A new political appointee arrives at his agency on Monday morning. He has the formal title. He has the formal authority. He has a list of changes he was hired to make. He moves fast — calls meetings, demands action, pushes through directives. The senior career staff smile, take notes, and go back to their offices.
Nothing happens.
He pushes harder. He demands explanations. The explanations come back wrapped in regulations, security factors, jurisdictional disputes, prior commitments, and references to procedures the appointee has never heard of. We've tried that before. It's a coordination issue. We need legal review first. The administrative molasses gradually deepens around him. Eighteen months in, exhausted and humiliated, he resigns. The senior career staff applaud his farewell speech at the going-away party.1
Siu's commentary on this scene is one sentence and worth pinning. That, of course, pleases the bureaucrats immensely, as they sit applauding his farewell speech at the going-away party. This is always the painless way of getting rid of an alien invader. The political appointee thought he was running the agency. The agency was running him. The Subterranean stance — operating from beneath the formal level of authority through control of process and momentum — won the engagement before the appointee understood the engagement had begun.
Siu treats the Subterranean as one of the seven strategic stances and gives it the most operationally detailed treatment of any stance in the book. Two classes — illegal Subterranean (Mafia and equivalents) and legal Subterranean (the senior career civil servant and the senior career military officer). The illegal version is more dangerous in obvious ways. The legal version is more potent. Siu's words: Much more potent among the subterranean congregations of power, although more amorphous, is the legally constituted bureaucrats of the modern governments.2
The political appointee comes and goes on a four-year clock. The senior career bureaucrat remains. The asymmetry is structural. Operating in the wings and backstage of the political drama and generally hidden from public view, the senior career civil servant and military officers exercise considerable power on their own with far greater security, comfort, and flexibility than the prominent political leaders.3
What the bureaucrat has that the appointee does not is accumulated information and historical momentum. The information is in the files. The momentum is in the institutional patterns of decision-making that have been refined across decades. They have the information in their files and the knowledge in their persons, which constitute an essential stuff of power. It is this historical momentum that is the source of the irresistible power in senior civil servants and ranking military officers.4 The appointee can override individual decisions; the appointee cannot override the cumulative weight of how decisions actually get made in the building. The institutional water flows downhill in patterns that pre-existed his arrival and will outlast his departure.
Siu's signature term for the Subterranean's primary instrument is servo-bureaucratic viscosity. The bureaucrat titrates just the right amount and kind of friction to produce the organizational sluggishness required to defend his domain. The catalog deserves to be read as one sentence because that is how it operates — a dense list of overlapping levers any one of which can stop forward motion:
Laws, executive orders, regulations, security factors, red herrings, pet peeves, unofficial commitments, what-happened-to-whom-whens, we've-tried-it-befores, conflicts of interests, jurisdictional disputes, coordinations.5
Each item in the list is a separate kind of friction. Laws and executive orders are formal. Regulations are semi-formal. Security factors and jurisdictional disputes are procedural. Red herrings and pet peeves are personal. Unofficial commitments are relational. What-happened-to-whom-whens and we've-tried-it-befores are historical. Conflicts of interests and coordinations are network-based. The bureaucrat can deploy any combination of these against any directive he disagrees with. Each one alone is defeatable; deployed in combination they are not.
The inertia seems to increase geometrically as the proposal encroaches closer upon what has been traditionally delimited as the bureaucrat's sphere of authority and power. The closer the appointee gets to the bureaucrat's actual interests, the more friction the bureaucrat can summon, and the friction compounds. Push too hard, and you find yourself wading in the gradually deepening and thickening administrative molasses. The exhaustion is the win condition for the Subterranean.
Siu makes a quiet observation that complicates the picture. Most bureaucrats are not actively resisting. Most bureaucrats are sincere, talented, and loyal to the political appointees and do not resist changes in policies. More frequently than not, the impasse is primarily engendered by the overconspicuous display of power on the part of the political tyros in power.6
The mechanism flips into resistance only when the appointee misreads the relationship. Read it correctly and the same bureaucrats who could grind any directive into administrative molasses will move it through the system at high speed. If the program has been developed in collaboration with the senior career civil servants and flag-ranked officers and the general conclusions and objectives meet with their concurrence, all obstacles are waved aside like magic, as it were, and the political boss is impressed with the speed of execution.7
The political appointee who is merging and the bureaucracy is winning. The political appointee who is fighting them is the one who is losing. Most appointees do not understand this and try to assert their formal authority directly, which triggers the viscosity catalog and produces their own going-away party.
Anthony Howard's review of the Crossman diaries (1976) compresses the actual deal. "It suits the vanity of politicians to pretend that they are all-powerful; it equally suits the convenience of civil servants to maintain that they have nothing more than an occasional influence on the margin. But the truth is rather different: civil servants willingly yield to their political 'masters' all the rewards in terms of fame, glamour and publicity — in return for which they expect, and are conceded, a continuing power, regardless of the political color of the Government, over the decisions that are actually taken."8
The bargain is publicity-for-substance. The minister gets the press conferences. The civil servant gets the actual decisions. Both sides are content because both are getting what they actually want. The illusion that the political class is in charge is itself part of the bargain.
