The Geometry of War
How a narrow strait moves the global economy, and why the architecture of interconnection matters more than the size of any single shock.
For decades the Strait of Hormuz has been one of the most quietly consequential places on earth. Roughly a fifth of global oil consumption passes through the 21-mile waterway separating Iran from the Arabian Peninsula, connecting the energy exporters of the Gulf to markets across Asia and Europe. When tensions rise there, oil traders reprice futures contracts within hours, marine insurers revise war-risk premiums before the week is out, and governments convene quiet reviews of strategic petroleum reserves they hope never to use.
The instinctive reading of this is a supply chain story: confrontation threatens output, output falls short of demand, prices rise. But this framing mistakes a network for a pipeline. The global economy does not relay shocks from one point to the next in tidy succession. It behaves more like a struck tuning fork, transmitting vibration simultaneously, in every direction, through every surface in contact with it. Causality here is not sequential. It is structural.
Grasping that distinction requires a different kind of map, one drawn not of geography but of topology.
Chokepoints and Centrality
Every day, roughly 20 million barrels of oil move through the Strait of Hormuz: crude from Saudi Arabia, Kuwait, Iraq and the UAE, alongside large volumes of liquefied natural gas bound primarily for Asian economies. No overland pipeline exists capable of absorbing even a fraction of that volume at comparable cost or speed. The strait is not the most convenient route. For the energy systems of three continents, it is the only route that works at scale.
The Strait of Hormuz belongs to a small and structurally distinctive class of locations: the Strait of Malacca between the Indian and Pacific Oceans; the Suez Canal linking Europe and Asia; the Bab el-Mandeb at the mouth of the Red Sea; the Panama Canal between the Atlantic and the Pacific. These are not simply busy waterways. In the language of network theory, they are nodes of high betweenness centrality, points through which an exceptional proportion of the network’s shortest paths must pass.
The formal grounding comes from graph theory. Represent the global economy as a network G = (V, E), where V denotes the nodes (countries, ports, refineries, financial institutions) and E denotes the connections between them (shipping lanes, trade agreements, capital flows, policy transmission channels). Betweenness centrality then offers a precise measure of a node’s structural weight: not how large it is, but how many of the network’s pathways run through it, and how many would be severed or degraded if it were removed.
Importance in a network is a property of position, not of scale. That distinction determines where systemic risk concentrates.
The practical implication is that disruptions in geographically minor locations can carry economically outsized consequences. Size and centrality are correlated, but they are not the same thing. A node can be small and irreplaceable, just as a node can be large and peripheral. The chokepoints of global trade fall squarely into the first category.
The Price of a Probability
One of the less obvious features of networked financial systems is that markets do not wait for physical disruptions to materialise before adjusting. They price the probability of disruption, and they do so the moment that probability shifts rather than when the event itself occurs.
Consider the global economic state at any given moment as a vector of coupled variables: oil prices, shipping capacity, insurance costs, inflation expectations, interest-rate forecasts, exchange rates. These are not independent gauges. They are entangled, each shaped by the others through feedback mechanisms that operate across asset classes and jurisdictions at once. When geopolitical signals shift near the Strait of Hormuz, marine underwriters revise war-risk premiums on the basis of intelligence assessments, not confirmed losses. Shipping firms reroute or idle vessels on the basis of projected costs, not realised ones. Oil futures markets reprice on probability-weighted supply scenarios, not current inventory drawdowns.
The aggregate effect is that risk is discovered and redistributed across the system at the speed of information, which in practice means almost instantly. A political development in the Gulf arrives in the bond markets of Frankfurt and Tokyo not because any single institution has traced the full transmission chain, but because every participant is updating expectations simultaneously, and those expectations share a common underlying structure.
This is what separates a networked economy from a sequential one. In a sequential system, cause precedes effect at each step, and a disruption travels only as fast as the underlying physical process. In a networked one, the adjustment is concurrent: it is triggered not by a shortage that has arrived but by the credible prospect of one.
The Architecture of Contagion
Once a disturbance enters a networked system, its path is governed not by its initial magnitude but by the topology it encounters. Think of it as a propagation problem. The stress level at any node in the next period is a weighted function of the stress levels at its connected neighbours in the current period, plus any new shocks entering from outside. The matrix encoding those connections, specifying who is exposed to whom and to what degree, determines whether perturbations decay, persist or amplify over time.
Get Gayathri Shankar’s stories in your inbox
Join Medium for free to get updates from this writer.
In a sparsely connected network, local shocks tend to dissipate as they spread. In a densely connected one, particularly where a small number of highly central nodes intermediate a large share of total flows, stress circulates rather than disperses. The eigenspectrum of the connection matrix governs this: when the dominant eigenvalue approaches unity, small disturbances no longer dampen. They persist, and in the presence of fresh shocks, accumulate. This is the mathematical signature of systemic fragility, and it is a property of the network architecture, not of any individual shock.
Applied to the Strait of Hormuz, the framework produces a non-obvious result. A disruption there does not produce a chain of sequential effects in which higher oil prices beget higher shipping costs, which in turn revise inflation expectations, which then alter central bank guidance. All of these adjustments occur concurrently, through subsystems that share a common exposure to the same underlying signal. What appears to be a downstream cascade is better understood as a simultaneous system-wide recalibration. The European bond markets are not responding to higher oil prices. They are responding to the same geopolitical signal that oil markets are responding to, processed through a shared expectation structure that connects them both.
