Network Power vs Territorial Power

The Core Thesis

Network powers gain leverage by controlling flows: trade routes, finance, mobility, alliances, standards, logistics and access. Territorial-production powers gain leverage by controlling land, population, industry, resources, proximity, coercive mass, and internal lines, meaning the ability to move forces inside their own territory faster than an enemy can move around it.

Historically the first kind looked like sea power, because the oceans were the long-distance network. The same logic now runs through payment rails, the systems that move money between banks, and through sanctions law, insurance, semiconductors, the computing power behind AI, software, technical standards, energy routes and data.

The political scientists Henry Farrell and Abraham Newman gave the modern form a name in 2019: weaponized interdependence, the use of a central position in a global network to watch, or to choke, everyone who depends on it. This framework treats that as one half of a pair.

The working frame is:

Network optionality versus territorial-production mass.

Shorter: flow systems versus mass systems.

Sea power was the original operating system. Later technology added modules.

Power Is Usable Access

A state may possess oil, grain, factories, ships, ports, people, money, minerals, territory or data.

Possession is not yet power.

Power depends on whether those things can be moved, converted, defended, financed, supplied, coordinated, repaired, trusted, replaced, scaled up and politically sustained.

Oil in the ground is not military fuel. Food in Ukraine is not bread in Berlin. A chip factory is not AI capability unless the design tools, the lithography machines, the packaging plants, the electricity, the software, the engineers, the financing and the shipping all connect. A fleet is not control of the sea unless it has bases, repair yards, fuel, munitions, surveillance, communications, political permission to use foreign ports, and crews willing to sail under threat.

Power is a property of the whole system. Inventory is only an input.

A territorial-production power often controls things. A network power often controls the movement and conversion of things. That distinction is the heart of the framework.

The Two Axes

The framework does not sort states into pure network powers and pure territorial powers. That is the first trap. Most serious powers are mixed, and some change over time. So it uses two axes and gives every state a position on each.

Network Optionality is the ability to keep critical flows moving, substituted, financed, defended, accepted as legitimate, and coordinated. A power has a great deal of it when it can draw on many suppliers, shift routes under pressure, finance a long contest, build and hold coalitions, control chokepoints, enforce sanctions and export controls, set the standards others build to, project force at a distance, keep its institutions trusted, keep its own public willing to pay for all this, and survive the loss of any single node. A navy is one expression of it. So are the dollar system, chip export controls, maritime insurance, the SWIFT bank-messaging network, cloud platforms, satellite constellations, access to ports, and the willingness of allies to stay aligned when costs rise.

Territorial-Production Mass is the ability to convert land, population, industry, resources, proximity and internal control into usable pressure. A power has a great deal of it when it can mobilize its population, command its industry, control resources directly, use nearness to create local superiority, absorb punishment through sheer depth, move forces along internal lines, coerce neighbors, build buffers, impose discipline at home, substitute mass for finesse, and sustain losses that would break a thinner system. An army is one expression of it. So are industrial basins, rail networks, internal security services, manufacturing scale, demographic reserves, missile stockpiles, and the ability to throw concrete, steel, bodies and fuel at a problem until it moves.

Put the two axes together and four positions appear.

The most dangerous powers are the hybrids. They combine territorial depth, population, industry, resources and protected geography with military reach, financial leverage, technological leadership, logistics, access to coalitions and cohesion at home. Rome after it learned the sea is one. Britain at its imperial height is another, and so is the United States after 1945. China is an aspiring one.

A hybrid is simply a power that scores high on both axes, and it has to earn both scores. It is not a magic category for rescuing the model when reality gets messy.

A caution about language. When this essay says “a network power” or “a territorial power,” it means a state leaning that way in the contest at hand, not a permanent type. The United States is a hybrid by its scores, and it fought the Cold War from its network side.

The Recurring Pattern

The classic sequence runs like this.

1. A territorial-production power rises.
2. It dominates its nearby land rivals.
3. A network power fears a unified regional bloc.
4. The network power builds coalitions and restricts flows.
5. The territorial power tries to break out.
6. It seeks buffers, resources, a decisive victory, or immunity from pressure.
7. Expansion creates more enemies, longer supply lines and higher policing costs.
8. The network coalition gains time to mobilize.
9. The territorial power either wins quickly or ends up overextended, isolated, strangled or exhausted from within.

The pattern does not always end in a network victory. Sparta beat Athens. Rome beat Carthage. Russia, with allies, broke both Napoleon and Hitler. Continental mass can absolutely win.

What recurs is the trap:

The territorial power expands to become secure, but expansion often creates the very encirclement, resistance and resource strain it was trying to escape.

That is the strategic treadmill. The political scientist Jack Snyder called the result self-encirclement.

Expansion is not the trap. The trap is expansion whose enforcement cost rises faster than the security it buys.

You conquer the province to secure the border.

Now you need a new border.

Congratulations, emperor. You have unlocked the next logistics nightmare.

The Time Gradient

The earlier draft said time usually favors the side with deeper optionality. That was directionally useful and too blunt. The hardened rule is:

Time favors the side whose cohesion, substitution capacity, replacement rate and enforcement discipline decay more slowly than the opponent’s shortages, extraction costs, production limits, legitimacy costs and coalition pressures accumulate.

Put plainly, both sides are wearing down, and the contest goes to the one wearing down more slowly. Neither network nor mass inherits the clock. The work is in finding out which side that is, and four questions do most of it.

Time is not an ally. Time is a stress test with a calendar.

