Before There Was AI, There Was the Fencer’s Mind: The Hidden Logic of “Physical Chess”
Watch a fencing bout for the first time and you might see a blur: two fencers lunging, retreating, blades flashing faster than the eye can follow. It looks like pure reflex. But ask any coach why fencing earns the nickname “physical chess,” and they’ll tell you the real contest is happening a beat ahead of the blade — in the head.
The footwork and the point control are the easy elements to admire. The hard part, the part that decides bouts at every level, is the thinking. And that thinking turns out to look a lot like math — and, as we’ll see, a lot like how an AI language model reasons through a problem one step at a time.
The bout is an inference loop
In fencing, almost nothing happens in isolation. You don’t just “go”; you initiate an action “because” of what you or your opponent did last, and you choose your next move “because” of how they responded to your action. Coaches sometimes call this “counting”: you keep a running tally of what’s been tried and what worked, and each new action is built on the outcome of the one before it — or, more often, on the outcomes of several actions that came before it.
A caveat is worth making here. A rare few fencers don’t really “count” at all. Their feel for distance and timing and their physical acuity run ahead of any conscious tally — for them, instinct sits above the system, and the right move arrives before the logic does. But even that instinct is pattern recognition that became automatic — the system compiled into reflex, not an absence of it. This is especially true today, now that sabre fencing has grown far more physical and athletic than it once was, rewarding raw speed and reflex in ways the older, more thoughtful game did not. These fencers are the exception, though, not the rule; for almost everyone else, the thinking still comes first.
That structure has a name in mathematics and computer science: it’s a feedback loop with a stopping condition. The process is iterative — each step takes the previous result as its input, and the sequence repeats until something resolves it.
A fencing exchange does exactly that (simplified example):
- You launch a simple attack. Your opponent parries it.
- That outcome becomes your new starting point. So next time you “feint” toward the same parry to bait it, then deceive the blade and finish.
- Your opponent notices the pattern and adjusts — maybe they stop parrying and counterattack instead.
- Their adjustment is now “your new input,” and you reason one layer deeper.
Each move is a function of the move before it, and the sequence keeps running until a stopping condition ends it — a touch lands, the distance breaks, or the referee calls a halt. Without that terminating condition, you’d just feint forever. Sound familiar? It’s the same logic a student uses when they break a big problem into a smaller version of itself and keep going until they hit a stopping point.
It is also, oddly enough, how today’s AI language models work. A model like the ones behind modern chatbots is built from algorithms that generate text one piece at a time: it predicts the next word based on everything that came before, then feeds that word back in as new input and predicts the next, and the next, running this autoregressive loop until the response is complete. Just as a fencer chooses each move “because” of the one before it, the model treats every word it has already produced as the starting point for the one it produces next. The bout and the sentence are both built the same way — not in a single action, but as a sequence where each step depends on the result of the last.
Where the math actually shows up
Plenty of parents sign their kids up having bought into the promise that fencing is “physical chess,” and then feel frustrated when those same kids struggle and look confused in their first seasons. So it’s worth taking the label seriously, because it isn’t just a poetic flourish. The tactical layer of fencing draws on the same reasoning patterns that power modern AI:
Pattern recognition (alternation). A simple attack, a deception of the parry, a deception of the counter-parry — these alternate in a predictable rhythm. Knowing whether you’re on an “even” or “odd” beat of the exchange tells you whether to finish now or deceive one more time. That’s modular, cyclical thinking: tracking alternation through a repeating pattern.
Decision-tree search. “If they parry, I faint. If they retreat, I lengthen my lunge. If they counterattack, I shorten my attack.” A fencer’s plan is a branching tree of conditions — nested if-then statements, the exact structure students use in logic puzzles and early programming.
Prediction and expected value. Good fencers read tendencies. “After a feint, this opponent parries four about two times out of ten.” That’s estimating likelihood and choosing the action with the best expected payoff — informal statistics, played out in real time. In truth, every action on the strip carries a probability — each has odds attached, and strong fencers are constantly updating those odds with every exchange. Also, we’re wired for safety, and tricking the brain into believing this is a game rather than real life is not easy for a mind that’s always analyzing.
That said, the counting itself — and more importantly, the ability to use it during the bout — comes as youth fencers mature and gain experience. You may have a mathematical mind, but executing actions based on strategy and tactics takes more time and experience to develop. Commitment to your next action requires confidence, and confidence is built with time. This is where math-minded kids often feel frustration: the thinking is there long before the execution catches up, and overthinking remains the biggest obstacle in the meantime. But the longer they stay in the sport, the better they connect the two — which is why results matter far more in cadet and junior events than in youth tournaments.
Pattern recognition and disruption. Once you can see your opponent’s pattern, the highest-level move is sometimes to “break” your own. You deliberately defy the count they’re expecting. That’s reasoning about your opponent’s reasoning — a step up in abstraction, and a deeply mathematical one.
Adaptation. The best fencers don’t just react — they learn. Touch by touch, they update their read of an opponent over many bouts, the same way an AI model is trained on more data and improves. Within a single bout the adjustment is faster and looser: lose a point to a counterattack, and a sharp fencer adapts the very next exchange, refining the system in their head on the fly.
So why do math-minded kids tend to do well?
Because the cognitive machinery overlaps. A kid who’s comfortable with sequences, conditional logic, and thinking several steps ahead — “if this, then that, then what follows from that” — already owns the mental toolkit that the sport rewards most.
This matters for two reasons.
First, the benefit runs both ways. Fencing trains that reasoning under pressure, with a clock running and an opponent actively trying to outthink you. It’s abstract logic made fast, embodied, and consequential. Many parents notice their fencers grow more decisive and more comfortable adapting a plan on the fly — useful well beyond a math test. And the pressure is unusually compressed: the reaction time a fencer has under fire is far shorter than what a tennis player gets between shots, so all of that reading, deciding, and adjusting has to happen in a fraction of the window other sports allow.
Second, it’s genuinely encouraging for young people who love numbers and logic but assume sports aren’t “their thing.” Fencing offers them a physical arena where their strongest skill — structured reasoning — is the deciding advantage, not a sidebar. The kid who plans three moves ahead in chess or sees the pattern in a sequence problem can become a real threat on the strip.
An honest caveat
The analogy is real, but it’s a model — not the whole truth. Fencing also rewards instinct, timing, deception, and sheer competitive nerve. The footwork and blade skills are trainable for “anyone,” regardless of how they feel about algebra, and plenty of brilliant fencers would never call themselves “math people.” The point isn’t that you need math to fence. It’s that the way a math-minded kid already thinks gives them a natural foothold in the tactical game — and that fencing, in turn, sharpens exactly that kind of thinking.
So the next time someone calls fencing “physical chess,” take it literally. Beneath the speed and the steel, every bout is a prediction problem being solved one move at a time. And the kids who light up when they spot the pattern? They’re often the ones who come alive on the strip, too.
Keep prompting. The blade, like the model, only answers the question you feed it.
