The Silent Lane: What an Empty Split File Taught Me About Verification

In the breaststroke leg of the 200m individual medley, what decides the race is rarely the swimmer who touches the wall first. It is the gap between the third and fourth arm cycle, where an athlete loses momentum within two-tenths of a second. That day I sat in the organizers' technical room, opened the split file to cross-check it, and found an empty table: column headers — 50m, 100m, 150m, 200m — with not a single number beneath them. A coach leaned over, looked at the screen and said: “Just comment by feel, nobody checks anyway.” I did not write. For someone who once recorded a striker's sprint distance wrong, “nobody checks anyway” is the sentence I fear most in this trade.

I came to swimming before I came to data. In 2026 I began my career in a newsroom as a swimming reporter, covering domestic and regional meets, learning to read a lane with my eyes: where the hand enters the water, how high the hip rides, how long a turn takes. I later moved into sports data analysis, and swimming is the sport that taught me the human eye is a poor measuring device. Over 200m a swimmer can look “slow” while their actual split is half a second faster; conversely, a powerful-looking sprint can be no more than the consequence of a rival easing off early. The eye sees motion; data sees cause.

Swimming has one of the densest data ecosystems in sport. A single race generates at least four layers of information: reaction time off the blocks, splits every 50m, stroke rate, and distance per stroke. At elite level, sensors are fitted to measure pulling force, shoulder angle and underwater depth after the start. No sport in Vietnam has such high analytical potential and is so little exploited. This is the paradox I meet every week: plenty of data, few people able to read it.

Vietnam's swimming scene sharpens that paradox. At the 2026 SEA Games in Singapore, Nguyễn Thị Ánh Viên won eight gold medals — a mark that, more than a decade on, remains the reference point for every domestic analytical table. At the 2026 Asian Games in Jakarta, Nguyễn Huy Hoàng won silver in the 1500m freestyle, Vietnam's first Asian Games swimming medal. Behind both milestones lie data files that were mostly never made public, never fully archived, and never analysed to the end. Fans remember medals; people who work with data remember that they once lacked the numbers to explain why those medals came.

In 2026, when I was the only data consultant in an analysis room, I made my first mistake and my biggest lesson. I miscalculated a striker's sprint distance: 1.2km recorded against an actual 0.8km. An expert in the room said, in front of colleagues, that women do not understand tactics. I did not argue. I stayed three months, re-checked fourteen thousand GPS samples and found three more systemic errors in the synchronisation software. From then on I set a rule: every number must pass at least two rounds of cross-checking before it enters an analysis. A small GPS deviation was enough to teach me: verification is everything.

Applying that rule to swimming, I understand why an empty split file is more dangerous than a wrong one. When data is wrong, you know you have a problem to fix. When data is empty, you have a silence that invites the imagination to fill it — and the imagination is always ready to serve the story you want to tell. In swimming, such silences are everywhere: a swimmer whose backstroke splits were not recorded, a lane missed by the camera, a youth-meet file never archived. If a writer is careless, he fills the blank with “form”, with “spirit”, with “character” — words that cannot be measured and cannot be verified.

My method for handling a data blank has four steps, and I apply it to both swimming and football. First, identify exactly what is missing: a split, a reaction time, or a stroke rate. Second, find an independent alternative source: broadcast footage, the referee's hand timer, or federation records. Third, compare the two sources and note the margin of error. Fourth, if it is still not enough, I write it plainly in the piece: the data is insufficient to conclude. The fourth step is the hardest, because it runs against a writer's instinct to tell a complete story.

My three rounds of checking are not a ritual. Round one checks provenance: where the number came from, which device recorded it, who operated it. Round two checks logic: does it match the footage and the related metrics. Round three checks repeatability: does a similar result appear at another meet. Only a number that passes all three goes into the piece; if it fails any round, it drops to the footnotes or is discarded.

In 2026, while supporting analysis for a sports channel during the World Cup, I learned the value of not rushing to conclusions. I collected the expected-goals figures for all sixty-four matches and found a team that reached the final despite creating only 5.3 xG in the knockout rounds, while its opponents combined for 7.1 xG. That team scored eight goals from 5.3 xG — an overperformance of more than fifty per cent. Look only at the result and you call it character. Look at the data and you see that a large part of it is the noise of a random variable. Croatia 2026 was not a miracle — it was xG written into history. The lesson followed me into swimming: a medal can come from repeatable skill, but it can also come from a lucky lane, and the analyst's job is to separate the two.

