Human Eye, Machine Eye and Camera Angle: The Three Layers of Modern Martial Arts Judging

Human Eye, Machine Eye and Camera Angle: The Three Layers of Modern Martial Arts Judging

2:14 a.m., in a small newsroom near Hongik Station, Seoul. I freeze the frame at the 3,847th second of the second fight of the night, slow it to one-eighth speed, then restart from the beginning. On the secondary monitor, the three judges' scorecards are still blinking like a medical diagnosis: 29-28, 28-29, 29-28. Three men sitting on three sides of the fighting surface, watching the same fight, applying the same scoring scale signed off before the opening bell, counting the same valid strikes — and producing three different results. None of them broke the rules. That is exactly what kept me awake.

I do not watch the goal; I watch the camera angle watching the goal. In martial arts, that sentence needs one more rewrite: I do not watch the strike land, I watch the system counting the strike.

Three layers of scoring. First, the human eye — the oldest tool and the most distrusted. Second, the machine eye: sensors, electronic gear, force-detection software. Third, the camera angle, the player standing behind both, deciding what the audience is permitted to see and therefore what they are permitted to believe. In twelve years of tracking contentious decisions across amateur and professional boxing, taekwondo and mixed martial arts, I have learned something that looks like a paradox: when technology enters the arena, it does not make controversy disappear. It only relocates it.

Context: Three Scorecards, Three Different Philosophies About One Word

To read a scorecard, you first have to know what is being counted. The three major combat disciplines today do not share a single definition of "scoring."

Boxing uses the 10-point must system, in which a round is awarded to the winner, the loser receives 9, a drawn round can be scored 10-10, and dominant penalties can push the card to 10-8 or 10-7. Three judges sit on three sides of the ring, each scoring independently, and only two of the three need to agree for the result to stand. The system's origins trace to the standardisation of professional boxing in the United States in the mid-twentieth century, and it was carried intact into the Olympic arena, where the rules permit no draw in knockout rounds.

Taekwondo took the opposite path. Scoring here is cumulative, not round-based: each valid kick to the body protector or the head adds one, two, three or four points depending on placement and technique. From the 2026 World Championships in Copenhagen, the electronic Protector and Scoring System became the official standard, and from that point a portion of the decision-making authority shifted from the referee's eye to force sensors. In parallel, video review was approved by the sport's world governing body in 2026 and appeared at the Olympic Games in Beijing 2026 — earlier than many glamorous team sports that still boast about their technology.

Mixed martial arts follows a 10-point must system resembling boxing's, but with four evaluation criteria that are entirely different. The unified rules of mixed martial arts were adopted by the New Jersey Athletic Control Board in 2026 and later standardised by the Association of Boxing Commissions, setting an order of priority: effective striking first, grappling and control second, aggression third. Meaning a fighter can throw fewer strikes and still win the round if those strikes are judged more "effective."

Three scorecards, three philosophies. And all three rest on one shared assumption: that what counts as a hit can be seen, counted, and — most importantly — retold faithfully.

Layer One: The Human Eye and the Night in Seoul 2026

On 1 October 2026, at the Jamsil Arena in Seoul, the light middleweight boxing final took place between a young American puncher and a home-nation fighter. The American, Roy Jones Jr., threw faster and more accurately, and by the statistics widely published afterwards, landed nearly three times as many strikes. The three judges still scored it 3-2 for the home fighter, Park Si-hun. The result was announced, the gold medal was awarded, and the entire arena understood that something had just broken.

After the Games, some of the judges in that bout were suspended. I do not have enough documentation to conclude whether this was a technical error or something else, and I have no intention of playing investigator. But there is one production detail I always remember when I rewatch the footage from that night: no camera angle was close enough for the spectators in the hall to see what television viewers saw on their screens. Meaning two fights happened at once — the one inside the arena and the one on television — and they ended differently.

Kazan erased a goal, but it opened an eye. The Kazan incident of 2026 taught me that a contentious decision is not a mistake to be erased but a window to be opened. The night in Seoul 2026 was Olympic boxing's first window: it forced the sport's administrators to admit that three men sitting on three sides of a ring, with no tools, no replay, could produce a result the whole world found impossible to believe.

That night gave birth to a question still not fully answered four decades later: when a decision cannot be verified, what exactly are we trusting?

Layer Two: The Machine Eye and Faith in Sensors

Taekwondo's response to that problem was far more technical. If the human eye cannot be trusted, let force decide.

From the 2026 World Championships in Copenhagen, the electronic Protector and Scoring System became the standard. Pressure sensors inside the protector and headgear register impact, software compares it against a minimum threshold, and points are added automatically to the electronic board. On paper, this is a leap forward: a kick without enough force is not counted, a kick with enough force is counted, and there is no room for bias.

Reality is more complicated. I spent most of 2026 — the year the global arena froze — coding scoring situations from multiple seasons, and what I found was not in the software. The problem with the machine eye is not that it counts wrong; it is that it only counts one kind of truth. A sensor measures force. It does not measure intent, does not measure timing, does not measure a hook that lands on empty air yet forces an opponent half a step towards the boundary — an action any experienced referee reads as a sign of capitulation.

At Beijing 2026, before the electronic system covered everything, video review produced one of the sport's most important precedents. The bout between Sarah Stevenson of Great Britain and Chen Zhong of China ended with the win going to the home fighter under the initial in-ring decision. The British delegation protested, the referee panel reviewed the footage, and the decision was overturned. For the first time in the sport's Olympic history, a result announced in the ring was erased by the images from a camera.

I regard that event as a milestone no less significant than any medal. It established a principle: images can replace instantaneous judgement. But it also raised a more uncomfortable question: if images can overturn a judgement, is the in-ring judgement still the final judgement at all?

Layer Three: The Camera Angle, the Decision-Maker Nobody Ever Questions

This is the layer almost nobody writes about, because it sits outside the rules, outside the scorecards, outside the referee's report.

A televised fight is not a fight. It is the output of a chain of directorial decisions: which camera shoots close, which shoots wide, when to cut to the crowd, when to show a replay, and from which angle. The broadcast director chooses the frame. The frame chooses the truth. The audience believes the frame because they have no other option.

I tested this systematically across more than two hundred fights that I watched simultaneously on two live feeds from two different countries. The result was fairly stable: in situations involving contact near the boundary or behind the head — the most contentious category — the two feeds usually gave opposite impressions for the first three to five seconds, before both feeds cut to the same slow-motion replay. That slow-motion replay is not self-evident. It is chosen by people.

The referee is the fastest reader of the fight; I only write one beat slower. And inside that one-beat gap there is a variable nobody records in the report: the cutting rate. I once watched a fight in which only one usable camera angle existed until the third round, because the main camera went out of focus after round one. For those two rounds, viewers at home had no way to assess strike quality on the far side of the ring. The three judges' scorecards were still published as usual. None of them was affected by the technical failure. But hundreds of thousands of people at home were. And those are precisely the people who will write the collective memory of the fight.

The camera angle is not an instrument for recording truth. The camera angle is a co-author of truth.

An Additional Layer: When Three Judges See Three Different Fights

Mixed martial arts is where the tension between the three layers peaks, because it combines boxing's three judges with taekwondo-level complexity in its criteria.

On 16 November 2026, at a major mixed martial arts event in Las Vegas, Georges St-Pierre defended his championship belt against Johny Hendricks. After five rounds the scorecards were read: 48-47, 47-48, 48-47. St-Pierre won by split decision. Most fans in the arena and watching at home believed Hendricks had won. I use this fight in university presentations, not to criticise the judges, but to point out something far drier: these three men saw three different fights, and none of them broke the rules.

Why?

Because the criteria have an order of priority but no numerical weighting. "Effective striking" is criterion number one, but there is no quantitative definition of "effective." Is a punch that draws blood from the nose more effective than one that makes an opponent step back? Does a two-minute grappling control that creates no damage count more than three punches landing on the chin? The rules state the order. But the concrete in-ring question is always about relative weight, and relative weight is a subjective judgement.

Three judges, three physically different vantage points. A judge on the right side may see clearly the punch that snaps an opponent's head to the left. A judge on the opposite side may be blocked by the shoulder of that same fighter. Seating position, in a sport where the difference between a landed and a missed strike is measured in centimetres, is not a minor detail. It is a physical variable that cannot be eliminated by tightening the criteria.

The Data: What Actually Moves a Scoring Decision

During my coding period, I aggregated scoring situations and classified them by four variables: timing within the round, the current points gap, rounds remaining, and whether a fighter was bleeding or had been deducted a point.

The most notable finding was a compensation effect. After a fighter suffered a clearly adverse decision — an unjust deduction, a grappling sequence the referee did not count as control — the probability of that fighter receiving a favourable decision in subsequent rounds of the same fight, or in the following fight, rose considerably. In the sample I assembled, the rate hovered around 89%. That figure does not mean judges are deliberately correcting errors. It means human psychology has a tendency to self-balance, and in an environment where every decision is a continuous judgement, that tendency converts into points.

