The king cannot be everywhere. His writ runs through distant provinces he will never see, governed by local rulers he must trust and cannot check. So the Han-shih Wai-chuan proposes a fix: station men in the four quarters whose job is simply to look. The text calls them Mu — "shepherd." Two shepherds for each direction, "deputed to secretly observe distant regions and oversee their masses."1 Not to govern. To watch the governors. If people in a far province go hungry and cold without clothing and food, if lawsuits are decided unfairly, if worthy men are passed over for office, the shepherd reports it straight back to the Son of Heaven.2 The word for the office is pastoral and gentle — shepherd, herder, keeper of the flock. What it actually describes is an early internal surveillance officer, the king's eye pushed out to the edges of the realm.
The Mu, in the one source that develops it, is a centrally appointed observer posted to peripheral and tribute regions and charged with intelligence gathering. The framing is benevolent. The shepherds exist, the Han-shih Wai-chuan says, "to open the Son of Heaven's eyes to the four quarters and allow his perception to penetrate them."3 When a local ruler comes to court, the king — already briefed by his shepherds — can ask, with apparent sorrow, how it can be that his subjects go hungry, or that lawsuits go undecided, or that worthy men go unappointed. The ruler, stunned that the distant king somehow knows, goes home and consults his ministers. The people, hearing of it, marvel: "Although we live in a rustic corner, he sees us as if we were nearby. Although we live in darkness, he sees us clearly. How can he be deceived?"4 The report is the smaller half of the shepherd's value. The larger half is the demonstration that the center sees everything — surveillance as a deterrent worn openly, even as the watching itself stays secret.
The trouble is corroboration. This rich picture comes from essentially one place. When Sawyer turns to the classic works of the Spring and Autumn period — the Ch'un Ch'iu and Tso Chuan — he finds only a single instance asserting that Mu should be established for China's anciently delineated nine regions, and even that one essentially echoes a line in the "Canons of Yao" in the Shang Shu.5 In every other case, the word just means what it plainly says: shepherds, herders, fodder overseers in the quartermaster corps, cattle managers.6 The grand intelligence office is hanging on very thin textual thread.
Even the lengthy Chou Li, which touches the office about a dozen times, gives Sawyer only three passages that hint at a centrally deputed observer's role — and none that says the work had a secret or clandestine character.7 In the first, the shepherd is one of nine factors by which the king gains and controls the people: he "gains people with land," set beside a teacher who "gains people through his worthiness" and a Confucian who "gains people through the Tao."8 In the second, a series of nine offices meant to rectify the states includes establishing Mu, whom the commentators read as "regional governors" — though Sawyer notes this is not necessarily right.9 The third passage hints they exercised some indefinite supervisory function over the states, especially when the king was about to make his tour of inspection, implying they fed him the information his administrative judgments would need.10
That is the whole case. Sawyer's own verdict is appropriately cautious: the material is "so sparse, the references so minimal," that no substantive conclusion is justified beyond recognizing that something like the later censors or independent inspectors — the kind eventually turned against the state's own administrators, becoming an internal affair — was conceptually present in the Spring and Autumn era.11 The Mu is less a documented institution than a documented idea of one.
The Mu is the inward-facing twin of the outward-facing intelligence office. Where the Hsing-jen collected on foreign states, the Mu collected on the king's own provinces — the same competence, aimed at home. That pairing is the office's main contribution: it shows early Chinese statecraft conceiving intelligence in two directions at once, against rivals and against subordinates, long before either had a settled bureaucracy. It hands Field Observation Methods a domestic theater — the shepherd reads a province's health the way an envoy reads a rival's, by hunger, justice, and the appointment of worthies rather than by armies and granaries. And it plants the conceptual seed of the censorate, the institution that would later turn the watching inward as permanent machinery.
Sit with the scene the Han-shih Wai-chuan stages, because the whole theory of the office is inside it. A local ruler arrives to pay homage. He has, let us say, neglected a famine in a far district and let some lawsuits rot. He expects ceremony. Instead, after the salutes, the Son of Heaven addresses him with feigned distress: "Alas, can it be that the instructions of our administration have not reached your domain? How can there be hungry and cold people who lack clothes and food, or criminal cases and lawsuits that have not been decided equitably, or worthies that remain unappointed?"12
The ruler has named nothing. The king has named everything — the three exact failures the shepherds were told to watch for. The power is in the asymmetry. The local ruler cannot tell how the king knows, only that he does, and so he must assume the king knows all of it. He retires and "certainly consults with his ministers and high officials about this incident."13 The fear does the governing. And when word filters down to the common people of that distant corner, they reach the conclusion the office was built to produce: he sees us in our darkness, he cannot be deceived.14 One briefing, deployed as a single startling question, converts a secret report into a realm-wide belief that the center is omniscient. That is the design — the report is private, but its effect is meant to be public and self-reinforcing.
