Why some countries can't be built

In 2007 I was twenty-two and working for the U.S. Department of Defense in Kabul, and my job for part of that year was ghost soldiers.

The Afghan National Army paid its men off paper. A commander reported how many soldiers he had, a roster went up the chain, time sheets were filled in by hand, and the payroll was released against those records. Nobody checked whether the soldiers existed. The commander said they did. Some of them didn't. The money for the ones who didn't went to the commander, or was split with whoever signed the roster, and the auditors who were supposed to catch it mostly didn't, for a reason that took me a while to understand. In areas where the auditor didn't know the commander, the audits worked. In areas where they knew each other, they shared the profit. That was the whole system: a ring, held together by who was related to whom.

Our answer was biometric ID cards. Every soldier gets a card tied to his fingerprints; he uses it to clock in and again when he's paid; no card, no money. We piloted it with the 201st Corps, which was based near Kabul, and it worked. Ghost soldiers in that corps stopped getting paid because there was no one to hold the card.

Then we tried to take it to the forward operating bases, which is where most of the army and most of the fraud actually were, and we couldn't. Not because anyone fought us. Because there was no reliable electricity, the readers broke, the logistics of getting equipment and technicians to the places that needed them were beyond what anyone could sustain, and the local oversight the system depended on didn't exist out there. The cards improved oversight where oversight already existed. Where it didn't, the ring kept running, and it ran until the day the republic fell.

I've thought about that project for nineteen years. It was a good solution. It was the right solution. And it hit the same wall that every reform in Afghanistan has hit for as long as anyone has been trying, and I want to try to describe the wall.

What a country is

Some places work. Other places don't. By "work" I don't mean rich. I mean the basic thing: millions of strangers manage to share roads, courts, money, schools, and police without robbing each other. When strangers can trust strangers, you have a country. When they can't, you can send any amount of money and soldiers and constitutions and it doesn't take. You don't have a country. You have a lot of households.

Everyone trusts their family. That's universal. What varies from place to place is how far trust extends past the family, and that distance decides everything else: what kind of government you can have, what kind of economy, whether a contract means anything. In Norway the radius is enormous. In the village my parents came from, it ends at the bloodline.

I'm Afghan by birth. Afghanistan has been trying to become a country for a long time and hasn't managed it. Not for lack of intelligence, or wanting a state, or trying. Every attempt, and there have been many, has collapsed. I think there are three reasons, and they stack.

Geography

Picture a farmer in Bamiyan eight hundred years ago. His valley is fifteen kilometers long. The mountains on each side go up three thousand meters. The next valley is two days' walk. Who does he trust with his life? His father, his brothers, his sons, his cousins to the second and third degree, the families who have farmed next to his for ten generations. That's the world. Everyone else is a stranger, and strangers are dangerous.

Run that across thousands of valleys for three thousand years and you don't get a country. You get thousands of small, hard, self-contained tribes, absolute inside and indifferent or hostile outside. The mountains don't cut Afghanistan off from the world so much as they cut Afghans off from each other.

But geography isn't enough on its own. Switzerland is mountains. Norway is mountains. Nepal, the Caucasus, the Scottish Highlands, all mountains, all tribal once, all different now. Geography is the soil. Something else is the seed.

Cousin marriage

About 46 percent of marriages in Afghanistan are between cousins, 28 percent between first cousins, and among Pashtuns the numbers run higher. In most of Europe it's under one percent; in the United States, 0.2. My parents were second cousins. Several people in my family married cousins. This isn't something I read about.

One cousin marriage looks small. Your son marries your brother's daughter, the land stays in the family, so does the money and the alliances, and your grandchildren are raised by people who already trust you completely. The family's loyalty doesn't extend outward; it loops back in and tightens. Do that for thirty generations and the most powerful institution in anyone's life is the extended clan. More powerful than the village, the city, the religion, or the state.

The anthropologist Joseph Henrich has measured this. His kinship intensity index, built from cousin-marriage rates and a few related variables, puts Afghanistan at the top of the world, and his argument, which I find convincing, is that a society's radius of trust is largely set by how much survival depends on kin. The more you need your cousins, the less you can afford to trust anyone else, and the harder it is to build anything that requires trusting anyone else.

The part of his work that surprised me most is that medieval Europe looked like this too. Cousin marriage was common, clans were powerful, and Europe in the year 900 resembled Helmand today more than it resembled Stockholm. What changed was the Catholic Church, which starting in the 6th century banned cousin marriage, kept widening the ban for five hundred years, and in doing so took the European clan apart. With the clan gone, Europeans had to build trust with non-relatives, and they did: guilds, parishes, universities, chartered towns, eventually corporations. The institutions the West is proud of grew in ground that had been cleared of clan loyalty on purpose. It took half a millennium.

Afghanistan never had that. Islam could have done it, and the Prophet preached against tribalism, but in Afghanistan the tribe absorbed the religion rather than the other way around. When Pashtunwali and Sharia conflict, Pashtunwali usually wins. So the clans never broke, and the radius never widened.

Resets

Other tribal, mountainous places got out. Switzerland's valleys controlled trade routes and had to cooperate or lose the money, and the canton replaced the clan. Tibet got monasteries, which asked for a loyalty above family. The Scottish Highlands got the British, who after 1745 abolished the chiefs' authority, banned the tartans, and cleared the glens, which was brutal and worked. China got two thousand years of a bureaucracy that demanded loyalty to the emperor over loyalty to family. In every case something climbed above the clan and took its place.

