Somewhere in the second year of a new posting, the foreign diplomat learns something that no briefing document prepared her for: the ministry she needs is physically on the wrong side of the city. Not wrong in any dramatic sense. Just inconvenient enough that her counterpart from the embassy two streets away from that ministry has lunch there twice a week, knows the deputy-director's assistant by name, and gets his calls returned the same day.
That gap, measured in taxi minutes, turns out to matter enormously.
The city plan that was never neutral
Every capital city is, among other things, a spatial argument about where power should sit and who should have easy access to it. The argument was usually made by a planner, a monarch, or a military engineer with specific intentions: project authority, separate the governed from the governors, or simply house everyone the state needed within walking distance of everyone else the state needed. The choice, wherever it was made, was never innocent.
Washington D.C. is the clearest example in the Western tradition. Pierre Charles L'Enfant's 1791 plan placed Congress and the White House at opposite ends of Pennsylvania Avenue, close enough to allow daily human contact but far enough to encode the separation of legislative and executive power in cobblestones. The departments that grew up between and around those anchors sorted themselves by proximity to one or the other, and that sorting persisted for generations. Treasury sits immediately east of the White House not because of any administrative logic worked out on paper, but because Alexander Hamilton insisted on it. The physical fact of that adjacency meant Treasury officials could walk to presidential meetings without scheduling them, and informal influence followed the footpath.
Brasília makes the counter-argument just as forcefully. Designed in the late 1950s by Lúcio Costa and Oscar Niemeyer on a blank plateau, the plan separated the three branches of government into distinct sculptural objects surrounded by vast open space. Beautiful, certainly. But the distances between ministries, between the legislature and the executive, between the official quarter and the residential areas where civil servants actually lived, were substantial enough that spontaneous contact became rare. Meetings required cars, which required scheduling, which converted every interaction into a formal appointment. Brasília's planners intended to build a rational, modern state. What they also built, inadvertently, was a bureaucracy that runs heavily on telephone calls and internal memoranda because the walk to a colleague's office is, in practice, a thirty-minute drive.
How informal networks fill the gaps
The formal hierarchy of any government is the org chart. The real hierarchy is the lunch route.
Consider two mid-level civil servants, call them Marta and Daniel, who work in adjacent but technically separate agencies. If their offices share a building, or even a street, they will bump into each other. They will talk about a problem before it becomes a formal interdepartmental dispute, agree on a number in a corridor before it ever enters a spreadsheet. That kind of ambient coordination is extraordinarily efficient, and it is entirely invisible to any official record. It is also, for that reason, chronically undervalued by reformers who read org charts and wonder why government moves slowly.
Now move their offices four kilometres apart. Every interaction requires intent. You don't bump into someone four kilometres away. The relationship between their agencies becomes mediated by email threads and formal liaison officers, channels that are slower, more defensive, and more likely to produce the turf protection that outsiders associate with bureaucratic dysfunction. The dysfunction isn't personal. It's architectural.
Paris demonstrates this in its most evolved form. The French state's grands corps, the elite administrative bodies that cross-cut formal ministerial lines, have historically reproduced themselves through the grandes écoles and through the physical concentration of power in a small patch of the 7th arrondissement. The Assemblée Nationale, the Quai d'Orsay, the Matignon, the Élysée: not identical institutions, but walkable from one another. The people who staff them at senior levels are often classmates, and the physical proximity reinforces a social network that was already dense, the two things winding around each other like strands of the same rope. An outsider appointed to a ministry but without those prior social ties finds the informal system almost impenetrable, not because anyone blocks entry, but because the invitations to lunch, the after-work aperitifs, the chance encounters at the nearby café, all flow along pre-existing lines.
The crust that builds up inside
There is a subtler effect that operates over decades, and it deserves more attention than it gets.
When a capital's informal networks calcify around a particular geography, they also calcify around the people who live and work within that geography. Reform becomes harder not because the org chart resists it but because the informal system that makes the org chart function is spatially embedded. Moving a ministry to a new building, or relocating a department to a cheaper district, disrupts those ambient connections. The civil servants who relied on them spend a year or two rebuilding, during which institutional memory fragments and productivity quietly drops.
The British government learned this when it began dispersing civil servants out of central London to regional offices. The policy was sound on paper: cheaper rents, less pressure on the capital's infrastructure. What it underestimated was the degree to which Whitehall's informal effectiveness depended on the density of the village. Departments that moved to Bristol or Sheffield found themselves in better buildings with lower costs and, for a period, measurably worse access to the informal channels through which decisions are actually lubricated. The lesson was filed, absorbed imperfectly, and then repeated.
The mechanism isn't conspiracy. It's friction, or rather the absence of it, and the absence is something most governments never think to count until it's gone.
What new capitals reveal
The most instructive cases are the purpose-built capitals that were supposed to solve the problem of an overcrowded, historically layered city. Canberra, Naypyidaw, Astana, Abuja: each was planned partly to escape the informal networks of the old capital, to build a rational state on neutral ground. Ask yourself whether any of them managed it.
The results have been consistent enough to suggest a pattern. The new capital does break up old networks. It also, within a generation, generates new ones, and those new ones cluster around the same spatial logic: whoever is closest to the executive residence, whoever shares a residential suburb with the senior bureaucrats, whoever's office sits on the route between the legislature and the ministry of finance, accumulates informal influence. The geography changes. The mechanism doesn't.
Naypyidaw is the extreme case, a city so vast and so car-dependent that informal contact between officials from different ministries is genuinely unusual. Observers of Myanmar's government noted, before the political convulsions that overtook it, that the isolation of the capital produced a bureaucracy that was formally organised but informally fragmented, one in which the absence of ambient coordination made it hard for anyone outside the innermost circle to understand what the state was actually deciding. The capital had been planned to project strength. It ended up projecting silence.
That is the sharpest version of the lesson, and it is worth sitting with. Physical layout doesn't just shape who knows whom. It shapes what the state can know about itself. A government whose ministries cannot easily talk to each other informally will eventually discover that the formal channels, however well designed, carry too little of what needs to be communicated. The city plan isn't background scenery. It is, in a very literal sense, the operating system on which the state runs, written by people with specific assumptions, most of which they never thought to explain, and revised by almost nobody since.