Systemic Risk as a Geometric Property
Conventional economic analysis tends to treat energy markets, shipping logistics, financial markets and monetary policy as distinct domains, each with its own equilibrium logic, modelled in isolation and occasionally linked by transmission mechanisms grafted on after the fact. The framework of complex systems offers a direct challenge to this habit: the variables that govern systemic fragility frequently reside not within any single domain but in the couplings between them, in the correlations, shared exposures and feedback loops invisible to any model that does not represent the full network.
The physicist Philip Anderson argued in 1972 that “more is different”: that as systems grow in scale and density of connection, new causal structures emerge at higher levels of organisation that cannot be derived by inspecting components in isolation. Herbert Simon extended this by describing complex systems as hierarchical assemblies of interacting subsystems, each generating patterns meaningful at its own scale, while the inter-level dynamics produce behaviours that no individual level of analysis can fully see. Emergence, philosophers call it. For risk analysts, it is a methodological warning.
The practical implication is captured in a deceptively compressed formalism. Systemic stress S can be written as a function of two variables: p, the probability of disruption, and c, the structural dependency of the network on the affected node, a quantity that tracks closely with betweenness centrality.
The relationship is multiplicative. Risk does not accumulate linearly as either variable rises independently. It accelerates when both rise in concert. A corollary follows: a moderate disruption probability at a node of high centrality can generate greater systemic stress than a severe disruption at a peripheral one. The Strait of Hormuz requires no imminent threat to command disproportionate attention from risk analysis. Its structural position ensures that even a marginal shift in disruption probability propagates across a wide range of asset prices and policy expectations, before a single barrel of oil has failed to arrive anywhere.
The Lesson of 2008
The 2008 financial crisis offers the most legible recent illustration of what happens when this logic is not accounted for. Early mortgage defaults in the United States were read, through conventional credit analysis, as a contained sectoral problem. The aggregate losses were real but not, in isolation, sufficient to explain the depth of the recession that followed.
The conventional analysis had missed the architecture. Mortgage risk had been dispersed throughout the global financial network through securitisation, collateralised debt obligations and derivative contracts, lodging itself in the balance sheets of institutions with no direct exposure to American housing markets. When stress entered the network at that node, it propagated through funding markets and counterparty exposures with a speed and reach that no model calibrated on sector-specific credit risk could have anticipated. The crisis was not a function of the shock’s initial size. It was a function of the network through which that shock moved, and of the prior failure of most risk frameworks to represent that network at all.
The connection to Hormuz is more than analogical. In both cases the analytically relevant question is not how severe the shock is, but how central the node. And in both cases, the answer to that second question was either unknown or underweighted before the event.
The Price of Efficiency
The deeper issue sits upstream of any particular geopolitical moment. Over recent decades, the global economy has grown substantially more interconnected: supply chains stretched across continents, energy flows concentrated through a small number of maritime passages, financial institutions coupled through cross-border funding arrangements and derivatives books spanning multiple jurisdictions. These linkages produce genuine value. Lower costs, faster capital allocation, more efficient allocation of productive capacity across the global economy. They also narrow the system’s tolerance for stress.
Efficiency and resilience pull against each other at a structural level. A system optimised for normal conditions progressively eliminates the buffers, inventory, spare capacity, redundant routes, that absorb disruption when conditions deteriorate. The tighter the coupling, the faster stress propagates, and the shorter the window in which any participant can respond before the disturbance reaches them. This is not a design flaw. It is what optimisation looks like from the outside when normal conditions no longer hold.
The relevant question for policymakers is therefore not whether to accept this trade-off but whether it was accepted knowingly. In most cases, it was not. The concentration of global energy flows through a handful of maritime chokepoints was not a deliberate architectural choice. It was the sediment of millions of individual cost-minimisation decisions, none of which had systemic resilience anywhere in its objective function.
A System Designed by No One
The global economy is an emergent structure. No single actor specified its topology. No planner decided that a fifth of global oil consumption should pass through a 21-mile channel, or that mortgage risk should be embedded in instruments held across six continents. These configurations arose from the accumulation of individually rational, decentralised choices over decades, each defensible in isolation, collectively producing a system whose fragility properties no single actor was ever positioned to assess in full.
For the most part this arrangement works, and works well. Energy crosses oceans, capital traverses borders, goods move through intricate supply chains with a reliability that would have been difficult to imagine a century ago. The same connections that generate vulnerability also generate the efficiency that makes the system worth sustaining.
But the true character of complex systems only becomes visible under stress. What stress reveals, with some consistency, is that fragility is not randomly distributed across a network. It pools at the nodes of highest centrality: the points that sit at the intersection of many pathways, that cannot easily be circumvented, that the rest of the system has come to depend upon without any formal decision ever having been made to do so.
The Strait of Hormuz is one such node. The American mortgage market in 2006 was another. The next will almost certainly share the same structural signature. The open question is not whether it will emerge at a point of high centrality. It almost always does. The question is whether the network will have been mapped carefully enough, in advance, for anyone to see it clearly.