President Carter, fifteen months into his term, broke the convention and named the problem out loud. "Before I became President I realized and was warned that dealing with the Federal bureaucracy would be one of the worst problems I would have to face. It's been even worse than I had anticipated."9 The press treated the comment as routine frustration. Siu treats it as the political class's recognition that the bargain is real and asymmetric. The bureaucrats had no incentive to confirm the recognition. They went on running the country.
Siu's closing line on Op#27 is the cleanest summary of where formal power actually sits in any large administrative state: Bureaucrats need have no fear of democracy.10
Siu treats the Mafia and equivalents briefly because the legal version is more important. The Mafia operated in mid-twentieth-century America by infiltration beneath the law — assessments on small businesses, control of gambling, distribution networks, juke box selections. By 1945 its hold in Sicily was reaching into every field of activity.11 The illegal Subterranean has lost ground in recent decades not because the structure became weaker but because the state itself began to operate gambling, lottery, and horse-betting on a legal basis with well-advertised and much higher grand prizes. The legal Subterranean (the bureaucracy) absorbed the illegal Subterranean's customers. The mechanism is the same; the legal version simply scaled.
The democracy problem. Siu's closing line is descriptively accurate and normatively unsettling. If formal democratic accountability is partly theatrical relative to the bureaucratic continuity that makes actual decisions, the legitimacy of the system depends on the theatrical cover staying intact. The cover is thinning under contemporary conditions (more transparency, more direct communication, less deference to expertise), and the operational consequences are visible — political appointees who refuse the bargain, bureaucracies that refuse the legitimacy of their nominal masters, and a political class less able to govern through the traditional mechanisms. Siu names the bargain. He does not address what happens when the bargain breaks down.
The good-bureaucrat / bad-bureaucrat asymmetry. Siu acknowledges that most bureaucrats are sincere and loyal. The viscosity catalog is operationally available to all of them and only deployed in resistance by some. The political appointee cannot reliably distinguish in advance which bureaucrats will deploy the catalog and which will not. The diagnostic is what the bureaucrat does after the appointee tries to push hard against the bureaucrat's interests, and by then the engagement is already lost.
The illegal-to-legal absorption. Siu's account treats the legal absorption of the illegal Subterranean (state-run gambling absorbing Mafia gambling) as a one-way trend. Modern conditions may be reversing it — ungovernable spaces (digital, transnational, encrypted) are creating new domains where the illegal Subterranean has structural advantages the legal version cannot match. The taxonomy may need updating for the post-2010 environment.
History — When the Bureaucratic Architecture Was Designed Deliberately: Meritocracy-Within-Subordination — The Loyalty-Capability Equation — Genghis Khan in 1206 faced the bureaucracy problem from the opposite direction. He had an empire to administer and no native bureaucratic class to administer it. He had to build the bureaucracy from scratch, recruiting capable administrators from conquered tribes and offering advancement on capability and loyalty. What he engineered, specifically, was the architecture that prevents the Subterranean stance from forming at the level of the bureaucracy itself. Advancement within subordination. You can advance — but only as far as Khan permits. The cap on the bureaucrat's potential power was structural, not personal.
The handshake reveals what Siu's analysis assumes away. Modern bureaucracies became Subterranean-capable because the political class did not engineer the structural constraints Khan engineered. The civil service was given continuity (correctly — institutional memory is a public good), accumulated information (correctly — administrative continuity is a public good), and procedural autonomy (correctly — the rule of law requires it). What was not engineered is the cap on how far the bureaucrat's power can rise relative to the political appointee's. Khan had a hard cap because the alternative was a bureaucratic faction that would eventually displace him. The modern administrative state has no equivalent cap, and the consequence is the legal Subterranean Siu describes — a structural feature of the political system that no individual political appointee can defeat. The two pages together name a finding that is unwelcome in democratic theory: the architecture that made modern bureaucracies effective is the same architecture that made them politically uncontrollable, and the political class accepted the bargain Khan refused because they did not understand the asymmetry it would produce. Khan's architecture would be illegal under modern democratic norms. The price of refusing it is the going-away party.
Psychology — The Inheritance Pattern: Generational Survival Pattern Cycling — Each generation's adaptive survival responses become the constraints the next generation inherits. What was an adaptation in the parent's environment becomes a worldview in the child's environment, and an invisible assumption in the grandchild's environment. Trauma becomes worldview becomes baseline. The mechanism operates at the family level and at the institutional level alike. The bureaucrat's historical momentum — Siu's term — is the institutional version of this pattern. The procedures that were adaptive when the agency was founded become assumptions in the second generation, become invisible default in the third, and become the we've-tried-it-befores and what-happened-to-whom-whens of the viscosity catalog.