You are a Mu posted to a far quarter. You do not sit in the governor's hall; you move through the districts under it. You walk a market and watch whether the poor have winter clothes. You linger near a magistrate's gate and listen for the litigant who lost a case he should have won. You note which capable man has been left without a post while a connected fool holds one. You carry nothing that announces what you are — the office has no secret apparatus, only a man with a route and a memory. When you have enough, you do not confront the governor. You send it home, past him, to the Son of Heaven. Months later, at court, you may hear that the king asked your governor a question that landed like a thunderclap. You said nothing. You were never there. You are already walking the next district.
The Mu fails first by dissolving back into its own name. The clearest warning sign in the record is the word collapsing into its literal sense — the moment "shepherd" stops meaning the king's secret observer and starts meaning, again, the man who keeps the cattle. Sawyer found exactly this: outside one or two passages, every use of Mu is just a herder or a fodder overseer.15 An intelligence office that cannot keep its own title distinct from the day job has not yet truly separated from the flock.
The second failure is the one Sawyer's caution points at directly: the office that exists in prescription but not in practice. The Han-shih Wai-chuan describes a beautiful machine; the contemporary chronicles barely confirm it ran. When the only vivid account of an institution is the one telling you how splendid it would be, the diagnostic sign is precisely that splendor — the gap between the idealized passage and the silent historical record.
The tension is unresolved and should stay that way. One source — the Han-shih Wai-chuan — gives a fully realized clandestine internal-surveillance office. The Chou Li gives only faint, non-clandestine hints. The Ch'un Ch'iu and Tso Chuan give almost nothing but cattle. Sawyer does not paper over this; he names the sparseness and refuses to conclude more than that the concept of an inward-turned inspector was present.16 The honest reading is that the Mu sits on the line between an institution and an aspiration, and we cannot say which side it falls on.
The open question that follows: did the Spring and Autumn states ever actually run such men, or did later writers retroject the censorate they knew onto a vaguer ancient term to give it a noble pedigree? The single corroborating line's echo of the "Canons of Yao" cuts toward retrojection — an old phrase reused to dignify a later idea.17 The Mu may be one of the earliest cases in the vault of an institution that exists more firmly in the texts describing it than in the world the texts describe.
Sawyer is at his most disciplined here, and the page should honor that. Where his broader project sometimes leans hard on cross-dynasty continuity — threading one office from Chou through Han and onward — with the Mu he applies the brake himself, declaring the evidence too thin for substantive conclusions. The convergence he allows is minimal and careful: that the concept of the later censor was present, not that the office existed. The reader's job is to keep his caution intact and resist the gravitational pull toward a cleaner story. The one place his continuity instinct does surface — linking the Mu to "the censors or independent inspectors of latter times" — is offered as conceptual lineage, not institutional proof, and that distinction is exactly the one to preserve.
To The Semblances Problem. The Mu turns the semblances problem inside out and aims it at the king's own officials. A governor presents a smooth surface to the throne — tribute paid, homage offered, all well in the provinces. The shepherd's whole function is to test that surface against the depth: are the people actually fed, are lawsuits actually fair, are the worthy actually employed? The insight neither domain reaches alone: the semblances problem is not only a hazard between states but within a single hierarchy, where every layer of subordinates has an incentive to show the layer above a flattering false face. The Mu is an institutional answer to the principal–agent version of the semblances problem — the center building a parallel channel of sight precisely because it knows its own agents will manage appearances. The shepherd exists because power, looking down its own chain, sees mostly the mask it is shown.
To The Hsing-jen. Reading the two offices together produces a claim neither makes alone: early Chinese statecraft split intelligence by target direction before it split it by method. The Hsing-jen watched outward, riding into rival courts; the Mu watched inward, walking the king's own districts. Same skill — reading a polity's health from observable particulars — pointed at two different threats. The tension that emerges is structural and durable: an outward intelligence service strengthens the state against enemies, but an inward one strengthens the center against its own provinces, and the second is far more politically dangerous to build, because the people you are watching are the people you depend on to govern. That may be exactly why the Mu stayed thin in the record while the Hsing-jen flourished — turning the eye outward is easy to justify; turning it on your own administrators is the move that later dynasties would institutionalize only with great friction, as the censorate.
Sharpest implication: The internal-surveillance state was imagined in China before it was built. The Mu is a fully theorized inward-watching office sitting on almost no evidence of actual operation — meaning the idea of watching one's own administrators predated the institution capable of doing it, and waited centuries for the censorate to catch up.
Generative questions: