Imagine a freelance strategist of the Warring States — a man like Su Ch'in or Chang Yi — who makes his living by traveling from court to court selling advice on grand strategy. This week he is dining with the king of Wei; last month it was Chao; before that, Ch'i. The powerful entertain him lavishly because his counsel might decide whether their state survives. And as he sits at one table after another, talking statecraft with one ruler after another, something accumulates in his head that no single ruler intended to give him: a working map of everyone's military strength.1
He did not steal it. Nobody handed it to him as a secret. It precipitated out of a hundred conversations — a general carelessly discussing martial affairs over wine, a minister boasting of his chariots, a count of mailed soldiers mentioned in passing to make a point about leverage.2 The persuader is a walking information market. He arrives at the next court already carrying, deliberately or inadvertently, the secrets of the last three.
The collapse of the old Chou nobility created exactly this class of man — disfranchised individuals "forced to seek employment through knowledge and talent," roaming the realm "seeking a receptive ear."3 They were hired for their ideas. They leaked as a side effect of being heard.
The peripatetic persuaders — itinerant strategists, professional advisers, the Vertical and Horizontal alliance men — were the Warring States' dominant intelligence conduit, both deliberate and inadvertent. Sawyer's framing is direct: their "importance as intelligence sources" came to exceed that of traditional defectors and exiles, because of "the information that might be easily elicited from professional advisers and peripatetic persuaders such as Su Ch'in and Chang Yi."4 A man who openly ventures from state to state, debating statecraft with rulers, "might easily divulge a critical plan to another state after simply having deduced its existence."5
What survives to prove it is a table. Sawyer reproduces a summary of mid-fourth-century force levels, "developed from persuasions exercised by Chang Yi and Su Ch'in" and preserved in the Chan-kuo Ts'e:6
This is order-of-battle intelligence — and it survives precisely because the persuaders carried it around and worked it into their pitches, where the historians later wrote it down.
The persuader leaks because of how he earns. His product is comparative strategic judgment — "ally with me against them," "their army is weaker than it looks," "Ch'in cannot hold the pass." To make that pitch credible, he must cite force levels. The numbers are not incidental to his persuasion; they are the persuasion. So every sale of advice is simultaneously a disclosure of intelligence, because the argument is the intelligence, spoken aloud to whichever ruler is buying.
This produces a one-way valve that leaks in every direction. The ruler of Wei tells the persuader things to win his counsel; the persuader carries them to Chao to win Chao's; Chao's reactions tell the persuader things he carries to Ch'i. No ruler can hire the persuader's knowledge without contributing to it. The more useful the adviser, the more thoroughly he has been briefed by everyone, and the more he leaks at the next stop.
Sawyer flags the crucial uncertainty in the numbers themselves: "Whether their numbers were mere guesses, calculated estimates, or precise figures remains an important but unanswered question."8 The persuader does not need exact counts. An experienced observer "could certainly have estimated total strength with basic facts about unit size and the number of armies."9 The leak works on estimates as well as on stolen ledgers — which is what makes it impossible to plug. You cannot stop a man from deducing your strength from what he saw at your gate and heard at your table.
This page hands the corpus its model of passive collection at scale. Most espionage pages turn on an active operation — a planted agent, a staged deception, a bribe. The peripatetic persuader is the opposite: an intelligence-gathering apparatus that runs with no operation at all, simply by existing and being welcome. It gives Sun Tzu's Five Categories of Spies a figure the five-spy taxonomy struggles to file — the man who is not your spy, not their spy, but a roving aggregator who sells to whoever listens.
It also supplies the raw material the deception programs spend. Systematic Bribery as Strategic Policy required knowing which ministers in which courts were corruptible and which alliances were forming — exactly the texture a peripatetic adviser accumulates. The persuader-network is the open-source layer beneath the covert layer; the bribery program is what a state does after the persuaders have told it where the seams are.
The force-level table is worth reading as a fossil — a hardened trace of a leak that was once fluid conversation.
Look at the asymmetries. Wei has 360,000 mailed soldiers — by far the largest figure — plus chariots and cavalry. Ch'in, the state that would eventually swallow everyone, is listed at only 100,000.10 If the table is even roughly accurate, it captures a moment before Ch'in's dominance was obvious, a snapshot from inside the uncertainty. The persuaders carrying these numbers did not know how the story ended. They were selling judgments based on a balance of power that the table preserves.
Now look at what the table omits. Sawyer notes "the composition and readiness of standing forces remain less clear."11 The persuaders captured headline strength — the number a ruler would cite to impress or intimidate — but not the harder, more decisive facts of training, supply, and morale. This is exactly the signature of leak-by-conversation: you absorb the figures men say out loud, not the operational realities they keep quiet or do not themselves fully know. The fossil records what was spoken at table, which is a biased sample of what was true in the field.