The handshake produces a finding neither page states alone. The Subterranean's power is not, primarily, the bureaucrat's deliberate strategy. The Subterranean's power is the accumulation of generational adaptations embedded in the institutional procedure — adaptations that no individual bureaucrat designed and that no individual bureaucrat could explain in full. The bureaucrat is the carrier of the pattern, not its author. Reading the two together explains a phenomenon the political-science literature names without explaining: bureaucracies resist innovation in ways that are coordinated across the building without being coordinated by any specific person. The coordination happens through the inherited pattern, not through deliberate organization. The political appointee who tries to fight the bureaucracy is fighting fifty years of accumulated adaptations, most of which are operationally useful and some of which are not, and none of which the bureaucrats themselves can fully separate. The fight is therefore mostly unwinnable in the form the appointee wants to fight it. The merging strategy works because it allows the inherited pattern to absorb the new directive on its own terms, transforming the directive into a form the institutional procedure can execute without breaking. Working with the pattern is the only mode that produces forward motion at scale, and the appointee who does not understand this is fighting the entire history of the institution with eighteen months of formal authority.
1. Diagnose the building you are walking into. It is your first morning in a senior role at a large institution. Before scheduling the welcome meetings, walk the floor. Notice who has been there twenty years and who has been there two. Ask the senior career staff what their previous bosses tried that did not work. Listen carefully to what they say, more carefully to what they do not say. The viscosity catalog is already pre-loaded against changes the building has decided cannot be made. You need to know what is already loaded before you propose anything.
2. Identify your three most senior career staff and meet them individually. Each meeting is one hour. The questions are: what would you change in this agency if you could? what have previous appointees gotten wrong? what do you need from me that you have never gotten from a previous appointee? The answers are the map. The bureaucrats who answer thoughtfully are the ones whose cooperation will determine your tenure. The bureaucrats who answer evasively are the ones who will deploy the viscosity catalog the moment you push.
3. Pick two changes, not twenty. The political appointee with twenty initiatives gets the viscosity catalog deployed across all twenty and accomplishes none of them. The political appointee with two initiatives can build the political and procedural support for those two and accomplish them both. The discipline is selecting the two whose accomplishment justifies the term. Most appointees over-select and underdeliver.
4. Run the merging move on each change. Find the senior career bureaucrats whose interests align with each of the two changes. Develop the proposals jointly with them. Let them shape the procedural form so it fits the building's existing patterns. Their shaping is what prevents the catalog from being deployed against the proposal. The appointee whose proposal carries internal bureaucratic sponsorship is operating in a different mode from the appointee whose proposal arrives as an external directive.
5. Recognize when you are running the failure mode. The signs are specific. Everything is taking longer than it should. Every meeting produces new procedural requirements you did not anticipate. Senior career staff are unfailingly polite and consistently unhelpful. Junior staff agree with you in private and revise their position in meetings. When two or more of these are present, you are running the failure mode and the going-away party is in the planning stage. The corrective is to stop pushing on the current frame, reset, and engage the merging move on a smaller proposal that has internal sponsorship. Continuing to push will not produce different results.
6. Notice what kind of bureaucrat you are becoming. If you are a senior career officer reading this in your second decade at the agency, the question is whether you are running the merging mode (good bureaucrat) or the catalog mode (resisting bureaucrat). Both are operationally effective. Only one is institutionally healthy. The catalog mode preserves your local power at the cost of the agency's adaptability. Notice when you are running it and ask whether the directive you are resisting actually deserves the resistance.
Modern democratic government does not actually produce the policy outcomes its formal architecture implies it produces. The political class wins elections, occupies offices, makes speeches, and produces the appearance of governance. The bureaucratic class produces the actual policy outcomes through the accumulated procedural patterns that no individual bureaucrat designed. The bargain Siu names is real and operationally stable as long as both classes accept it. The current period is one in which both classes are increasingly refusing the bargain — political classes that demand actual control over policy outputs, bureaucratic classes that resist political directives they consider illegitimate. The result is a governance system whose nominal architecture and operational architecture are diverging visibly, with neither class able to operate the system as it was previously operated. Siu describes the equilibrium that held for most of the twentieth century. We are in the period in which the equilibrium is breaking down. The implication is not partisan. The implication is that the political-bureaucratic relationship is being renegotiated in real time across most administrative states, and the renegotiation is producing the same kind of dysfunction in democracies of all political colors. The Subterranean stance, as Siu described it, is the pre-renegotiation stance. What replaces it is being figured out now, and badly.
The Crossman bargain (publicity for politicians, substance for bureaucrats) was operationally stable for most of the twentieth century. Modern transparency norms have made the bargain less viable — the bureaucratic class is more visible than it used to be, and the political class less able to claim credit for outcomes it did not actually produce. What replaces the bargain, and what does the political-bureaucratic equilibrium look like once the publicity-substance trade has broken down?
Khan's architecture (advancement within subordination) prevented the Subterranean from forming at the level of the bureaucracy. Modern democratic norms preclude that architecture. Are there democratic-compatible structural caps that would produce equivalent constraints, or is the Subterranean an inevitable feature of any large administrative state with constitutional protections for civil servants?
The illegal Subterranean was largely absorbed by the legal Subterranean across the twentieth century (state lotteries, regulated gambling, etc.). The post-2010 information environment is creating new domains where the illegal Subterranean has structural advantages the legal version cannot match (encrypted networks, transnational distributed organizations, decentralized financial systems). Is the absorption pattern reversing, and what does a renewed illegal Subterranean operating across digital domains look like at scale?