A boxer lets his right hand sag. To the man across from him it looks like a mistake — an opening, a tired arm falling. So he steps in to punish it. The instant he commits, the sagging hand comes up as a hook and the whole exchange reverses. Here is the question that the Questions and Replies spends pages chewing on: was that dropped guard a trick, or a real lapse that got lucky? And the deeper one behind it — could you have told the difference before the hook landed? In a real fight, a dropped guard and a feint look identical from the outside. What separates them is not the gesture. It is whether the boxer still had control of his feet while it happened.1
That distinction — between a move that looks like collapse and a collapse that is one — is the spine of this page. The Questions and Replies (the dialogue between the T'ang emperor T'ai-tsung and his general Li Ching, one of the Seven Military Classics, probably written in the tenth century) holds the most extensive discussion of unorthodox theory anywhere in the Chinese corpus.1 Its central move is to demolish the comfortable rule — "first be orthodox, then unorthodox" — and replace it with something far less manageable: the labels are decided live, in contact, by the commander, and they keep trading places while the fight is still going.
The page that grounds the whole vocabulary, The Zheng/Qi Duality, makes the case that the orthodox (zheng, the force you fix on) and the unorthodox (qi, the force that wins) are not fixed kinds of troops but designations that live in the enemy's mind. This page is where that claim gets argued out at full length, by two men who actually ran armies.1
The Questions and Replies takes apart two slogans that had hardened into doctrine. The first: "First be orthodox, and afterward unorthodox." The second: "When troops advance to the front it is orthodox; when they retreat to the rear it's unorthodox."1 Both are tidy. Both are, in Li Ching's hands, exposed as too simple to survive a real battle.
His instrument for breaking them is a single engagement — the battle at Huo-yi, where T'ai-tsung destroyed Sung Lao-sheng. When the fronts clashed, the Army of the Right "withdrew somewhat." T'ai-tsung then led elite cavalry down from the southern plain, cut across, severed Lao-sheng's troops from the rear, and captured him.1 T'ai-tsung's own question is the hinge of the page: were those orthodox troops, or unorthodox? He is genuinely unsure — because the withdrawal nearly turned into a rout. "At that time our slight withdrawal almost defeated our great enterprise, so how can you refer to it as unorthodox?"1
Li Ching's answer reframes the whole event. The withdrawal was the bait. "If the Army of the Right had not withdrawn somewhat, how could you have gotten Lao-sheng to come forward?" Lao-sheng, brave and impatient, saw a retreating enemy and chased it — and chasing it is what got his rear severed. "This," Li Ching says, "is what is referred to as using the unorthodox as the orthodox."1 The near-collapse and the trap were the same object, seen from two sides.
The engine of this page is a test you can apply in the middle of the fight. T'ai-tsung asks the obvious follow-up: can every withdrawal be called unorthodox? If a retreating army is automatically running a clever play, the word "unorthodox" means nothing.1
Li Ching says no — and gives the discriminator that makes this page matter. A withdrawal is a true defeat when "the flags [are] confused and disordered, the beating of the large and small drums not responsive, and their orders shouted out clamorously." A withdrawal is unorthodox strategy when "the flags are ordered, the drums respond to each other, and the commands and orders seem unified, then even though they may be retreating and running."1
Read that carefully. The men are running in both cases. The footwork — flags, drums, the chain of command — is the only thing that tells a feint from a rout. The body is moving backward either way; what differs is whether the brain is still attached to the body. A real break is the moment communication fails. A staged break is communication holding firm underneath a surface that looks broken. This is why a dropped guard and a feint look the same from across the ring: you cannot see the feet from there. You can only see the falling hand.
Out of this falls the question that, in Sawyer's account, baffled commentators for centuries: are the orthodox and unorthodox distinguished beforehand or determined at the time of battle?1 Li Ching first hands T'ai-tsung the pre-designation school. He quotes the Hsin Shu — Ts'ao Ts'ao's New Book: "If you outnumber the enemy two to one, divide your troops into two, with one section being orthodox, and one section being unorthodox. If you outnumber the enemy five to one, then three sections should be orthodox and two sections unorthodox."1 Clean arithmetic. Assign the roles before you march.
Then he detonates it with Sun-tzu. "The changes of the unorthodox and orthodox cannot be completely exhausted! The unorthodox and orthodox mutually produce each other, just like an endless cycle. Who can exhaust them?" And the verdict: "So how can a distinction be made beforehand?"1 If the two are a wheel that never stops turning, then any number you stamp on your troops at dawn is obsolete by noon.
