In the small hours of August 1, 2004, on a stage in Pomona, California, a thin Japanese man named Daigo Umehara saved his tournament life with a piece of athletic perception so improbable that fighting game fans still speak of it the way cricket fans speak of Bradman’s last innings — quietly, and with the slightly dazed reverence reserved for things that probably shouldn’t have happened. His character had one pixel of health left. His opponent, Justin Wong, launched Chun-Li’s Houyoku-sen, a multi-hit super move that would, on contact or even on block, end the round. Umehara chose instead to parry — a technique that requires tapping forward at the precise instant a strike connects, within a window of six frames, or roughly a tenth of a second at sixty frames per second. He then did it fourteen more times in a row, against a move whose timing varies hit to hit, before launching a counter-combo that took the match. The crowd’s reaction is, by some margin, the loudest noise ever recorded in a competitive gaming venue.
To understand why grown adults still wear T-shirts commemorating a Tuesday night in a California university hall, one has to understand the strange unit of time that governs fighting games. These games run at sixty frames per second, which means each frame is 16.67 milliseconds long. Everything — every punch, every block, every step — is measured in these slivers. A jab might have three startup frames, a sweep nine, an overhead twenty-five. Professional players don’t merely know this data; they internalise it the way a sommelier internalises tannins. They can tell you, without thinking, that a move which leaves them at minus four on block can be punished by a four-frame normal but not a five-frame one, and they can tell you this while simultaneously executing the punish.
The slightly inconvenient fact lurking behind all this precision is that the human nervous system was not designed for it. The average human reaction time of about 265 milliseconds equates to roughly 16 frames in a 60fps game — meaning that any attack with a startup time under sixteen frames is, in the cold language of physiology, unreactable. You will be hit before your brain has finished deciding to move. The dirty secret of high-level fighting games is that almost nothing important is, in the laboratory sense, reacted to at all.
What is happening instead is closer to what happens in the mind of a top-order batsman or a chess grandmaster. Decades of research on chess perception, beginning with Chase and Simon in the 1970s, established that expert players process larger configurations during each fixation, acquiring memory representations for “chunks” of domain-specific visual information that are supplemented by larger memory structures called templates. The grandmaster does not see thirty-two pieces; she sees four or five familiar shapes. The fighting game professional, similarly, does not see a character lifting one leg by a few pixels; he sees “Chun-Li starting Hoyoku-sen,” a pattern he has watched ten thousand times and which, crucially, he has already begun reacting to before the move even exists in any objective sense.
This is prediction rather than reaction, and it turns out to be how most of the brain works most of the time. In the primary visual cortex, neurons fire particularly strongly when prediction errors occur — that is to say, the visual cortex is constantly forecasting what is about to appear and reporting back only the surprises. A professional player has spent years calibrating these forecasts to a very narrow domain. They are, in effect, watching a slightly more confident version of the future. When Umehara parried Wong’s super, he was not reacting fifteen times in a row to the impact animation; he was reacting once to Wong’s decision to commit to the move, and then executing a rehearsed rhythm. The hard part wasn’t the parrying. The hard part was knowing, before there was anything to know, that the super was coming.
There is a forum post about Street Fighter 6 that puts the point with unexpected elegance. “Even though everything you have ever reacted to in your life was 1/5 a second ago,” the writer observes, “you never noticed. We can literally measure it happening, but you don’t know it happens because your brain auto-compensates the timing differences.” We all live a fifth of a second in the past; the difference is that fighting game players have agreed, by professional convention, to care.
Which brings us to the bleaker side of all this — what happens when the equipment refuses to honour the bargain. If the entire edifice of competitive play rests on a sixty-frame-per-second contract between the human and the screen, then a violation of that contract is not merely annoying. It is a small theft from one player and a small gift to the other, redistributed silently. And it happens.
In October 2018, at a Dragon Ball FighterZ tournament held at the Thailand Game Show in Bangkok, a Filipino visitor named Alden Jacob lost the championship match to local player Peerapat “nickey_e” Ekpoothorn. Jacob explained afterwards that the monitor connected directly to the PlayStation 4 didn’t have input lag but the other one did; he raised the concern with the organisers and was told, with a flat managerial pragmatism that has since become infamous, that the players should just coin toss on who gets to play on the lagless monitor. Jacob had offered the obvious remedy — move both players to the working screen — and been refused. He played for the trophy on a display that was, in any meaningful sense, lying to him about what frame the game was on. He lost.
Two screens, identical to the untrained eye, can differ by a frame or two of latency, and a frame or two is the difference between a punish and a counter-hit, between a tournament won and a flight home. The fighting game community has been arguing about this for as long as flat-panel televisions have existed. “It’s daunting to know that minor differences in timing can cost you as much as $15,000 in a tournament, especially when you know it wasn’t really your fault,” one Mortal Kombat veteran told an interviewer some years ago, in the kind of sentence that makes equipment managers everywhere reach for the Scotch.
The problem has not gone away. At Evo Japan 2025, Victor “Punk” Woodley — winner of Evo 2024, one of the most decorated American players of his generation — placed an underwhelming seventeenth in Fatal Fury: City of the Wolves and went on social media to explain why. According to multiple competitors at the event, PlayStation 5 consoles were overheating, resulting in controller inputs not responding and frame rates dropping dangerously low. Punk’s complaint was unusually pointed for a man known otherwise for measured politeness: “PS5’s controlling the outcome of a set is literally the same feeling as someone spitting on you,” he wrote. He threatened to stop attending events if organisers continued to run them on the troubled hardware.
The complaint sounds, to outsiders, faintly absurd — a grown man undone by a warm console — until one remembers what is actually being claimed. Punk is not saying the game looked a bit choppy. He is saying that the contract between his finger and the screen had been quietly altered, that the sixty-frame heartbeat he had spent fifteen years synchronising himself to had skipped, and that a discipline which had been refined to the level of single-frame punishes had been rendered, for the duration, a kind of guessing game. Imagine asking a concert pianist to perform on a piano whose middle C occasionally produces no sound. It is not the loss that stings. It is the indignity of having one’s craft made suddenly meaningless by something so banal as thermal management.
What makes fighting games strange and beautiful as a spectator sport is that the action you are watching is happening on two different timescales at once. There is the visible drama — the dragon-punches, the projectile zoning, the desperation jumps in the final seconds — and there is the invisible one, in which two human beings are conducting a high-speed negotiation in sixteen-millisecond increments, each of them attempting to predict the other’s predictions about their predictions, all while their visual cortexes hum away forecasting tree-trunks and apples. The miracle is not that anyone wins. The miracle is that, on a good day, with a good monitor, the better player usually does.
Daigo Umehara, asked years later about the parry, has always been pleasingly unsentimental about it. He didn’t see fifteen kicks coming. He saw one. The rest was a piece of trained music, executed in a window of attention so narrow that, by every reasonable measure of the human brain, it should not have been there at all. That it was — that it is, every weekend, in basements and convention halls around the world — remains one of the more underrated arguments for what people can do with their minds when given a sufficiently strange thing to do with them.
