken, and few people comment on it.

The high-tech swimsuit era ran from roughly 2026 until the international federation banned polyurethane fabric in 2026. During that window, more than a hundred world records were erased by swimmers wearing suits that increased buoyancy, cut drag and pressed the body into a favourable posture. Many of those marks took more than a decade to fall, and some still stand.

Any progression curve that blends the two eras — before and after the ban — is contaminated data. A forecasting model built on it will repeatedly mispredict at the record tier. Labelling the suit era clearly is a mandatory step, yet it is routinely skipped when data is pushed onto visual comparison charts.

None of this makes analysis meaningless. It makes analysis harder, and it demands something data tables cannot generate on their own: caution about provenance.

When governance speaks a different language

The story of the twenty-three swimmers is another example of the limits of numerical analysis.

The data said one thing: the 2026 samples returned anomalous results, the national anti-doping agency determined the cause was food contamination, and WADA decided not to appeal. That is a procedural conclusion, grounded in records and process.

The media said another thing: a national body cleared its own athletes, and the global watchdog did not object. That is a systemic suspicion.

Neither is the full truth. The full truth lives in procedural documents, in appeal deadlines, in the regulatory definition of contamination, and in technical meetings no journalist was allowed to attend. It is a slow, tedious kind of truth that cannot be compressed into a single figure.

Data analysis can only speak about what is measured. It cannot replace governance.

What someone standing close can feel

Based on my years of watching swimming finals up close, there is a signal no instrument records that is almost always accurate: the interval between a swimmer stepping onto the blocks and the whistle sounding.

Someone about to swim well tends to stand still. Someone about to struggle tends to adjust a lot: tugging at a suit strap, wiping the face, glancing at the neighbouring lane, looking up at the stands. At La Défense Arena, Pan Zhanle stood so still that I thought he was holding his breath longer than necessary. He did not glance sideways once.

That is not evidence. It is observation. But after many years I have learned that in a sport where everything is measured, the unmeasured signals are the ones that arrive first.

I once sat in Kazan in June 2026, when Mbappé ran at thirty-seven kilometres an hour and I understood that speed, past a certain threshold, stops being movement and becomes a kind of dance. Kazan taught me that speed knows how to dance too. I carried that lesson into swimming, and at La Défense Arena I saw it hold in a different way.

But in Paris I heard no roar for Pan Zhanle. The stands were full, yet no mass of sound belonged to lane four. He swam inside a silence of his own making — and inside that silence he swam faster than anyone had ever swum.

The language of silence is not the absence of information. It is another kind of information, one that requires the reader to stand closer and listen longer.

What remains after the number

I am not writing this to deny data. I have spent more than a decade reading split sheets and I still read them every week. But I have learned that swimming data is strongest at exclusion, not prediction. It tells me who cannot win a medal. It very rarely tells me who will win one, and it almost never tells me how.

At La Défense Arena, amid the noise of eight thousand people and the clatter of camera shutters, I stood in a very quiet place: the area behind lane four, minutes before the final began. Nobody spoke. There was only the faint stir of water as a swimmer dipped a hand in to check the temperature. In the emptiness, I hear the breathing of the race more clearly.

Silence is not short of language — it owns a language of its own.

The next world record will come from a blank column again. There will again be a dataset that failed to predict it, a model that will explain it after the fact, and a heat map that will redraw it as a perfectly reasonable curve. The writer’s job is not to chase the spreadsheet. The writer’s job is to stand close enough to hear what the spreadsheet does not record.

What remains after a final is not the number. It is the quiet after the number lights up, when eight thousand people fall silent, and only one person is still breathing on the surface of the water.