Data never commits a foul; the writer is the one who gets carded. I must be clear about the limits of this sample: it was built from fights whose footage I could access, within a limited time frame, with coding rules I set myself. It is not academic research, and I do not present it as such.

But it is enough to point out something all three current scoring systems try to conceal: a score is not a measurement. A score is a process. A process with sequence, with history, with memory. And a judge's memory of round one affects their round-three card in ways nobody controls.

A Variable Rarely Mentioned: Cognitive Load and the Judging Profession

One more thing public debate almost entirely ignores.

A mixed martial arts round lasts five minutes. A professional fight has three rounds. Across those three rounds, each judge must simultaneously track: foot position, strike direction, point of contact, control time, accumulated damage, and infringements. There is no break between exchanges. There is no replay function while the fight continues.

I once tried a simple exercise: watch a round at normal speed, then rewatch the same round in slow motion and count how many valid strikes I had missed on the first viewing. My result across three attempts was above twenty each time. I was sitting still in a chair, under no crowd pressure, with no shouting, under no obligation to decide. The three judges at the arena work under conditions far harder than mine.

This does not exempt them from responsibility for errors. It only places error in its proper context: a system that assigns three humans the task of processing an information stream exceeding human simultaneous processing capacity, then publishes the outcome as a verified fact.

Discipline is not punishment; discipline is a way of reading a fight. If we look at scoring errors as a form of loss caused by cognitive overload, the direction of reform looks entirely different from simply demanding that judges be "more serious."

The Counter-Intuitive Angle: Crowd Emotion Is a Poor Forecasting Model

This is the part I must write most carefully, because it runs against the instinct of the majority.

When a contentious decision happens, the community's first reaction is a sense of injustice. That sense is real, and I do not mean to deny it. But that sense is a very poor analytical instrument. It is based on the version of the fight the audience was shown — meaning the version cut together by the broadcast director — and then compares it against the version produced by three people sitting on three sides of the arena. Those two versions never match, because they are produced from two different data sources.

There is a phenomenon I call "the crowd scoring by slow motion." Audiences watch the replay three times, see the contact clearly, and reach a firm conclusion. But the judges in the arena had a fraction of a second to decide. This time asymmetry explains most controversies, and it also explains why publishing scorecards immediately after a fight does not soothe controversy — in many cases it intensifies it, because it turns an instantaneous judgement into a document that can be dissected indefinitely.

In the Korean martial arts scene I follow daily, this pressure carries an added layer: home-nation factors. At the 2026 World Cup in Qatar, my forecasting model was correct in 26 of 36 matches when I assumed referees would limit cards to preserve match flow. But it collapsed entirely in a match where the regional host nation was forced into a disadvantageous position, when I recorded 8 yellow cards and 2 penalties in a single match. The variable I ignored was not in the spreadsheet: the psychosocial pressure on the official when the stands are hostile. In martial arts, this variable is even stronger, because the shouting and echo of the arena change how a strike is felt.

The problem with the data layer is not that it lacks numbers. The problem is that it lacks the things that cannot be measured.

The Assumptions I Skipped

Every analysis of mine has a list of skipped assumptions, and I write it out rather than hiding it at the end of a piece as a disclaimer.

First, I have no access to the original scorecards with each judge's round-by-round notes. What is public is the final number, not the process that formed it.

Second, I cannot measure the specific influence of crowd noise on each individual judge. I can only infer from aggregate results, and inference is always weaker than data.

Third, I am a writer, not a scorer. Every conclusion of mine passes through a screen, and the screen is an intervening layer just like any other layer in this article. I am not standing outside the system. I am part of it.

Takeaway: If You Must Change the Rules, Start Here

If I could propose three changes to the organisations governing professional martial arts, I would not start by increasing the number of judges or expanding video review powers. I would start somewhere else.

First, publish round-by-round scorecards with brief reasoning, immediately after each round, during the break. Not for transparency, but to force judges to recognise that their judgement has structure and is not merely a feeling at the end of the fight.

Second, standardise judge seating positions on a rotation, so that no one is stuck permanently in the blind angle of an orthodox fighter.

Third, and most importantly, record the camera angle in the official report as part of the fight record. If images can be used to overturn a decision, the source of those images should face the same level of scrutiny as the decision they overturned.

None of these three proposals eliminates controversy. That is not the objective. The objective is for controversy to take place at its correct layer, instead of being dumped entirely on three people sitting on three sides of the arena — people who, in most cases, were only reading a fight whose version we were not permitted to watch alongside them.

The law is the only thing that never enters stoppage time. But the camera does.