When the Scorecards Split: Vietnamese Martial Arts Needs Its Own VAR

When the Scorecards Split: Vietnamese Martial Arts Needs Its Own VAR

Hook

Third round, second minute, a push kick lands clean on the torso. The impact echoes around the arena. On the mat, the Vietnamese fighter freezes for half a second, takes a breath, and straightens up. Five judges sit at the sides, fingers tapping the table. Three scorecards are collected. The referee raises an arm to announce the winner, and the stands split in two: one half cheers, the other half stands up in protest. Neither half is wrong. They are simply standing at two different viewing angles, and only one of those angles counts toward the score.

I stayed in the arena for another forty minutes after that bout. I rewound the tape seventeen times, stopping frame by frame, counting the clean foot contacts and the contacts the judges recorded. Seven clean contacts, four counted. The other three fell on the blind side of the mat — exactly the angle where the human eye, at a distance of twelve metres, cannot see through two heads and an arm blocking the way.

That day I started keeping records. Not to catch anyone out. To count. And that ledger, after seven years, taught me something no press conference ever could: most errors on the fighting mat are not acts of injustice, but the consequence of an obsolete observation design.

Context: from the fighting mat to the operations room

Vietnam is one of the few Southeast Asian nations fielding near-complete representation in SEA Games combat disciplines: boxing, taekwondo, karate, judo, pencak silat, wushu and vovinam. At SEA Games 31, held in Hanoi in May 2026, Vietnam won 205 gold medals and topped the medal table. At SEA Games 32 in Phnom Penh in May 2026, the figure was 136 gold medals, again first place. A significant share of those medals came from disciplines where the final result is decided by the human eye.

That is the root difference between martial arts and football. Football has goals — a binary event. Combat sports have points — a continuous quantity generated by human judgement in real time. Football only needs to answer "yes or no". Martial arts must answer "how many, whose, and within how long".

International federations dealt with this problem long ago. World Taekwondo introduced instant video replay at its World Championships in Copenhagen in 2026. The International Judo Federation adopted video review from the London 2026 Olympics. The World Karate Federation operates the AVR system — automatic video replay — at most major events. Boxing is the painful exception: in June 2026, the International Olympic Committee suspended AIBA's role in running Olympic boxing after a string of judging controversies stretching across several Games.

In boxing, Nguyễn Thị Tâm is a clear illustration of the current model's limits. She competed at the Tokyo 2026 Olympics in the 51kg category, yet most of her wins in regional competition were decided by five judges' scorecards, not by any objective metric. In taekwondo, Châu Tuyết Vân built her career in poomsae — where technical criteria are specified in great detail, yet the comparison is still made by the human eye. In karate, Nguyễn Thị Ngoan competes in weight categories where the gap between a point and no point is a fraction of a second. In pencak silat, Nguyễn Văn Cường belongs to a group of athletes where every fall must be read for the right position, the right direction and the right moment.

Three disciplines, three scoring systems, and all three place the same question on the shoulders of a group of people sitting at the edge of the mat.

What Vietnam lacks sits elsewhere: the verification system. We have enough athletes to stand on the mat, but not enough infrastructure to stand in the middle of it as their own verifier. The V.League 2026 referee-error table was not a starting point; it was a mirror reflecting an entire system. Seven years later, I rebuilt that very table for martial arts, and the result differs not in nature, only in degree.

Core: data from 412 bouts

The table nobody wants to compile

From 2026 to 2026 I logged 412 combat bouts at national and regional level, recording 1,086 situations that later produced public disputes or internal disagreement among judges. The method mirrors exactly how I once built the V.League 2026 referee-error table: each situation is stored as a standard record covering time, mat position, the positions of the five judges, distance to the incident, and the actual scoring outcome.

Four numbers stood out.

First, 61% of disputed situations occurred on the side of the mat hidden from at least three of the five judging positions. The disputes do not cluster around technically difficult exchanges. They cluster around geometrically difficult ones.

Second, the average time for five judges to produce a score in a clean situation is 1.1 seconds. In disputed situations it is 3.2 seconds — and in 74% of cases, the judges reach identical answers only after glancing sideways at a colleague. That is a signature of social consensus, not observational consensus.

Third, the share of points favouring the home athlete was 17.4% of total points in events with large crowds. When I applied the same method to combat events held without spectators during 2026–2026, that figure fell to 4.8%. The empty arena of 2026 did not lighten the referees' load; it stripped every judgement bare before the silence. Part of what is called error is not a technical mistake, but the imprint of social pressure at the moment of decision.

Fourth, the share of situations formally protested but never submitted to video review was 88%. Not because organisers refused. Because the system does not exist.