Network powers stretch time when they can

Network powers try to lengthen a contest with blockade, sanctions, coalition-building, credit, the denial of insurance, export controls, interdiction at sea, attrition on the periphery, the denial of technology, diplomatic isolation and pressure on supply chains.

Immediate collapse is rare. The usual aim is to make the opponent’s system slower, costlier, less trusted, less adaptive, harder to resupply, harder to insure, harder to bank and further behind technologically. Pressure of this kind works best when the target has no substitutes.

Territorial powers fear time when shortages compound

Territorial-production powers often fear a long contest. Coalitions form. Shortages worsen. Occupied territories resist. Fronts multiply and supply lines stretch. Legitimacy erodes, losses outrun replacements, enemies rearm, and factions at home start blaming one another.

This is why Napoleon and Hitler both felt pressed to force a decision. As long as Britain stayed in the war, the continental hegemon could not declare the game won. The longer Britain survived, the more coalition finance, naval blockade and Russian depth came to matter, and against Hitler, American weight as well.

But network pressure can leak

Russia since 2022 is the warning label.

A territorial power with resources to sell can answer network pressure by recruiting substitute nodes: tankers with obscured ownership, known as the shadow fleet; trade routed through third countries; alternative payment systems; gray-market supply chains; neutral intermediaries; partners who are sympathetic or merely opportunistic; overland routes; making at home what can no longer be imported; wartime repression; forced austerity.

If the network cannot enforce the restriction, the pressure leaks. If the target can recruit substitutes faster than its shortages accumulate, time may not favor the network at all.

The modern network is powerful. It is not magic.

A blockade with holes is not a blockade.

It is a sieve wearing admiral’s stripes.

How Network Power Breaks

Network power is strong because it is distributed, adaptive and full of options.

It is brittle because it depends on trust, legitimacy, coordination, predictable institutions, and the willingness of people to bear real costs for benefits that are abstract and shared.

A network power breaks when the nodes stop believing the network is worth maintaining. There are six ways that happens.

Domestic consent

Network systems are expensive. They need bases, fleets, diplomatic commitments, subsidies, open markets, open institutions, immigration, technical standards, legal enforcement, and sometimes wars in places most citizens could not find on a map. The core of the network has to keep asking its own people to pay now for optionality later.

That argument can fail. Consent erodes when casualties rise, when inequality grows, when allies seem ungrateful and elites seem corrupt, when the benefits feel abstract and the rivals feel distant, when the public wants to pull back, and when the core starts to feel exploited by its own network.

A network power can dominate the world and still lose patience with the job.

Empire does not always fall because barbarians breach the gate. Sometimes the accountant, the voter and the exhausted sailor quietly vote to stop paying the maintenance bill.

Alliance cohesion

Network powers are usually coalition managers, which is a strength and a vulnerability. Allies can defect, free-ride, hedge and delay. They can demand subsidies, refuse to escalate, cut side deals, and disagree about what the threat is. They can fear being abandoned and fear being dragged into someone else’s war, often at the same moment.

Network power needs alignment across many actors with different interests.

The territorial power often needs only one crack.

The network power has to keep the whole bridge standing.

Institutional trust

Network power rests on systems that others trust: contracts, courts, reserve currencies, insurance, standards, payment rails, alliances, compliance regimes, technology licences and intelligence sharing.

Turning those systems into weapons can be effective. Every use also teaches everyone else that the system is a weapon.

Rivals cannot replace such systems quickly. Replacement is hard. But overuse speeds up the search for alternatives. Sanctions, export controls, asset freezes, exclusion from payment systems and the denial of insurance are not free moves.

They spend trust.

Sometimes spending it is worth it.

Sometimes the bill arrives later, with interest.

Substitutable nodes and routes

A network is resilient when flows can reroute. It is brittle when too many of them depend on a handful of nodes: one cluster of chip factories, one payment rail, one canal, one cloud provider, one rare-earth processor, one strait, one cable route, one software ecosystem, one dominant insurance market.

Network power likes chokepoints when it controls them.

Network power hates chokepoints when it depends on them.

The noose is attached to both necks.

Excellent engineering, gentlemen.

Rival-network recruitment

This variable matters enough to stand alone. Imposing a sanction is the easy half. The operational question is whether the target can recruit enough alternative flows to make the sanction survivable: other buyers, neutral shippers, non-aligned banks, friendly ports, permissive jurisdictions, and the rest of the list above.

The more the target recruits, the less time favors the network power.

Legitimacy at the core

The hegemon’s legitimacy at home is part of the network. If the core loses faith in the mission, the network degrades even while the ships still float and the cables still carry packets.

That faith decays when voters come to see allies as parasites, when elites cannot explain the strategy, when institutions lose credibility, when infrastructure at home looks neglected, when wars seem optional or endless, when trade feels like deindustrialization, when financial power seems to serve insiders, and when national identity no longer supports the work of maintenance.

The network can be strategically brilliant and politically unsellable.

That still counts as brittleness.

How Territorial Power Breaks

Territorial-production power is strong because it can command mass.

It is brittle because mass has weight. The more territory, people, borders, occupied zones, factions and security commitments a state controls, the more energy it burns just to remain coherent.

Network powers can die of entropy.

Territorial powers can die of gravity.

Command rigidity

Territorial systems prize control, and control can harden into rigidity. It does so when bad news is punished, when subordinates stop improvising, when local knowledge is ignored, when doctrine sets like concrete, when the leader’s preference overrides reality, when success means pleasing the center, when retreat becomes politically impossible, and when military decisions turn into tests of loyalty.

