A Southern Sung border officer watches his own army fall apart from the inside. The nomadic enemy has done something simple and devastating: they have copied his flags and pennants exactly.1 Now his soldiers see a unit advancing under familiar colors and take it for friendly — and it is the enemy.1 Worse, these peoples know his army's whole signaling organization so thoroughly that, as the Ts'ui-wei Pei-cheng Lu puts it, "our armies are constantly betraying themselves because their structure is not secret."1 The drums, the gongs, the colored flags — the entire language by which a Chinese general commanded his troops — has been read by the other side and turned into a weapon.
That is the problem this page is about. A battlefield command system has to be loud and clear to work; a general "commanded rather than led," directing movement by drums and gongs and direction by colored pennants, and to cut through the din the writers insisted the drums multiply and sound in unison so every soldier responded.2 But the louder and clearer you make the signal for your own men, the easier the enemy hears and learns it too.3 Visibility is the whole point and the whole vulnerability.
This is the open command-signaling system of the Chinese army — and the doctrine for what to do once the enemy has learned to read it. From antiquity, China's forces were directed by signal flags, drums, and gongs, which let them act in segmented, articulated formations as early as the Spring and Autumn period.2 The general controlled the drums that signaled type and speed of movement and the battle pennants that signaled direction and which units should respond.2 This is the opposite of the secret-communications problem: here you want maximal clarity, because the message has to reach soldiers in the heat and noise of combat.4
The counter-leak half comes from the "Countering Leaks" chapter of the Southern Sung Ts'ui-wei Pei-cheng Lu. Its premise is a clean strategic symmetry: "when we maintain security while the enemy leaks information, victory will always be ours. Conversely, when the enemy is secretive while we leak, victory will always be theirs."5 Once the enemy has adopted your flags and learned your structure, the answer is not to abandon signaling — you cannot — but to rotate the codes so the language they learned no longer means what it meant.1
The Ts'ui-wei Pei-cheng Lu gives four methods for countering leaks, and each one works the same way: take a signal the enemy has learned and reassign its meaning.6
First, "summoning by signals." A green flag formerly directed generals and commanders; now a green flag summons ordinary officers and troops. A white flag formerly summoned the commander in chief; now it summons divisional commanders.7 The flag the enemy learned to read as "the top commander is being called" now calls someone else.
Second, "flags and pennants." Green formerly signaled the left flank, white the right; now exchange them, so the enemy who knows your old designations misreads which wing is which. Green formerly meant a straight formation, black a curved one; swap black and green so the enemy cannot tell whether your strategic power will deploy straight or curved.8
Third, "gongs and drums." This is the sharpest inversion. The army formerly advanced to the drums and retreated to the gongs — the most basic convention in the system. Now reverse it: when the drums sound, the army halts; where it formerly halted at the gongs, it now advances to them.9 The enemy who has learned that drums mean "they are coming" now watches them halt.
Fourth, "beacon fires." One torch formerly signified enemy invaders, two torches a request for help; now invert — one torch signals a request for help. No smoke formerly meant no incidents, smoke a warning; now the presence of smoke signifies the absence of incidents.10 The fire-codes the enemy decoded now read backward.
The mechanism in every case is the same arms-race move: the signal stays visible, but its meaning is rotated out from under the eavesdropper. The enemy's hard-won knowledge of your system becomes a liability the moment you flip the key.
The dynamic, adversarial face of communications. Where Secure Military Communications and Secret Tallies and Fragmented Letters build codes meant to stay secret, this page admits the code will be broken — the enemy will copy your flags and learn your drums — and supplies the response: rotate. It turns communications security from a static fortress into a live game where the move is always the next reassignment.
It also feeds Contrary Employment Doctrine a concrete instance. The drum that means "halt" instead of "advance" is the contrary-employment principle applied to your own signal language — deliberately doing the opposite of what is expected, here against an enemy who expects to read you. The page shows that contrary employment isn't only about troop movements; it's a technique you can run on your own communications.
Take the third method — advancing to the gongs and halting at the drums — and feel why it is both the cleverest and the most dangerous of the four.9
The drum-advance, gong-retreat convention is the most deeply trained reflex in the army. Soldiers do not think when the drum sounds; they move forward, because that is what every drill has built into them. That is exactly what makes it valuable to an eavesdropping enemy: hear the drums, and you know the Chinese are advancing. So reversing it is a beautiful trap. The enemy hears the drums, prepares for an advance that does not come, and watches the line halt instead — and if you have timed it, you advance to the gongs while they brace for a retreat.9
But now read the line Sawyer drops at the end, the one that turns the whole chapter sour: "Whether these countermeasures would deceive the enemy or one's own troops more thoroughly may well have been problematic."11 There is the danger. The drum-advance reflex is trained into your own soldiers too. Reverse it, and you are asking men in the chaos and terror of battle to override their deepest conditioning on the day it matters most. The enemy had time to learn your old code; your own troops had years to internalize it. A signal reversal that confuses the enemy for one battle may confuse your own line into advancing when you signaled halt, or freezing when you signaled charge. The counter-leak measure designed to exploit the enemy's knowledge may detonate on the very reflexes it depends on. The case study's lesson is that signal security and command reliability are at war: the more you scramble the meaning to fool the listener, the more you risk un-training the men who must obey it.
You are the general the night before contact. Your scouts have confirmed it: the enemy is signaling with your colors, advancing units under your green and white, and your men have already been fooled once.1 You cannot stop using flags — there is no commanding a battle line in silence. So you gather your divisional commanders and you rotate the language.
