You cannot command a man to walk into an enemy city, lie for months under a false name, and die without a word if he is caught. There is no rank that compels it and no drill that produces it. The work is dangerous, it is dishonorable by the standards of the literati who ran the state, and a single careless mouth can get the agent killed before he begins. So the Chinese military writers settled on a hard, unsentimental conclusion: you cannot order this work, you have to buy it — and you have to buy it well. Two things make an agent: you keep his mission absolutely secret, and you pay him more generously than you pay anyone else.1
The Ts'ao-lu Ching-lüeh puts the money first and bluntly. "If you want to gain knowledge of the enemy's situation, you must not begrudge the expense of a thousand pieces of gold. To be parsimonious about a thousand pieces of gold and lose agents is to be defeated. To expend a thousand pieces and thereby learn the enemy's situation is to be victorious."2 Stinginess on intelligence is not thrift; it is how you lose.
Two pillars hold up every agent operation, and the writers Sawyer assembles name them together over and over: secrecy and generous reward.3 They are not separate virtues. They are the two halves of the answer to one question — how do you get an honorable, intelligent person to do work that is deadly and shameful? You hide it so the danger does not become fatal, and you pay for it so the shame becomes worth carrying.
Sun-tzu set the bar and everyone after him cleared it: "Of all the Three Armies' affairs no relationship is closer than with spies, no rewards are more generous than those given to spies, no affairs are more secret than those pertaining to spies. If before the mission has begun, it has already been exposed, the spy and those he informed should all be put to death."4 Closest relationship, most generous reward, most extreme secrecy — the superlatives are deliberate. Spies are not a category among others; they are the limit case, the place where the army's normal rules of distance, frugality, and openness are all pushed to their maximum.
Behind both pillars sits something larger that the Warring States thinkers built deliberately: a worked-out psychology of rewards and punishments. The military and philosophical writers of that period "developed a complex psychology of rewards and punishments... that fully explored the power of incentives to motivate and manipulate men."5 Paying spies was not an isolated budget line. It was one application of a general theory about how incentives move people — the same theory that staffed the central government.6
Start with why the reward has to be lavish rather than fair. A fair wage buys ordinary work from ordinary motivation. Spy work runs against the agent's deepest constraints — his fear of death, his sense of honor, the contempt his own culture pours on the deceitful. The reward is not paying for his hours; it is paying to overpower those constraints. The text is explicit that the writers emphasized generous reward precisely "because of the risks undertaken by secret agents and the need for courage, determination, and self-control."7 The danger sets the price. The more abhorrent and lethal the task, the higher the incentive must climb to outweigh it. This is why incentives for spies are described in superlatives and not in proportions.
Now the secrecy, which is even more extreme than the secrecy a state already needs against its enemies. Counterintelligence requires keeping plans hidden; "the measures to be implemented with regard to running spy operations must be even more extreme."8 The reason is structural: exposure does not merely waste the mission, it kills the man, and it kills whoever he informed. Sun-tzu's death-penalty clause for premature leaks is not cruelty for its own sake — it is the recognition that the agent's life is the asset, and a leaked operation has already spent that life for nothing.9 The Wu-pei Chih-yao compresses the whole doctrine to a line: in employing spies "the main point is to prevent people from fathoming them, the subtle essence is secrecy."10
The deepest claim in the passage is economic, and it inverts how a ruler usually thinks about defense. The Kuan-tzu's "Administrative Distinctions" chapter: "The ruler does not begrudge gold and material wealth to the acute of hearing and sharp of eye. Thus a minor campaign occurring within one thousand kilometers is known about... A major campaign anywhere under Heaven will be known about within five days because the ruler disperses gold and material wealth in order to employ those who have acute hearing and sharp eyes. Thus one who excels at employing the military may lack ditches and ramparts because he has ears and eyes."11 Read that last line slowly. Information is a substitute for fortification. Gold spent on sharp-eared agents buys early warning, and early warning lets you do without the ditches and ramparts you would otherwise have to dig. The spy budget is not a cost; it is the cheapest wall you can build. Sawyer underlines the point: clandestine information reduces the state's military expenditures.12
So the two pillars resolve into one economic argument. You pay extravagantly for ears and eyes, and you guard them with absolute secrecy, because a network of well-paid, well-hidden informants is a better and cheaper defense than stone.
This page supplies the why underneath several others. The economic argument for spies — that a thousand pieces of gold spent well wins, spent stingily loses — gets its motivational engine here: the gold has to be lavish because the work is abhorrent, and it pays off because information substitutes for fortification. The recruitment and selection pages inherit a constraint from this one: you are not recruiting volunteers, you are buying reluctant talent, and the price is set by how much the task violates the recruit's honor and safety. The systematic-bribery page extends the same logic outward — bribery of enemy officials is the incentive psychology turned against the other side's people, using their desires the way you use your own agent's.
Take the Kuan-tzu line literally and run it as a budget decision. A ruler has a fixed sum. He can spend it on ditches and ramparts — physical fortification that sits in one place, costs continuously to maintain, and only tells him about an attack once it arrives at the wall. Or he can disperse the same gold among men with "acute hearing and sharp eyes" scattered across a thousand kilometers.13 The fortification is passive and local; the network is active and distributed. The wall learns of the enemy when the enemy is upon it. The network learns of a major campaign "anywhere under Heaven" within five days — before a single soldier has marched into range.14
The "Administrative Distinctions" claim is that the gold-financed network is the superior buy, to the point that an excellent commander "may lack ditches and ramparts because he has ears and eyes."15 This is a startling substitution. It treats early information as functionally equivalent to physical defense, and then argues the information is cheaper. The case rests on the incentive premise: the ears and eyes only stay sharp if the ruler "does not begrudge gold." Underpay them and the substitution collapses — a half-funded network goes deaf, and a deaf network is worse than a wall, because the ruler thinks he can see and cannot. The lavishness is not generosity; it is what keeps the cheaper wall standing.