A territorial power can have vast mass and still misapply it catastrophically.

The map says the army is there.

The radio says the army is still there.

The graves say otherwise.

Corruption and the cost of extraction

Territorial power often extracts directly: grain, taxes, labor, conscripts, minerals, fuel, tribute, industrial output, occupied land. Extraction can strengthen the center. It can also rot the system. The more a regime depends on coercive extraction, the more it tends to produce corruption, black markets, evasion, sabotage, hollow reporting, local resentment, predatory elites and collapsing morale.

Resource wealth can multiply the brittleness, if it pays for perpetual war without ever forcing the institutions to adapt.

You can loot the province.

You cannot always make the province productive.

Over-centralization

Mass systems centralize because the center fears fragmentation, and over-centralization creates blindness. The ruler cannot see the periphery clearly. The periphery lies to survive. The bureaucracy optimizes for pleasing the center. The center then decides on the basis of curated fiction.

This is how a state can have enormous mass and still repeatedly punch itself in the face.

Factional militarism

A territorial-production power can be captured by factions that turn the state’s power toward their own prestige, an ideological fantasy or a bureaucratic rivalry. The danger is greatest when military factions act on their own authority, when civilian control is weak, when the regime needs victory for its legitimacy, when stories of humiliation drive strategy, when anxiety about resources tips into panic, and when rivals inside the regime compete by escalating.

Imperial Japan is the textbook case. The warning applies to any state where the formal strategy says one thing and the factional machinery does another.

Demographic drag

Territorial mass depends on people.

People age. People leave. People stop having children. People resist mobilization. People lose faith.

Demographic drag cuts into military recruitment, the labor supply, innovation, fiscal room, the appetite for risk, the capacity to care for the old, and the willingness to absorb casualties. A territorial power can look enormous on the map while its usable human base is shrinking.

Maps do not show age pyramids.

They should.

Legitimacy that depends on victory

Some regimes need victory to survive, and that is dangerous. Such a state may prefer a bad war to visible stagnation. It may escalate because admitting failure is politically lethal. It may double down because the leadership cannot come home empty-handed.

The pattern is brutal:

The regime needs victory to maintain legitimacy, but the pursuit of victory raises the costs that undermine legitimacy.

The territorial treadmill runs through the mind as well as the map.

Territorializing the Network

To territorialize a network is to convert flows into dependencies that can be controlled.

A territorial-production power tries to make the network less optional and more geographic, so that other people’s flows pass through systems it can influence, price, delay, monitor, deny or coerce. Ships and ports are instruments of that conversion. They are not the conversion itself.

The mechanism

A power territorializes the network by building or controlling the things that flows have to pass through: ports, rail corridors and pipelines; payment rails and platforms; industrial and technical standards; rare-earth processing and other dominant manufacturing inputs; data centers, cable routes and satellite services; chip supply chains; shipping capacity; commercial infrastructure with military uses; access to a market that others cannot do without; and debts, contracts or concessions that carry strategic leverage.

The conversion is from

“You can choose among flows”

to

“Your flow depends on my node.”

That is the stone learning to flow: mass acquiring the habits of a network.

The idea is older than the vocabulary. In 1945 the economist Albert Hirschman showed how Germany in the 1930s had turned trade with its smaller neighbors into political leverage, by making itself the partner they could not afford to lose.

Plate · Territorializing the network

The Continental System as archetype

Napoleon’s Continental System is the early model. Unable to beat Britain at sea after Trafalgar, he decreed from Berlin in 1806 that the ports of France and its allies were closed to British trade, and spent the next six years widening the ban. He was trying to reorganize a continent’s trade geography by coercion: to turn flows he could not control into territory he could.

The failure is instructive. Policing the system meant coercing more states, occupying more coastline and punishing more defectors. Portugal would not comply, so France invaded it in 1807. That led into Spain the following year, and a war that bled the French army for six more. Russia left the system at the end of 1810. The breach was a leading cause of the march on Moscow in 1812.

The attempt to territorialize the trade network produced the very resistance and overextension it was meant to solve. The cost of enforcement exceeded the optionality denied. That gives the rule:

A territorial power that tries to territorialize flows pays an enforcement cost and provokes coalitions. If those costs rise faster than the control gained, the strategy recreates the encirclement it was designed to escape.

China’s version

China’s program is much wider than a navy. It runs through shipbuilding, ports, manufacturing chains, payment channels, standards, rare earths and access to its market. Section XIII lists it in full.

One recent episode shows the mechanism working. China refines about nine-tenths of the world’s rare earths and makes a still larger share of the permanent magnets built from them. In April 2025, in answer to American tariffs, Beijing put seven of those elements and the magnets under export licences. Within two months a Ford plant in Chicago had stopped for a week, and parts suppliers in Europe were shutting production lines. That July, Washington let Nvidia resume selling an AI chip to China, and the Commerce Secretary said the decision was part of the bargaining over magnets. Beijing widened its controls in October and, after the two presidents met, suspended the widening for a year. The April licences remain in force.

A material the world had treated as a commodity turned out to pass through one node. The rule above held too: in June 2026 alone Washington committed nearly $3 billion to building magnet supply outside China.

A fair objection arises here. When Washington restricts chips, this essay calls it network power. When Beijing restricts magnets, it calls it territorializing the network. The act is the same. The two terms mark where each side started: one inherited its nodes with a network that others chose to join, and the other is a mass power building nodes by out-producing everyone else. Once built, a node works the same way for whoever holds it, and it invites the same search for substitutes.