In swimming, separating skill from luck is harder than in football, because each lane gets a single swim and there is no rematch within the same session. A 100m freestyle swimmer has exactly one chance to prove it. That makes the sample small, and a small sample makes every conclusion fragile. This is why I always look at a season-long sequence of results rather than a single swim. A swimmer who drops from 55 seconds to 54 in one meet may simply be reacting to pool conditions — colder water, a weaker swimmer in the next lane, or just a day when the body worked better. But if that drop repeats across five consecutive meets, you are looking at a trend, not a one-off.

Even with complete data, reading it demands care about context. A short-course 25m time cannot be compared directly with a long-course 50m time, because there are twice as many turns and every turn is a chance to gain or lose time. A national record in short course does not mean the swimmer will repeat it in long course. When I read a results table, the first thing I check is not the final time but the pool type, the water temperature and the time of day the race was swum. Ignoring those variables is the fastest way to turn a correct analysis into a wrong conclusion.

I am often asked why I do not simplify swimming analysis to make it easier to read. My answer is that simplifying and oversimplifying are two different things. Cutting jargon is fine; cutting evidence is not. When I write about a slow turn, I have to show how many per cent slower it was than that same swimmer's turn at the previous meet — I cannot just say a slow turn. When I judge that a swimmer started well, I have to cite the specific reaction time, not rely on the feeling that they reacted fast to the gun. Vietnamese readers are getting sharper, and they deserve numbers instead of adjectives.

There is another temptation I must remind myself of every day: the temptation to turn data into a defence of a model I have already built. In 2026, when competitions were suspended by the pandemic, I spent seven months building a recovery-index model from the GPS data of hundreds of players. When the league returned, I predicted that the three teams pressing at the highest intensity faced a twenty-three per cent rise in injury risk. My club cut training load and lost no key player. But I was obliged to state clearly in the piece that the model had assumptions, a sample size and an error margin. The pandemic season taught me to measure a league by its recovery index, not by its points. Had I hidden that limitation, I would have turned analysis into advertising for myself.

Back to that empty split file. After confirming the file was genuinely blank, I followed the four steps exactly. I took the broadcast footage, hand-timed each 50m split, then compared them with the referee's timer. The margin between the two sources reached three-tenths of a second at the 150m mark — enough to change how the whole final leg read. I did not hide that margin. I wrote plainly in the piece: the original data does not exist, these are reconstructed numbers, and any conclusion drawn from them carries only medium confidence. Readers did not complain. A coach even messaged me to say he valued that I had made my limits clear.

But a data blank is not only an enemy. After many years, I have learned that it is precisely where the numbers fall silent that the human story shows itself most clearly. When there is no split to read, I am forced to ask the swimmer how the third leg felt. When there is no reaction time, I am forced to watch the footage again and notice a detail no number ever captures: the way she clenched her hand on the starting block. Data does not tell stories; it records everything so that I can tell them myself. And sometimes a shortage of data opens a door that an abundance of data had closed.

The danger lies elsewhere. Correlation is not causation, and in swimming that shows more clearly than anywhere. A swimmer raises her stroke rate and goes faster — but the cause may be that she improved her turn technique, not that she stroked faster. A national team changes coach and wins a medal — but the cause may be a cohort of swimmers just reaching maturity, not the new coach. If a writer ignores that distinction, he turns analysis into a series of plausible-sounding stories that are wrong at the core. With an empty data file, the temptation is greater still, because a blank contradicts nothing you write.

What I carried away from that empty split file was not a conclusion about any swimmer, but a question I now set myself before every piece: if all the data vanished tomorrow, would I still have enough evidence to defend what I wrote? For Vietnamese swimming, where data remains scattered and archiving is not yet systematic, that question is not hypothetical. It is the working condition. I believe in numbers, but only after a number has passed three rounds of checking. And when a number does not come, the most honest thing a writer can do is say that it did not.