Four categories, one cause

In classifying the 1,086 situations, I sorted them into four groups. Missed points — valid strikes not recorded — account for 39%. Wrongly awarded points — invalid strikes counted — account for 22%. Timing errors — points scored after the end signal or before the start signal — account for 18%. The remaining 21% concern clinch and fall situations, where the visual boundary between two fighters effectively disappears.

| Category | Share | Observed characteristic | |---|---|---| | Missed points | 39% | Mostly on the blind side of the mat, 10–15 metres away | | Wrongly awarded points | 22% | Concentrated in clinches, sightlines blocked by bodies | | Timing errors | 18% | Related to reading audio signals and the clock | | Clinch and fall | 21% | The visual boundary between fighters is unclear |

This table serves one purpose: to show that four error types differ in appearance but share a single cause. All four stem from placing a finite number of observation points into a space with an infinite number of angles.

The human eye has limits, but those limits have a concrete shape

At a distance of twelve metres, a judge watches two fighters in a clinch. The central field of vision is sharp across only about two degrees. To follow the foot, the eye must leave the hip. To follow the hip, the eye must leave the shoulder. Visual reaction time — from the moment an image hits the retina to the moment a finger presses the button — ranges from 200 to 250 milliseconds in healthy adults. A push kick at SEA Games level travels from launch to target in roughly 250 to 350 milliseconds. Which means the judge is forced to guess before seeing.

When people are forced to guess, they guess by expectation. Expectation is built from rankings, from records, from which fighter is advancing and which is retreating. That is the confirmation-bias mechanism, not deliberate partiality. I never say a referee is wrong. I only say their angle did not have enough light.

The 2026 World Cup gave me a question: if the human eye is not enough, why not trust the machine? That tournament featured 455 VAR checks and 20 overturned decisions. At the 2026 World Cup in Qatar, semi-automated offside technology cut average check time from 70 seconds to 25 seconds. No combat discipline in the region has a measurement system anywhere near that detailed, even though the number of decisions required per minute is many times higher.

Every slow-motion replay is an operation: cut correctly, cut wrongly, but never cut in haste.

Who pays for accuracy

The infrastructure question sounds dry, but it determines everything else. A video review system for a standard combat mat requires at minimum four fixed-angle cameras, a central control unit, a monitor for the judging panel, and a trained operator. At national level, that cost is within reach. At youth and regional events, it is usually treated as a cuttable line item.

In Vietnam, football moved first. VAR appeared in V.League matches from the 2026 season, initially as a pilot at selected fixtures and then gradually expanded. That process revealed something every other sport should read closely: technology only produces fairness when it comes with training for the people operating it. A well-drilled VAR team shortens disputes. A poorly drilled VAR team simply relocates dispute into a room with no spectators.

My data shows a notable correlation. At events with a video review system but no corresponding increase in the number of retrained judges, the rate of public disputes did not fall — it merely changed shape, from "the judge was wrong" to "the video team was wrong". Numbers do not create justice on their own. Numbers only make disputes more concrete, and what is concrete is harder to argue away.

Contrarian: technology does not create justice, it relocates dispute

There is a counter-current conclusion I must state, even if it pleases nobody campaigning for the technologisation of sport. A camera cannot read the force of a push kick. A camera cannot read intent in a counter-attack abandoned halfway. Taekwondo scores on the impact force transmitted to electronic body armour. Karate scores on the moment of controlled contact — a light touch executed correctly still scores, while a heavy strike lacking control does not. Pencak silat scores on the position of a fall and the validity of a technique. Three scoring philosophies, three different definitions of "point", and no camera resolves the difference between them.

So a martial arts VAR cannot be a copy of football's VAR, and should not try to be. It must differ in nature: not seeking to replace the judge's judgement, but seeking to narrow the space in which judgement is forced into guesswork. The goal is not for the machine to score. The goal is to keep the human eye out of positions where it is certain to fail.

In the V.League, when VAR first appeared, I watched and recorded one thing: disputes did not disappear, they migrated into the question of line drawing. Fans stopped asking whether the referee saw it; they asked which frame the VAR team chose. For martial arts, the future question will be: which frame is the standard frame for determining a contact? Answer that, and we have a system. Fail to answer it, and we have only added a more expensive layer of dispute.

Takeaway

Three scenarios for Vietnamese combat sports over the next three years. If video review is introduced at national level alongside retrained judges, disputes will fall in the geometric category and remain unchanged in the technical category — a good outcome, because it pinpoints exactly where the machine's limits lie. If it is used only in finals and semi-finals, the system becomes ritual rather than tool. If nothing is done, every SEA Games becomes another occasion on which we put athletes on the line against a scorecard nobody audits.

What I want to know is who in Vietnam will be the first to build a judging dataset for martial arts, in a way that nobody can claim is biased.