You tell them green now summons the troops, not the generals; white now calls the divisional commanders, not the chief.7 You swap the flank colors — green is the right now, white the left — and you swap straight and curved.8 Then you reach the hard one. Tonight the drums mean halt and the gongs mean advance.9 You watch the faces of men who have spent their lives advancing to the drum, and you know the line you are walking: by tomorrow this reversal either turns the enemy's stolen knowledge into their grave, or it scatters your own front when a frightened soldier obeys his trained reflex instead of your new order. You drill it, hard, all night, because the only thing that makes a reversed code safe is making the new reflex deeper than the old one before dawn. You are not just deceiving the enemy. You are racing to re-train your own army faster than the battle arrives.
It fails in the gap between the enemy's reaction time and your own troops'. Reverse a signal and you create a window where two armies are confused — and the question Sawyer raises is which army is confused more.11 If your soldiers' old conditioning is stronger than your hasty re-drilling, the reversal hurts you. The counter-leak measure that looks like cleverness on the planning table can read as chaos on the line.
It also fails by escalation. Counter-leak is an arms race with no terminal state. You rotate the codes; the enemy, who learned the first set, eventually learns the rotation; you must rotate again. Each rotation costs your own troops a re-training and a moment of vulnerability. An army that has to flip its signal language repeatedly is an army whose command system is permanently half-trusted by its own men — every drumbeat now carries a flicker of "but which code are we on?" The deepest failure is that secrecy and clarity cannot both be maximized: the signal loud enough to command in battle is loud enough to be learned, and the only fixes degrade either the secrecy or the clarity. There is no settling point, only the next move.
The honest tension is the one Sawyer himself plants: he records the four elegant counter-leak methods and then immediately doubts whether they fooled the enemy or one's own troops more.11 That doubt is the most valuable thing on the page, and it should not be resolved. The Ts'ui-wei Pei-cheng Lu presents the rotations as solutions; Sawyer treats them as solutions that may have been self-defeating. Both readings stand.
The historical frame is specific and worth keeping: this is Southern Sung doctrine responding to nomadic enemies who had adopted Chinese flags and learned Chinese military organization.1 It is born of a real, named failure — Chinese armies "constantly betraying themselves" at the frontier.1 That specificity is a strength; it also means the doctrine is a reaction to one strategic situation, not necessarily a universal Chinese practice, and projecting it across other periods would overreach.
Open question the source leaves: the four methods all assume you can re-train your troops to the new code faster than the enemy can re-learn it. Sawyer never tells us whether any army actually pulled this off in battle, or whether the "problematic" doubt was borne out — the doctrine is prescription, and its battlefield record is absent here.
Sawyer's framing and his sources mostly converge — the Ts'ui-wei Pei-cheng Lu offers counter-leak as the answer to signal compromise, and Sawyer presents it faithfully. The single sharpest divergence is his closing aside, where he steps back from the text's confidence to ask whether the cure was worse than the disease.11 That aside is Sawyer the skeptic pulling against his source's prescriptive certainty, and it is the page's live nerve. The Sung author writes as if reversing the drums simply works; Sawyer, reading across the practical reality of trained reflexes, isn't sure it does. Preserve that gap. The text says "here is the fix"; the translator says "this fix may have broken the wrong army." Neither is wrong, and the tension is the point — it is the difference between doctrine on the page and doctrine in the noise and terror of a real line, where men obey their conditioning before they obey the latest order.
Handshake with Secure Military Communications. These pages are the static and dynamic faces of the same problem, and together they expose a tension neither resolves alone. Secure communications builds codes meant to stay secret — invisible ink, split keys, blinded couriers. This page concedes the battlefield code cannot stay secret, because it must be loud enough to command men in combat, and answers compromise with rotation rather than concealment. The structural insight is that signal security splits by audience reachability: a message to a distant ally can be hidden, but a message to your own front line must be broadcast, and broadcast cannot be hidden — only periodically reassigned. The two pages map the full spectrum: at one end, hide the message so thoroughly the enemy never sees it; at the other, show the message to everyone and change what it means. The same army runs both regimes simultaneously, and the insight neither generates alone is that "communications security" is not one discipline but two opposite ones, sorted by whether the recipient can be reached in private.
Handshake with Contrary Employment Doctrine. Contrary employment is the principle of doing the opposite of what the enemy expects; this page is that principle turned inward onto your own signal language. Reversing the drums so they mean halt, inverting the beacon fires so smoke means safety — these are contrary employment applied not to troop movements but to the codes that govern them.9 The structural parallel is exact: both exploit the enemy's confident expectation, and both depend on the enemy having learned the normal pattern well enough to be trapped by its reversal. But this page surfaces a cost that the troop-movement version of contrary employment hides: when you reverse a signal, you also reverse it for your own men, who were trained on the same normal pattern.11 The insight neither page generates alone is that contrary employment is cheap when the expectation you invert lives only in the enemy's mind, and expensive when it also lives as a trained reflex in your own troops — feinting a march costs you nothing, but reversing the drum that orders the march costs you your own army's conditioning. Contrary employment has a hidden price that scales with how deeply your own side has internalized the thing you're inverting.
Sharpest implication. The signal that is clear enough to command an army in battle is, for that exact reason, clear enough for the enemy to learn — so battlefield signaling can never be made secret, only periodically reversed, and every reversal risks confusing your own conditioned troops more than the enemy. Security and command reliability are not allies here; they are competitors trading off against each other on the day it matters most.
Generative questions.