And look at how it survived. These numbers reached us because they were embedded in the recorded persuasions of Chang Yi and Su Ch'in — written into the Chan-kuo Ts'e as part of their arguments.12 The intelligence outlived the states it described because it was load-bearing in a sales pitch. The persuader's leak is, paradoxically, the reason Warring States order-of-battle is knowable at all.
You are a king of the Warring States, and a famous strategist has arrived at your gate. You want his counsel badly — it might save you. So you seat him at your finest table, pour the good wine, and let your generals talk freely to impress him, because a man who thinks your army strong will give you bolder advice. Over three nights you tell him, without quite meaning to, how many mailed soldiers you can field, how many chariots, where your border garrisons sit. You are buying his judgment with your secrets and you barely notice the price. When he leaves, satisfied and well-paid, he carries your order of battle in his head to the next court down the road — where another king will do exactly the same, and the strategist's map of the realm grows one state more complete. You have hired an oracle and fed him your throat.
You are exposed when your court mistakes being heard for being secure. The persuader is welcome, flattered, trusted — and every one of those conditions is what makes him porous. Warmth toward a roving adviser is leakage in disguise.
You are exposed when your generals perform their strength for a guest. The boast that means to intimidate is the disclosure that informs. A persuader does not need access to your archives if your commanders narrate the archives to him over dinner.
You are exposed when you assume secrecy requires theft. The persuader's most dangerous intelligence is deduced, not stolen — estimated from unit sizes and army counts a careful man can see without a single document. You cannot embargo what a guest can infer from your gate.
And the structural failure: trying to plug a leak that is identical to the service you are buying. The persuader's value to you and his porosity to your rivals are the same property. Refuse to tell him anything and his counsel to you is worthless; tell him enough to advise you well and you have armed everyone he visits next.
The numbers are the open wound. Sawyer himself will not vouch for them — "mere guesses, calculated estimates, or precise figures" is left explicitly unresolved.13 The table is invaluable as evidence that order-of-battle thinking existed and circulated, and unreliable as a literal census. Both things are true and must stay true on the page.
There is also a selection problem. We know about the persuaders' leaking because the leaks were preserved in their recorded persuasions. The advisers whose pitches were not written down — the majority — left no trace, which means the surviving table is a sample skewed toward the famous and the eloquent. Su Ch'in and Chang Yi are over-represented in the record precisely because they were good enough to be remembered.
Sawyer holds two claims that sit uneasily together. He argues the persuaders displaced defectors as the era's prime intelligence source — yet in the same passage he concedes that "prominent traitors, especially key generals, could still prove critical."14 The displacement is real but partial: the persuader became the better source for general strategic intelligence (force levels, alliance dynamics, court moods), while the defecting general remained the better source for specific operational intelligence and, increasingly, for the simple fact of his absence from the enemy. Sawyer's strong version — that advisers eclipsed traitors — overstates a shift that the evidence shows as a division of labor. Where he is firmly convincing is the social mechanism: the collapse of the Chou nobility manufactured a mobile, talent-selling class, and that class could not sell talent without leaking the knowledge that made it valuable. The convergence is sound; the ranking of source-types against each other is where his reading pushes past what the texts establish.
To The Difficulties of Persuasion. The persuader's leak and the persuader's danger are the same coin. The difficulties-of-persuasion corpus is about how perilous it is to advise a ruler — say the wrong thing and you die. This page shows the mirror peril facing the ruler: the very man you must open up to in order to be advised is the man who will carry your openness elsewhere. The handshake produces an insight neither page holds alone: persuasion in the Warring States was a mutual exposure. The adviser risks his life by speaking; the ruler risks his secrets by listening. Each party is the other's vulnerability. The persuader walks into a room where candor can kill him and walks out carrying candor that can kill the king. The transaction only functions because both sides accept a leak — of safety on one side, of intelligence on the other — as the cost of the exchange.
To Estrangement Techniques. Estrangement works by inserting a false belief into an enemy court; the persuader is the natural delivery vehicle, because he already has standing to speak strategy to rulers. The handshake reveals the pipeline: the same roving adviser who collects a court's secrets is positioned to plant the rumor that wrecks it. Sawyer notes the period's covert weapons were "used in combination rather than singularly" — subversion through bribes, disaffection through calumny, estrangement through rumor — all "facilitated by the emerging psychological and political analyses of court dynamics."15 The persuader is where collection and estrangement fuse into one body: he learns the seams on the way in and seeds the split on the way out. Neither the intelligence page nor the estrangement page, read alone, shows that the leak and the attack can ride the same traveling man.
Sharpest implication: The Warring States built its most effective intelligence network by accident — out of the men it hired for advice — and could not dismantle it without going deaf, because the leak and the counsel were the same service.
Generative questions: