A late-Ming officer named Chieh Hsüan looked at the old word for spy and decided it was too small. For two thousand years chien had meant a person — somebody you sent into the enemy's house to come back and tell you things. Chieh Hsüan said: a rumor can do that. A song can do that. A bribe slipped into the women's quarters can do that. A forged letter sitting on a general's desk can do that, and the letter never has to come home. So in his manual the Ping-fa Pai-yen he stopped counting agents as people and started counting them as channels — sixteen of them — every distinct way you can reach into an enemy and move him without facing him.1
That move sounds like wordplay until you sit with it. He is saying the spy was never really the point. The point was always the leverage. A human agent is just one delivery vehicle for leverage; a circulating prophecy is another; a sack of gold is another. Once you see it that way, "espionage" stops being a roster of brave men in disguise and becomes a menu of pressure points. The translator Sawyer notes the awkwardness this creates: there is no clean English for it. You almost have to render each item as "agency through" — clandestine action through writing, through rumor, through song.2
Chieh Hsüan's contribution is a typology of covert leverage. He keeps the inherited frame — agents exist to "strike fear in the enemy's general staff, slay the enemy's beloved generals, and cause chaos in the enemy's estimates and strategies" — but then he explodes the methods column into sixteen entries.3 His own list runs: living agents, dead (expendable) agents, written, civil, rumors, prophecy, songs, bribes, things, rank, the enemy, villagers (local guides), friends, women, goodwill, and awe.4
Read the explanations and you see he is mixing three different kinds of thing under one roof, which is exactly what makes the list interesting. Some entries are people-types in the old sense: living agents go and return; dead agents go and do not.5 Some are media: forged letters are clandestine action through writing; stupefying discussions are clandestine action through civil affairs; verses circulated among the people are clandestine action through prophecy; the dispersing songs once used against Ch'u are clandestine action through song.6 And some are inducements — the currency you pay an enemy insider in: ten thousand ounces of gold (bribery), a promised title (rank), seized or gifted objects (things), the pull of a man's friends (friends), an exploited personal bond (goodwill), and bodily threat (awe).7 The women's-quarters entry sits across two of these at once: it is both a channel into a household and a description of who carries the bribe.8
The closing line is the one that gives the whole list its spine. "It is not that implementing clandestine activities is difficult, but employing men that is difficult. Therefore employing agents is more difficult than employing the army."9 Sixteen channels, and the bottleneck is none of them — the bottleneck is the human judgment that decides which channel to open and who to trust with it.
Walk down the list and a single logic shows itself: every entry is a way of getting the enemy to act against himself while believing he is acting for himself. A forged letter works because the general reads it as his own intelligence. A prophetic verse works because the people repeat it as their own wisdom. A bribe through rank works because the traitor experiences his betrayal as ambition rewarded. The agency is always through something the target already trusts — his correspondence, his folk songs, his sense of his own future. Chieh Hsüan's preposition is doing real conceptual work. "Through" names the carrier the deception rides on.
This is why he can call an inanimate object an agent. A planted document has no loyalty to test, no nerve to fail, no face to give it away under interrogation — and yet it penetrates the enemy's decision loop exactly as a recruited insider would. The object inherits the function of the spy without inheriting the spy's fragility. What it cannot inherit is the spy's judgment, which is why the whole apparatus still collapses back onto the one hard problem: a person has to choose the channel, write the letter, compose the verse, decide which courtier is bribable. The sixteen agencies multiply the ways in; they do nothing to multiply the wisdom required to use them. That asymmetry is the manual's real teaching.
There is a tidy irony Chieh Hsüan himself flags. Having built this elaborate sixteen-fold scheme, he concedes it cannot finally be harmonized — sorting each agent by living-or-dead and then again by whether he was moved through benevolence or fear gets "too complex." So, he writes, Sun-tzu was right to collapse everything into a single chapter of five types.10 He builds the cathedral and then admits the old hut was more livable. Sawyer agrees the categories fold back under Sun-tzu's five rubrics, most of them amounting to local guides plus an explicit listing of the lures used to recruit them.11
For the vault's intelligence corpus this page is the bridge between the human-agent tradition and the operations tradition. Anywhere the vault treats deception as a tool — disinformation, planted documents, rumor warfare — Chieh Hsüan supplies the unifying claim that all of these are the same act viewed through different carriers. It lets a reader stop asking "is this a spy story or a deception story?" and start asking "which channel is the leverage riding?"
It also hands Systematic Covert Programs a vocabulary. A state running a coordinated covert apparatus is, in Chieh Hsüan's terms, running several agencies at once — gold into one court, a forged letter into another, a song into a third — and the difficulty of "employing men" is precisely the management problem that a program exists to solve.
Take the single entry most foreign to the older tradition — "creating forged letters is clandestine written action" — and follow what Chieh Hsüan is actually claiming.12 In the inherited Sun-tzu scheme, if you wanted to feed a false plan into an enemy command you needed an expendable agent: a human being primed with the false information, sent in, allowed to be caught, executed when the deception surfaced. The cost of the lie was a life.
Chieh Hsüan's written agency cuts the human out of the loop. The forged letter carries the same false plan, penetrates the same decision-making staff, produces the same "chaos in the enemy's estimates" — and there is no courier to break under questioning, no body to recover, no face that might be recognized. The letter is the dead agent's payload without the dead agent. And yet notice what does not change: somebody had to know the enemy general well enough to forge a letter he would believe, had to know his hand, his idiom, his current anxieties, had to arrange for the document to surface where it would be read as genuine and not as a plant. The forgery removes the courier's risk and relocates the entire difficulty into the forger's judgment. This is the sixteen-agencies thesis in miniature: the channel is easy to name, ruinous to use badly, and only as good as the man choosing it. It is exactly why the manual ends not on a method but on a warning about employing men.
It is past the evening meal and the officer is at his writing table with three of his staff. He does not say "send a spy." He asks a different question, the Chieh Hsüan question: which way into this man is open? One of his staff knows the enemy general is vain about a promotion he was denied — that is an agency through rank, dangle the title his own court withheld. Another knows the enemy's soldiers sing a homesick tune from the southern marches — that is an agency through song, seed it through the camps until the ranks ache to go home. A third has a man who served as a clerk in the enemy's secretariat and knows the general's hand — that is an agency through writing, and a letter can be made.
The officer does not pick one and stop. He weighs them against the harder column. The song costs nothing and risks nothing but may do nothing. The forged letter could shatter the enemy's plan in a night — or expose his own clerk and warn the enemy that his correspondence is compromised. The promised rank is the deepest hook and the longest fuse. He is not choosing a tactic. He is doing the thing the manual says is harder than commanding the army: deciding which channel a particular enemy, this week, with this temperament, can actually be moved through — and which of his own men can be trusted to carry it. He keeps two channels in reserve. If the letter is detected, the song still runs.
You are misusing Chieh Hsüan when the list becomes a checklist. The failure mode is mechanical breadth: running all sixteen channels at once because the manual names sixteen, flooding an enemy with rumor and song and forged letter and bribe until the very volume of clumsy approaches tells him he is under coordinated attack. The list is a menu, not an order to be filled.
A second failure is loving the inanimate agency because it feels clean. The forged letter has no nerve to fail — so the careless officer treats it as free. He forgets that an unconvincing forgery is worse than no forgery, because it teaches the enemy that his correspondence is being targeted and hardens the very channel you wanted open.
The deepest failure is the one Chieh Hsüan ends on. You can have all sixteen agencies memorized and still fail at the only thing that matters: employing the men who work them. The diagnostic sign is an officer who can recite the channels fluently and cannot tell you which of his own people he would actually trust to carry one. Fluency in the menu masquerading as competence at the table.