The debt-trap guardrail

This concept must not collapse lazily into “debt-trap diplomacy,” the claim that Chinese lending is designed to seize assets from borrowers who default.

Some claims of that kind can be sourced.

Some are contested.

Some are garbage wearing a think-tank tie.

A disciplined version sorts the claims into three tiers.

Recipient states have agency. Local elites have incentives. Domestic politics matter. Dependency can be real without turning every borrower into a puppet.

A useful model does not need cartoons.

The cartoons are for television panels and emotionally needy empires.

Chokepoints

Chokepoints are where flows become terrain.

The classic ones are straits, canals and the fortresses that command them: Gibraltar, Suez, Malta, Panama, Singapore, Hormuz, Malacca, Bab el-Mandeb, the Bosporus and the Dardanelles.

The modern ones are less visible: bank-messaging and payment rails, maritime insurance, the extreme-ultraviolet lithography machines that only the Dutch firm ASML makes, the leading-edge chip factories of TSMC in Taiwan, the plants that package finished chips, the data centers of the big cloud providers, undersea cables, satellite constellations, rare-earth refining, the graphics processors that AI depends on, app stores, operating systems, logistics software, and the licensing offices that administer export controls.

The principle is the same:

If many flows must pass through a small number of controllable points, those points become strategic terrain.

Plate · Where flows become terrain

Chokepoints are not automatically decisive. How much one matters depends on how dependent the target is, what substitutes exist, whether the closure can be enforced, how long it lasts, what stockpiles and alternative suppliers are available, how much pain the politics will tolerate, how far the military risk escalates, whether the coalition holds, and how fast everyone adapts.

A chokepoint with no substitutes is a weapon.

A chokepoint with leakage is a tax.

A chokepoint with many substitutes is a nuisance wearing a scary hat.

Hormuz, 2026

As I revise this in October 2026, one of the oldest names on that list is being tested in public. On February 28 the United States and Israel began striking Iran, and Iran closed the Strait of Hormuz to shipping it considers unfriendly. About a fifth of the oil the world consumes normally passes through it, and the International Energy Agency called the closure the largest supply disruption in the history of the global oil market. A deal signed in June fell apart after three weeks. In early October fewer than twenty cargo ships a day were making the passage, against about 125 before the war.

I did not forecast this, and reading it through the framework now is the after-the-fact fitting that the model-loss rules forbid. So I will not count it as a confirmation. I will record what it puts to the framework.

First, the state holding the chokepoint is no great power of either kind. It is a regional power beside a strait, with mines, missiles and drones, and it has kept that strait mostly shut for seven months with the world’s strongest navy nearby. This framework lists control of chokepoints under network optionality. Hormuz says proximity, which belongs to the mass axis, can do the same job.

Second, the three-line rule held up reasonably well. For importers with thin reserves the closure was a weapon: the Philippines declared a national energy emergency within a month. For exporters with pipelines to other coasts it became a tax, and by October the region’s crude exports were back near their prewar volume by way of the Red Sea and the Gulf of Oman. Iran also sold exemptions: Lloyd’s List reported in March that at least one operator had paid for passage.

Third, the cost did not fall along alliance lines. The importers who suffered included American allies. China, the great power most exposed on paper, cut its crude imports by about a third in the spring and has so far ridden it out on 1.4 billion barrels of stockpiled crude, a diverse set of suppliers, and an electric-vehicle fleet that had already cut its need for fuel.

The episode is not over. It is still the best evidence in this essay for asking about substitutes, stockpiles and enforcement before calling any dependency fatal.

Logistics: Where Map Fantasy Dies

Territorial powers draw arrows.

Network powers attack the arrow’s fuel, food, rail, credit, ports, repair yards, spare parts and political legitimacy.

Conquest expands requirements. It does not automatically expand usable capacity. A captured thing is only the first link in a chain.

Break any link and the possession stays inert. A captured port is not maritime power unless ships can dock, unload, refuel, repair, communicate and survive attack there. A captured chip factory is not semiconductor sovereignty unless the whole chain upstream and downstream of it survives too.

This is why Germany’s hunt for oil in the Second World War kept creating more oil problems. In the summer of 1942 the army drove for the Caucasus. It took the small field at Maikop, which Soviet engineers had wrecked before leaving, and never reached Grozny or Baku.

It needed oil to reach Baku.

Baku was too far away to solve the shortage that was stopping it.

The solution required the thing it was trying to obtain.

That is not strategy.

That is a snake eating its own logistics tail.

Coalition Management

Network powers are usually coalition managers. They do not always need to defeat the territorial power directly. They need to keep enough enemies, pressures and constraints in play that it never fully consolidates.

Britain paid subsidies to one coalition after another against Napoleon. It sustained the resistance to Germany twice. The United States turned British survival into an Atlantic strategy, then built NATO and its Pacific alliances against the Soviet Union. Today it relies on Japan, South Korea, Australia, the Philippines, Taiwan, NATO, a partial alignment with India, and European cooperation on technology and finance to balance China.

The territorial power’s nightmare is a rotating cast of enemies that never fully goes away.

You win the campaign.

Now fight the coalition DLC.

Again.

Forever.

Alliances are an asset and a liability at once. They add bases, legitimacy, manpower, geography, intelligence, industry, economic depth and diplomatic cover. They also add partners who can say no, fights over who pays, disagreements over escalation, other countries’ domestic politics, incompatible readings of the threat, and the twin fears of abandonment and entrapment. When Washington called on its allies in March 2026 to help reopen the Strait of Hormuz by force, most of them, Germany, Britain and Japan among them, at first refused.