A general sits with his quartermaster over the season's accounts. The quartermaster wants to thicken the eastern ramparts; the stone is bought, the labor is waiting. The general pushes the ledger back and asks a different question — how much gold is going to the men with sharp ears in the enemy's three border towns? The quartermaster names a figure and winces; it is a large figure for a handful of unseen men who produce nothing you can stand on. The general doubles it. He explains, quietly, because the explanation itself must not travel: those men are walking into death wearing false names, and if any one of them is caught the rope is real. A man does not do that for a clerk's wage. Pay him like the most valued soldier in the army, or he will not be there when you need to know whether the eastern road is filling with carts. And tell no one — not the quartermaster's deputy, not the men's own families — because a single loose word turns the whole budget into a row of fresh graves. The stone can wait. The ears cannot.
You are doing this wrong when you treat the spy budget as discretionary — the first line cut in a lean season. The Ts'ao-lu Ching-lüeh names the result: parsimony with the gold "is to be defeated."16 A network starved of reward goes quiet, then goes over to the higher bidder.
You are doing it wrong when you pay an agent the same as a soldier doing honest, survivable work. The reward has to outweigh death and dishonor; a fair wage cannot do that, and an underpaid agent is a half-motivated agent who will not hold under pressure.
You are doing it wrong when the operation is known to people who do not need to know it. Every additional person who knows is another mouth, and Sun-tzu's penalty for the leak falls on the agent first — he dies for someone else's careless word.17
And you are doing it wrong when you build stone but no ears — when you can feel an attack at the wall but cannot see it form a thousand kilometers away. The deaf, well-fortified state believes it is safe right up to the morning the carts arrive.
The passage's confidence has a soft edge it does not examine. Lavish reward buys motivation, but money is also exactly what the enemy can offer to turn the same agent. The incentive that recruits him is the incentive that can flip him; an agent loyal to gold is loyal to whoever has more of it. The text emphasizes paying generously without fully confronting that the most purchasable agent is also the most purchasable away. The corroboration and double-agent material elsewhere in Sawyer is the implicit patch, but on this page the tension sits unresolved.
A second open question: the Kuan-tzu's claim that information substitutes for fortification is asserted, not demonstrated. Five-day knowledge of a distant campaign is only worth a wall if the state can act on five days' warning. The substitution assumes a responsive command that can convert foreknowledge into defense in time — an assumption the passage leaves unstated.
Sawyer presents secrecy-and-reward as a doctrine of remarkable continuity, threading Sun-tzu, the Wu-pei Chih-yao, the Kuan-tzu, and the late Ts'ao-lu Ching-lüeh into one unbroken emphasis.18 The convergence he shows is real — every writer he quotes does stress both pillars. But the continuity claim deserves a careful eye. He is stitching together texts separated by many centuries and treating their shared emphasis as a single evolving doctrine. The Kuan-tzu's economic substitution argument and Sun-tzu's death-penalty secrecy clause are genuinely different in spirit — one is a budget theory, the other a security rule — and Sawyer's smooth assembly can make a chorus out of voices that were answering somewhat different questions. The shared insistence on generous reward is well attested across the sources; the seamless "all the military writers" framing is Sawyer's synthesis, and the cross-dynasty continuity it implies is worth verifying independently rather than taking on the strength of the juxtaposition.
Handshake with The Semblances Problem. The reward pillar quietly creates the problem the semblances page describes. An agent bought with gold gives you the appearance of loyalty — he shows up, he reports, he takes the money. But the appearance is rented, not owned. The same desire that makes him show for your gold makes him available to the enemy's larger purse, so the loyalty you can see is precisely the loyalty you cannot trust. The structural tension is sharp: the incentive that solves the recruitment problem manufactures the semblance problem. Neither page generates this alone — the incentive doctrine on its own reads as straightforward economics, and the semblances problem on its own reads as a general epistemology of appearance. Put together, they produce a specific and uncomfortable insight: every purchased agent is a standing semblance risk, and the more purely he is motivated by reward, the less his visible loyalty means. This is why the corroboration apparatus exists — it is the structural answer to the semblance that lavish reward inevitably buys.
Handshake with Confirmation Bias as Ancient Problem. The Kuan-tzu's "ears and eyes" argument assumes a ruler who will believe what his sharp-eared agents tell him even when it is unwelcome — early warning is worthless if it is dismissed. But a ruler who has spent a thousand pieces of gold on a network has a powerful motive to believe his network's reports are correct, simply because he paid so much for them. The expensive source feels authoritative because it was expensive. Holding the two pages together exposes a failure the incentive doctrine cannot see on its own: the lavishness that recruits good agents also biases the ruler toward over-trusting them, and a generously funded network can become an echo the ruler cannot bring himself to doubt. The insight neither domain produces alone is that the cure (paying enough for real ears) carries the seed of a cognitive disease (over-weighting the costly source), and that the substitution of gold for stone only holds if the ruler can stay skeptical of the very thing he paid most for.
Sharpest implication. You cannot conscript intelligence; you can only buy it, and you must buy it above market and hide it below sight — but the gold that buys an agent is the same gold that can buy him away, so reward and secrecy together produce loyalty you must never fully trust.
Generative questions.