The sharpest tension is internal and Chieh Hsüan states it himself: he builds sixteen categories and then concedes they will not stay sorted, deferring to Sun-tzu's five.13 Is the sixteen-fold scheme a genuine analytic advance or an elaborate gloss that collapses back into the classic on inspection? Sawyer leans toward the latter — the categories "may easily be subsumed under Sun-tzu's five rubrics," most being local guides plus an explicit listing of lures.14 But subsumption is not refutation. The list's value may lie precisely in making the lures explicit, naming the carriers Sun-tzu left implicit inside his five clean types.
A second tension is dating and continuity. This is a late Ming thinker reading and extending a corpus whose core is two millennia older. Sawyer's larger project repeatedly asserts continuity of military thought across that span, and his cross-dynasty claims need independent verification — a late synthesizer can read coherence back into a tradition that was messier in its own time. Whether Chieh Hsüan's typology reflects long-running practice or his own theorizing about an inherited vocabulary is not settled by the text alone.
Sawyer the translator and Chieh Hsüan the source pull gently against each other here. Chieh Hsüan presents the sixteen agencies as a real expansion of the concept of chien; Sawyer presents them as a finally reducible elaboration. The translator's footnote energy goes into the awkwardness of rendering "agency through" — and that awkwardness is itself the convergence, because it is the linguistic fingerprint of a genuinely new idea straining against an old word. If the concept were merely Sun-tzu re-listed, it would translate cleanly. The fact that it does not, that Sawyer has to reach for "agency through," is quiet evidence that Chieh Hsüan caught something the five-type scheme had left unnamed: the carrier, the medium, the through.
Handshake one — Sun-tzu's Five Spies. This is the parent the page argues with. Sun-tzu sorts espionage by the kind of person employed — local, internal, double, expendable, living. Chieh Hsüan sorts it by the channel the leverage rides. Lay the two schemes side by side and you get an insight neither produces alone: Sun-tzu's "types of spy" and Chieh Hsüan's "agencies through" are not rival taxonomies but two axes of the same matrix — one axis is the human vehicle, the other is the medium and the lure. Chieh Hsüan even tries to cross-tabulate them (living-or-dead × benevolence-or-fear) and finds the matrix too complex to harmonize.15 The handshake's payoff: covert action has always been at least two-dimensional, and the classic five-type list quietly compressed the second dimension into the first. The expendable agent and the forged letter occupy the same cell on the channel axis (both deliver false information into the enemy's decision loop) while sitting in different cells on the human axis (one is a person, one is paper). That is why they keep getting confused — they are functionally adjacent.
Handshake two — Systematic Covert Programs. A covert program, as the vault treats it, is a standing institutional apparatus rather than a single operation. Chieh Hsüan's closing line is the program's founding problem stated in one sentence: "employing agents is more difficult than employing the army."16 An army's difficulty scales with logistics and discipline — things you can systematize. Agents' difficulty scales with judgment about individual men, which resists systematization. The insight the two pages generate together: a covert program is an attempt to industrialize precisely the activity Chieh Hsüan says is least industrializable, and the tension between the sixteen reproducible channels and the one non-reproducible skill (employing men) is the permanent internal stress of any intelligence bureaucracy. The channels can be written into a manual; the judgment cannot.
Handshake three — Women as Intimate Agents (sibling). Chieh Hsüan lists "agency through women" as one channel among sixteen — "bribes that penetrate the women's quarters."17 His framing is revealingly thin: woman-as-conduit, the household as a wall with a gap in it. The intimate-agents page restores what the list flattens — that this channel runs on a person with her own position, leverage, and risk, not merely on a breach in a perimeter. The handshake produces a critique the list cannot produce of itself: Chieh Hsüan's typology, by reducing a human agent to "agency through women," demonstrates the very compression the intimate-agents tradition exists to resist. Naming a channel can erase the person who is the channel — and the manual's own breadth makes that erasure systematic.
Sharpest implication. The deepest covert thinking does not ask "what kind of spy do I need?" It asks "which channel into this enemy is open, and who can I trust to carry it?" Chieh Hsüan's real discovery is that the human spy was always just one carrier among many, and that the scarce resource was never the carrier but the judgment selecting it.
Generative questions.