A coalition is not a single organism.

It is a sack of cats duct-taped to a strategic concept.

Sometimes the cats kill the enemy.

Sometimes they claw through the bag.

The Historical Gauntlet

This section does not prove the framework.

It stress-tests it.

A clever framework can explain almost anything after the fact, and every reading below was made after the fact. The cases show what the mechanisms look like in operation. They cannot validate the model. Only forecasts made in advance can, which is the business of Sections XVI and XVII. So read each case as a test of mechanism, and keep asking what would have made the framework lose.

Athens and Sparta

Athens was a maritime, commercial network power. It lived on its fleet, its walls, the tribute of its allies, its trade and its sea routes, above all the grain route from the Black Sea. Sparta was a land power that relied on its army, its discipline and a coalition of its own. When the war began in 431 BCE, Athens could watch its countryside burn summer after summer and survive, so long as the Long Walls to its port held and the sea stayed open.

The labels “sea” and “land” do not explain why Athens lost. It suffered a compound failure: a plague that killed perhaps a quarter of its people and its leading statesman, Pericles; faction at home; revolts among its allies; the destruction of an entire expeditionary force in Sicily in 413; the loss of men and money that followed; and, at the end, a Spartan fleet paid for with Persian gold, which destroyed the Athenian navy at Aegospotami in 405 and closed the grain route. Athens starved and surrendered the following year. The land power won at sea, with ships someone else had bought.

Athens shows that network power collapses when internal brittleness makes the interdiction of its flows decisive.

Internal collapse and external interdiction are not rival explanations. They interact. A resilient Athens might have survived the external pressure. A brittle one could not.

Rome and Carthage

Carthage began as the stronger maritime and commercial power. Rome began as the stronger power in territory and manpower.

Rome’s achievement was to learn enough of the sea without giving up its capacity to mobilize the land. According to the Greek historian Polybius, it built its first war fleet in 260 BCE by copying a Carthaginian warship that had run aground. It then lost whole fleets to storms and built them again. When Hannibal invaded Italy overland in the next war, Rome’s command of the sea helped keep him from being reinforced.

Once Rome was a hybrid, Carthage’s network advantage eroded.

The most dangerous competitor is the territorial power that learns network power without losing its mass base.

This matters for China today. The reason is not that China is Rome. The reason is that the conversion is a familiar one.

Napoleon and the Continental System

Section VIII told this story. Napoleon dominated the continent and could not beat Britain at sea. Britain survived, financed coalitions, controlled the maritime flows and remained a platform for resistance. Closing Europe’s ports to it meant coercing more territory than France could sustainably manage.

Enforcement problems became wars.

Wars became overextension.

Overextension revived the coalition.

Territorializing a network can create the very coalition pressure it seeks to escape, when the cost of enforcement rises faster than the control gained.

The Dutch Republic

The Dutch Republic of the seventeenth century was a classic network power, built on trade, finance, shipping, commercial institutions and maritime reach. It was also small: under two million people, beside an England nearly three times that size and a France ten times.

Its case shows that a network advantage can decay when rivals build network capacity of their own and bring more mass to it. In 1672, which the Dutch still call the Disaster Year, England and France attacked it together. It survived, through its fleet and by flooding its own countryside. But forty more years of war with France loaded it with debt that a state of its size could not carry, and commercial and naval primacy passed to England.

The case warns against romanticizing finance and trade as self-defending. They are not. A network position has to be defended by institutions, military capacity, fiscal resilience and political cohesion.

Network power can be outscaled if it lacks the mass to defend its network position.

Imperial Japan

Imperial Japan is a test in two parts, and they must not be blurred.

1931 and 1937. When officers of the Japanese army engineered the seizure of Manchuria in 1931, and when a clash near Beijing in 1937 was allowed to grow into full war with China, Japan still had real network options. It traded, borrowed and bought its oil on world markets. The turn toward territorial expansion came from factional militarism, an army acting on its own authority, ideology, the politics of prestige, anxiety about resources, weak civilian control, and a legitimacy that had become tied to expansion.

The Depression had narrowed Japan’s markets, and its planners feared blockade. But necessity did not force the choice. The fever drove the machine.

This half tests whether the model can handle choices that are irrational, or driven by faction. States do not always take the economically sensible path.

1941. By the summer of 1941 the position had changed. The United States, which supplied most of Japan’s oil, froze Japanese assets and cut the oil off. Japan’s time gradient was compressed, its substitutes narrowed, and its anxiety about resources became acute. The choice for war was catastrophic, but it was no longer the same kind of choice as 1931 or 1937. Japan’s leaders chose a desperate territorial breakout, toward the oilfields of the Dutch East Indies, because the network constraint had become severe.

A power may choose territorial expansion out of factional fever before it is necessary, or out of compressed time after its flow options have been closed. Do not blur the two.

Germany

Germany in both world wars was an industrial land power with severe maritime and resource weaknesses. It had a strong army, strong industry, a central position, and real tactical and operational skill. It lacked secure oil, protected geography, global finance, maritime reach, depth of alliances and durable access to flows from outside. In the first war a British blockade helped wear it down to collapse.

In the second, Germany’s problem after the fall of France was real. Britain was still in the war. The blockade continued. The United States was moving toward Britain. The Soviet Union remained a massive factor to the east.

Hitler’s answer was catastrophic: invade the Soviet Union, as his ideology had always wanted, without having defeated Britain, without secure oil, without adequate logistics, and with an occupation policy that maximized resistance.

A real strategic dilemma can still produce a stupid strategic answer.

The model should explain the dilemma.

It should not excuse the decision.

The United States and the Soviet Union

The Cold War set a network coalition, led by a hybrid, against a continental empire.

The United States had two oceans, the dollar, naval reach, NATO, Japan, bases around the world, technology networks, a dynamic consumer economy, and alliances that were legitimate at least by comparison with the Soviet ones. The Soviet Union had land mass, military depth, resources, ideology, internal coercion, buffer states, an enormous army and nuclear weapons.

The American strategy was containment, the word the diplomat George Kennan supplied in 1947: prevent Soviet consolidation outward, sustain the alliances, control the sea lanes, compete in technology, apply financial pressure, and leave the Soviet system to carry its own inefficiencies. Nuclear weapons kept the contest from being settled by force, which left endurance to settle it.

The Soviet Union was not defeated by one magic lever. It was constrained by a network coalition while its internal burdens compounded.

Network power often wins by preventing decisive expansion and letting the territorial system’s internal costs accumulate.

Russia since 2022

Russia is the live warning against network triumphalism.

Western network tools imposed real costs: sanctions, the freezing of some $300 billion in central-bank reserves, export controls, pressure on insurers and shippers, the denial of technology, diplomatic isolation and a price cap on Russian oil.

Russia adapted. It sold its oil to China and India instead. In the summer of 2026 between two-thirds and three-quarters of its seaborne crude moved on tankers outside Western insurance. It found alternative suppliers, substituted at home under state direction, mobilized for war and repressed dissent.

That does not mean the sanctions failed. It means network pressure did not automatically produce a decision.

At the same time Russia’s territorial brittleness has become visible: casualties that the British government puts at about a million and a half killed and wounded, corruption, lost equipment, command failures, demographic strain, inflation, dependence on coercion, and a legitimacy tied to the outcome of the war. In the war’s fifth year the central bank’s interest rate stands at 14 percent, the liquid part of the state’s reserve fund has fallen from more than 6 percent of GDP to under 2, and Ukrainian drones have at times shut down a quarter of the country’s oil refining. The fighting continues.

The Russia case has to be held with both hands:

Network pressure leaked because rival-network recruitment worked. Territorial mass endured because coercion and resources bought time. But territorial brittleness accumulated, because extraction, casualties and command rigidity did not disappear.

The model fails if it reaches for whichever half of that reading is convenient after the fact. Before analyzing a case, it has to name the variables it expects to dominate.

No retroactive prophecy.

That is astrology with footnotes.

The Present Case: The United States and China

The modern contest is this:

American-led network power against a China-centered territorial-industrial power that is trying to become a hybrid.

The United States controls much of the high-end connective tissue.

China controls enormous industrial mass.

That is the tension. Much of it plays out along the first island chain, the arc of islands that runs from Japan through Taiwan to the Philippines and encloses China’s coast.

The dollar

The dollar is a core asset of American network power. It is strategic infrastructure, and it is a system of trust.

Use it wisely and it compounds leverage.

Overuse it and others slowly search for alternatives. In 1999 the dollar made up about 71 percent of the world’s official currency reserves. It now makes up under 57. Sanctions are only one cause of that drift, and most of the lost share went to a scatter of smaller currencies, not to a rival.

Slowly does not mean never.

America’s other exposures are the ones Section VI describes.

Shipbuilding

China’s shipbuilding position deserves particular attention. In 2025, by China’s own industry figures, its yards delivered 56 percent of the world’s new shipping tonnage and took 69 percent of new orders. The American share of commercial shipbuilding is about a tenth of one percent.

Shipbuilding is latent maritime power.

Not every commercial yard becomes a navy yard. But industrial capacity, infrastructure that serves both uses, trained labor, steel, ports and a whole ecosystem of maritime production all count.

When a territorial-industrial power learns the sea, it converts mass into network reach.

Rome smiles from the grave.

Carthage does not.

Where China is exposed

China is powerful, and it is not invulnerable. Its weaknesses include dependence on imported energy, exposure at maritime chokepoints, imports of food and animal feed, bottlenecks in semiconductors, exposure to the dollar and to foreign demand, an aging population, the risk of encirclement by a coalition, the geography of Taiwan, limited combat experience far from home, and a domestic legitimacy tied to economic performance and nationalism.

The “Malacca dilemma,” a phrase attributed to Hu Jintao in 2003, is real. In the first half of 2025 some 7.9 million barrels of crude a day reached China through the Strait of Malacca, about 70 percent of its crude imports. But closing Malacca does not collapse China on a timetable anyone should bet a war on. Hormuz in 2026 is a mild test of that, since Iran let cargoes for China through. Even so, imports fell by a third, and the system has bent without breaking.

The hard version of the claim is:

“China’s energy and trade system contains exposure at maritime chokepoints that becomes strategically dangerous if the disruption lasts long enough, substitutes are insufficient, stockpiles run down, the enforcing coalition holds, and escalation stays manageable.”

That is less sexy.

It is also less stupid.

Taiwan is a society, not terrain

Taiwan is many things at once: a geographic barrier, a political community, a democratic society, the place where most of the world’s most advanced chips are made, a military tripwire, a test of American alliance credibility, an anchor of the first island chain, a chokepoint for the world economy, and a multiplier of human cost.

The earlier draft called Taiwan “Malta, Gibraltar, and a semiconductor fab cluster rolled into one geopolitical heart attack.”