Li Ching does not leave Ts'ao Ts'ao humiliated. He supplies the saving context: the pre-assignment "is merely for the purpose of instruction." You divide troops into fixed orthodox and unorthodox blocks to teach them — so green soldiers learn the flags and drums, learn to divide and combine on command. But once trained, "they can be raced about like a flock of sheep, following wherever their general points." At that point the fixed labels dissolve. "By determining the changes at the moment of battle, they become inexhaustible."1 T'ai-tsung's gloss seals it: "Ts'ao Ts'ao must have known it but the Hsin Shu just teaches what he conveyed to his generals, not the fundamental method."1
The deepest line in the exchange is the one that closes the Huo-yi discussion. T'ai-tsung asks whether the lucky withdrawal was "a question of Heaven" — fate, accident — and whether capturing Lao-sheng was "due to the efforts of man." Li Ching: "Unorthodox and orthodox lie with man, that's all. Because they changed in spirit-like fashion, they are attributed to Heaven."1 The transformation looks like magic from the outside, so people credit Heaven. But it was the commander reading the enemy and deciding live. There is no fate in it. There is only a man who kept his drums responding while his front fell back.
This page is the load-bearing argument under The Zheng/Qi Duality's "decide-live, never pre-designate" position — the duality page names the fork; this page walks both sides of it with the actual transcript. It feeds:
The case the dialogue itself reaches for is its own cleanest demonstration, so follow it as Li Ching tells it. At Huo-yi the army "had been mobilized out of righteousness, so it was orthodox."1 The cause was just; the deployment was the expected, head-on kind. Then Prince Chien-ch'eng fell off his horse and the Army of the Right "withdrew somewhat." That stumble — a genuine accident in its origin — is the seed of everything.1
Lao-sheng read the withdrawal the way a brave, undisciplined commander reads any retreat: as an advantage to be seized. The Liu-t'ao line Sawyer quotes catches exactly the appetite Lao-sheng was acting on — "One who excels in warfare will not lose an advantage when he perceives it."1 So Lao-sheng raced forward. "He relied upon courage and made a hasty advance. He didn't anticipate his rear being severed."1 T'ai-tsung's cavalry came down from the southern plain and cut the trap shut.
Now watch what Li Ching does with the labels. The withdrawal started as a near-rout — Chien-ch'eng on the ground, the line bending, T'ai-tsung's "great enterprise" nearly lost.1 It became the bait the instant Lao-sheng chased it. Nobody in the T'ang command planned the prince's fall. The unorthodox character of the move was not built into it at dawn; it was conferred on it, live, by what the enemy did in response. The same backward motion was a defeat and a lure depending on one thing: whether the T'ang flags and drums held while it happened. They held. So it was a feint. Had they not held, the very same withdrawal would have been the rout it nearly was — and Sawyer's two slogans, "first orthodox then unorthodox" and "advance is orthodox, retreat is unorthodox," would both have been useless to predict which.1 Li Ching's verdict: "using the unorthodox as the orthodox."1 The commander out-read the enemy. Heaven got the credit.
You are losing ground in a negotiation, or a campaign, or a contest where the other side is watching you give way. You feel the pull toward the two questions everyone asks — is this going badly, or is this my plan? Stop. Those are the wrong questions, because from the outside a retreat and a feint look the same, and you are on the inside where you can check the one thing that distinguishes them.
You run the drum-test. You ask: are my flags ordered? Are my drums responding to each other? Is my chain of command still unified while I give ground? If yes — if your people are coordinated, if the giving-way is happening on purpose and everyone executing it knows the next beat — then you are not retreating, you are baiting, and you let the other side lean in. You watch them commit to your sagging hand. The moment they overextend to punish the opening you offered, you bring the cavalry down from the southern plain. You sever their rear. That is the hook coming up.
But if your flags are confused, your drums silent, your orders shouted in a clamor — then this is not unorthodox strategy. This is the thing it resembles. You are in a real rout, and dressing it up in the language of "it was all part of the plan" is exactly the self-deception that gets armies destroyed. You do not pretend the wallet-lift was intended. You stabilize the line or you break clean.
And you refuse, all the way through, to have decided in advance which this was. You did not stamp "feint" on the withdrawal at dawn. You held the label open and let the enemy's response, plus your own coordination, assign it in the moment. That is what "it lies with man, not Heaven" means with your back against the table: the decisive quality of your retreat is something you confer live, by reading him and keeping your own drums responsive — not something you scheduled.
This doctrine fails in three recognizable ways:
The evidence is the dialogue itself, preserved in the Seven Military Classics, and Sawyer's reconstruction of how its arguments sit against the earlier theorists it quotes — Sun-tzu, the Ssu-ma Fa, Wu Ch'i, the Liu-t'ao, Ts'ao Ts'ao's Hsin Shu.1
The central tension is unresolved and should stay that way: pre-designate versus decide-live. Ts'ao Ts'ao, running real armies, wanted the split fixed before the fight so the machine could be commanded. Li Ching insists the split must float until contact or it stops being unorthodox at all. Li Ching's "ameliorating context" — pre-assignment is only for training — is generous, and Sawyer notes the generosity may be partly face-saving, since "the Hsin Shu fragment undermines Ts'ao Ts'ao's reputation for strategic acumen."1 So the reconciliation is real but politically convenient, and a skeptic can hear Li Ching smoothing over a genuine doctrinal defeat rather than a misunderstanding.
🚩 Bias flag — Sawyer's pro-Chinese-distinctiveness lean. Sawyer's larger project treats the unorthodox/orthodox cycle as a uniquely Chinese, uniquely sophisticated theory of war. The drum-test distinction between feint and rout is not, on its face, culturally special — every tradition that ever used a feigned retreat had to solve the same problem of telling a staged break from a real one. Reading this page, hold the difference between the historical fact (Li Ching articulated a clean discriminator in the tenth century) and Sawyer's framing (that this represents a depth Western thought lacked). The first is well-sourced; the second is the author's thesis, and the feigned-retreat problem was hardly China's alone.
A genuine open question: the Huo-yi event "was not a deliberate ploy but accidentally unfolded."1 Chien-ch'eng fell off his horse. Li Ching retroactively classifies the lucky accident as unorthodox strategy. Is that legitimate analysis — the unorthodox is whatever the enemy can't fathom, regardless of intent — or is it a theorist laundering an accident into doctrine after the fact?
Only Sawyer is cited here, but the page is built from his staging of a three-way argument among Chinese theorists, and it inherits that argument's friction. The sharpest pairing is Ts'ao Ts'ao versus Li Ching, mediated by Sun-tzu, whom both claim.
Ts'ao Ts'ao the field commander wants designations he can command: two-to-one, split your force; five-to-one, three orthodox and two unorthodox. Li Ching the theorist wants designations that float free until contact. Both quote Sun-tzu's endless-cycle line, and that is the strange convergence — they agree on the metaphysics (the wheel turns without end) and split entirely on the practice (so do you stamp roles on troops, or not?). The tension reveals something neither says outright: a doctrine is maximally lethal exactly when it is least teachable. You can only train soldiers by giving them fixed roles and drills, which is precisely the readable, commandable structure that kills qi. Ts'ao Ts'ao's arithmetic is how you build an army that can be unorthodox; Li Ching's fluidity is what that army must shed the instant the fighting starts. The boxer drills a thousand fixed combinations so that, in the ring, he can throw the one his opponent didn't drill against. Li Ching's "merely for instruction" is the seam where the drilled becomes the improvised — and the whole page lives on that seam.
The problem this page solves — telling a staged break from a real one — recurs anywhere one mind acts in front of another mind that is reading it back. The drum-test is a detection problem and a production problem at once, and both sides of it have rich counterparts outside Chinese military history.
These three together make the scale-invariance argument: the same detect-or-produce-a-false-break logic runs at the level of two swordsmen, two negotiators, and two armies, which is the strongest sign it is a logic of minds reading minds, not a fact about T'ang cavalry.
The Sharpest Implication If the unorthodox character of your move is conferred live by the enemy's reaction — and not built into the move at dawn — then you cannot know in advance whether your retreat is a feint. You can only know whether your flags are ordered. The feint-ness itself is decided in the enemy's response, which you don't control. This is brutal for any planner: the one quality that makes your withdrawal a weapon rather than a defeat is located partly inside the other commander's head, and the only test is contact. You manage the half you own — keep the drums responsive — and gamble on the half you don't.
Generative Questions