That line stays, because it is accurate enough and fun enough to earn its food.

It needed a correction:

Taiwan is a society with agency, sitting on terrain that other people’s strategies keep flattening into a node.

Plate · Human node, not terrain

Taiwanese identity, political preference, willingness to resist, civil defense, cohesion among its elites, military readiness and the sheer difficulty of occupation all enter the strategic equation.

A fait accompli is not accomplished because an arrow on a map reaches Taipei.

Occupation is not control.

Control is not legitimacy.

Legitimacy is not obedience.

And obedience is not guaranteed when 23 million people have their own vote in the matter, whether Beijing, Washington or the nearest clever analyst likes it or not.

What China is trying to do

China’s strategic project is to reduce its vulnerability to network pressure while increasing its ability to impose a local decision. That means building an ocean-going navy; making its coastal waters too dangerous for an opponent to enter, which strategists call anti-access and area denial, or A2/AD; raising its domestic chip capability; diversifying its energy routes through pipelines and overland corridors; expanding alternative payment rails; strengthening its security in food and energy; dominating key manufacturing chains; building maritime infrastructure with military uses; splitting American alliances; making intervention over Taiwan too costly and American escalation too risky politically; building alternatives to the systems the West controls; and making foreign firms and states dependent on Chinese production.

China does not need to defeat the United States globally. It needs to make American intervention inside the first island chain slow, costly, uncertain and politically divisive.

That is the modern territorial power’s breakout strategy.

What the United States is trying to do

The American strategy is to preserve network optionality and prevent China from closing East Asia into a Chinese sphere. That means maintaining alliances; holding the first island chain; preserving Taiwan’s autonomy, or at least preventing its annexation by force; controlling the semiconductor chokepoints; keeping financial leverage; maintaining dominance on and under the sea; strengthening Japan, the Philippines, Australia and South Korea; encouraging India and Vietnam to hedge; hardening space and cyber systems; diversifying supply chains; preserving credibility without tipping allies into fatalism; and keeping China from gaining control of critical flows.

The United States does not need to conquer China. It needs to prevent China from achieving a decisive regional fait accompli.

That is classic network-power logic.

The central question

The contest can be put as one question:

Can China convert industrial mass, proximity and territorialized dependencies into regional dominance before the American-led network converts optionality, alliances, finance, technology and maritime reach into durable containment?

Or, from the American side:

Can the United States preserve enough maritime, financial, technological and alliance control to keep China from creating a closed regional sphere?

That is the modern version of the old struggle between sea and continent.

Same physics.

More domains.

Worse acronyms.

The Diagnostic Checklist

Use this when analyzing any great-power contest. It has five parts.

Network optionality

• What flows must keep moving?
• How many substitutes exist?
• Who controls the chokepoints?
• Can routes shift under pressure?
• Can finance and insurance continue?
• Can sanctions be enforced and technology denial sustained?
• Can allies absorb the costs?
• Can domestic support survive a long contest?
• Can logistics reach the theater under fire?
• Can the core of the network keep its legitimacy?

Territorial-production mass

• What land, population and industry can be mobilized directly?
• What resources are controlled at home, and what must be imported?
• How secure are the internal lines?
• How much local military superiority exists?
• Can the state coerce its neighbors?
• Can it absorb casualties and replace equipment?
• Can its command adapt?
• Is corruption hollowing the system?
• Does legitimacy depend on victory?
• Are the demographics improving or deteriorating?

The time gradient

• Whose shortages accumulate faster?
• Whose cohesion decays faster?
• Whose replacement rate is stronger?
• Whose coalition improves or fractures with time?
• Whose domestic legitimacy is more fragile?
• Whose flows are easier to substitute?
• Whose pressure campaign leaks more?
• Who needs a decision sooner, and who can survive delay?

Never ask “who benefits from time?” in the abstract. Ask which specific variables improve or deteriorate, and over what horizon.

Territorializing the network

• Is a power converting flows into dependencies it controls?
• Which of those are physical, institutional, technological, financial, legal or political?
• What does enforcement cost?
• What coalitions does it provoke?
• Are the recipient states passive, coerced, opportunistic or bargaining hard?
• Does the strategy reduce vulnerability or create new resistance?

Human cost and agency

• Who lives inside the node?
• What do they want, and how stable are those preferences?
• How much resistance is plausible?
• What would occupation require, and how much coercion would control demand?
• Would the attacker inherit a prize, a ruin or an insurgent society?
• Are analysts treating people as terrain because the map looks cleaner that way?

If your model has no place for human agency, your model is not strategic.

It is cartographic taxidermy.

Evidence Discipline

The framework uses ordinal labels, meaning rankings and directions. It does not pretend to a precision it lacks.

No decimal-point cosplay.

No “Network Optionality = 73.4.”

No fake confidence wearing a spreadsheet helmet.

It uses four scales:

• High, Medium or Low
• Improving, Stable or Deteriorating
• Substitutable, Bottlenecked or Non-substitutable
• Observed Fact, Inference or Speculation, the three tiers set out in the debt-trap guardrail above

Every assignment should come with five things: the proxy, meaning the observable stand-in for the variable; the source and its date; the evidence tier; the reason it matters; and the counter-indicator that would flip it.

An entry has this form:

“Alliance Cohesion: Medium, Deteriorating. Proxy: public disputes over defense spending, delayed basing agreements, divergent enforcement of sanctions. Evidence tier: Inference. Counter-indicator: renewed multi-year basing commitments together with synchronized enforcement of export controls.”

A real entry would also name the source and date behind each proxy.

This is discipline, not mathematical certainty. Judgment stays. The goal is to make it inspectable before the outcome is known.

Model-Loss Rules

The model loses if it can survive only by cheating.

It cheats when it:

• renames the loser after the fact
• calls every hard case a “hybrid”
• adds a missing variable only once the outcome is known
• treats “brittleness” as a magic explanation for everything
• refuses to specify a time horizon
• refuses to name the dominant variables before the analysis
• converts every failed prediction into “still consistent with the model”
• uses vivid language to hide a weak mechanism

Before running a case, state the variable you expect to dominate, the time gradient you expect, the mechanism, the evidence that would confirm it and the evidence that would falsify it.

A model that cannot lose cannot teach.

It can only recruit.

That is how frameworks become cults with maps.

Forecast Hooks

The framework should not pretend to forecast the probability of a war to the percentage point.

It is not a crystal ball.

It is an engine for conditional forecasts, of this form:

If variables X, Y and Z change in observable ways, then the model expects pressure A to increase, option B to degrade, or strategy C to become less viable by a stated date.

An outcome forecast

A fait accompli over Taiwan becomes more plausible if several conditions converge: American legitimacy at home and alliance cohesion abroad deteriorate; China’s ability to substitute for critical flows improves; its local military coercion stays high; the human cost of subduing Taiwan is underestimated, or Taiwan’s capacity to impose that cost is degraded; financial and blockade tools leak badly; and the regional allies hedge instead of mobilizing.

State it conditionally, and in terms of likelihood. Do not write:

“China will take Taiwan.”

Write:

“If at least two major variables of network deterrence degrade while China’s local coercive capacity and its flow substitutability improve, the model expects the risk of a successful fait accompli attempt to rise.”

Human agency remains active throughout. Taiwanese resistance can raise the cost of occupation. Allied decisions can change the gradient. American domestic politics can recover or decay. Chinese internal politics can speed up or restrain risk-taking.

The structure pressures.

It does not abolish choice.

A leading-indicator forecast

A more useful near-term forecast may not involve a war at all. By a stated audit date, assess whether China’s alternative payment channels grew materially; whether allied enforcement of export controls converged or fragmented; whether Taiwan’s civil defense strengthened or weakened; whether American public support for defending allies rose or fell; whether the concentration of critical supply chains shifted; whether maritime insurance and sanctions regimes leaked; and whether the states of the region moved toward China, toward the United States, or toward hedging.

This creates a forecast trail before the crisis.

Otherwise the model sits around waiting for catastrophe like a ghoul with a clipboard.

Limitations and the Normative Hazard

This framework describes pressures.

It does not declare destiny.

Structure matters. Geography matters. Flows matter. Mass matters. So do ideology, leadership, morale, doctrine, intelligence, luck, weather, disease, personality, corruption, accident, domestic politics, technological surprise, human courage and human stupidity.

Especially human stupidity.

Never underrate the strategic power of some idiot in a room with authority, a bad map and a deadline.

The fatalism trap

A clean model can become dangerous.

If leaders believe conflict is structurally inevitable, they may make it so. If analysts describe Taiwan as a “node” and forget that it is a society, they may lower the perceived cost of war. If strategists treat sanctions, blockades and export controls as frictionless levers, they will ignore adaptation and blowback. If territorial powers believe encirclement is inevitable, they may lunge for a breakout. If network powers believe their system always wins with time, they may drift into complacency.

The model should increase clarity.

It should not sedate moral responsibility.

The agency correction

People are not terrain.

Allies are not chess pieces.

Small states are not just corridors.

Ports are not just dots.

Taiwan is not just a semiconductor node.

Ukraine is not just a buffer.

Sailors, voters, engineers, soldiers, workers and citizens all affect the system. Agency is not moral decoration. It is predictive. A model that ignores it will misread resistance, the cost of occupation, the durability of alliances, legitimacy and the risk of escalation.

That is not ethics floating above strategy.

That is strategy refusing to be stupid.

The Compact Version

Great-power conflict often reduces to a struggle between network optionality and territorial-production mass.

Network powers control flows: trade, finance, sea lanes, alliances, information, technology, logistics, standards and chokepoints.

Territorial-production powers control mass: land, population, industry, armies, resources, buffers, proximity, coercion and internal lines.

Network powers usually seek optionality, coalition pressure, the denial of flows, and time.

Territorial powers usually seek a decision, consolidation, secure resources, buffer space and immunity from outside pressure.

Time favors neither side automatically. It favors the side whose cohesion, substitution capacity, replacement rate and enforcement discipline decay more slowly than the opponent’s shortages, extraction costs, legitimacy burdens and coalition pressures accumulate.

The strongest powers are hybrids, which means high performance on both axes and nothing more mysterious than that.

The modern contest stacks sea, air, space, cyber, finance, semiconductors, cloud computing, AI, supply chains, sanctions, standards and human legitimacy on top of the old geography. The core dynamic remains:

One side weaponizes connectivity. The other weaponizes proximity and mass. The winner is the side that converts its advantage into usable power without breaking its own system in the process.

In one line:

Sea power was never just about the sea; it was about controlling the flows that make power usable.

And a warning label, for the tool itself.

Use this framework like a tool.

Not like a flag.

A good tool makes reality easier to interrogate.

A bad flag makes the tribe feel clever while marching into artillery.

The purpose is to ask better questions before reality answers them with explosives.

First published June 2026 · Revised October 2026