Afghanistan never got that, and the third reason is why: every time something started to climb, an empire came through and knocked it down. Persians, Alexander, Kushans, Arabs, Ghaznavids, Mongols twice, Mughals, the British three times, the Soviets, the Americans. That's not a list of invasions. It's a list of resets.

The modern resets did something worse than destroy. They paid. When Washington funds the government in Kabul, that government doesn't need to negotiate with its own people, doesn't need a tax base, doesn't need to build anything its population trusts. It cashes the check. The republic I worked for collected less than a quarter of its budget from Afghans; the rest came from abroad, most of it American. When the money stopped in 2021, the government was gone in eleven days. The army I'd tried to put ID cards on didn't fight. There was nothing underneath.

I saw the small version of this in 2007 and didn't recognize it. The ghost-soldier payroll was foreign money flowing through a structure that had no reason to be honest with itself, and our cards could fix the part of it that was already half-fixed. The rest was the country.

The reformers

The usual story is that Afghan reformers were incompetent or corrupt or puppets. I don't think so. Some were brilliant. They failed for the same structural reason: they tried to widen the radius faster than the society could carry.

Amanullah Khan came back from a European tour in 1928 and called the tribal elders to a Loya Jirga to announce compulsory girls' education, a ban on the veil in public, Western dress for officials, civil courts in place of the jirgas, and to make the point unmistakable he brought his wife Soraya to the assembly, unveiled, in a European dress, in front of two thousand men who had never seen such a thing. Within a year the eastern tribes were in revolt, a Tajik bandit with no royal blood was sitting on the throne in Kabul, and every reform was rolled back. Amanullah died in exile in Zurich in 1960.

The communists in 1978 did it faster. Decree 7 outlawed forced marriage; Decree 8 broke up the estates. Both defensible on paper, both issued with no consultation and no capacity to enforce them, and the countryside was in revolt within months. The Soviets invaded two years later and lost a million Afghans, fifteen thousand of their own soldiers, and the war. The mistake wasn't Marxism. Marxism took root in China and Vietnam. The mistake was thinking you could remove the tribe in one generation. They pulled out the load-bearing wall and the building fell on them.

Karzai and Ghani, with American money, built a system everyone on the ground knew was hollow, and I include myself in everyone. We told ourselves it would catch on. It couldn't.

Then there's the one who almost made it. Zahir Shah took the throne at nineteen, four years after the country threw his cousin Amanullah out, and he drew the right lesson: don't fight the tribes. He left the countryside alone and spent forty years slowly building a state in Kabul, a university, a bureaucracy, a small real army, foreign relations. For four decades Afghanistan was stable. Not modern or rich, but stable, which no one else has managed since.

In 1964 he gave the country a constitution with parliamentary democracy. It was the most enlightened thing he ever did and it broke everything. Parties formed along the only lines that meant anything, ethnic and ideological, and within nine years his own cousin Daoud overthrew him, within fifteen the communists overthrew Daoud, and within twenty the country was at war with itself and the Soviet Union.

That's the claim I'd stake the essay on. Trust between strangers can't be widened faster than a society can carry, and even forty patient years can be too fast. If someone can show me a low-trust society that introduced multi-party competition without reverting, I'll revise. I haven't found one.

The Taliban

If the problem is that nothing ever climbed above the tribe, what do we do with the fact that, for the first time in fifty years, something homegrown has?

The Taliban control all thirty-four provinces with no active insurgency, which the republic, the Soviets, and the mujahideen government never did. They collect around two billion dollars a year in domestic revenue, close to the republic's peak, from an economy a quarter the size and with no foreign budget support. The ghost soldiers vanished from the payroll the week they took over; there was no one left to pay them. And they've run the state for five years under sanctions without it collapsing.

They've widened the radius. A Pashtun from Helmand and a Pashtun from Khost used to share a language and not much else; now they share a movement, sit in the same assembly, fight in the same units, and trust each other in a way their fathers wouldn't have. For many rural Pashtuns that's the first time trust has reached past the kinship network.

And it stops there. It doesn't extend to Hazaras, who are Shia and have been persecuted by every Pashtun-led government in the country's history. It doesn't reliably extend to Tajiks or Uzbeks. It doesn't extend to women. Half the country. The circle got wider, and it still excludes more people than it holds.

What I hear from family in Kabul is not the same as what I read. The leaders have changed how they look: tailored white clothes, groomed beards, polished turbans, men who want to be taken for statesmen. Their policy has turned toward the economy. The Qosh Tepa canal is being dug across the north, the kind of project that stalled for the whole of the republic. Kabul is being rebuilt. On September 7 a Saudi company signed a $200 million exploration deal for the Kushk-Tirpul basin in the west, on a 25-year contract; as I understand it, an earlier deal of that kind fell apart because the Amir wanted Sharia to govern any dispute, and this one didn't. In the capital, the modesty police have stopped enforcing on women. They dress the way they did under the republic, with a scarf and an uncovered face. Beards aren't checked. Men wear Western clothes.

Two things haven't moved, and they're the two that matter. Girls can attend English courses and madrasas and cannot attend school or university. And there's no constitution, nothing that makes the state something other than the movement, nothing that would outlast the men currently running it. Until those move, people won't trust it, and they're right not to.

I expect them to move, slowly, for a reason Ibn Khaldun wrote down in the fourteenth century. Tribal conquerors moderate. The Arabs who took Persia in 651 were desert tribesmen with no bureaucracy; within a hundred years their grandsons were hiring Persian viziers and their great-grandsons were building Baghdad. The Mongols who took Persia were Persian within four generations. The Manchu who took China became Chinese. The mechanism is always the same: the fighters move into the city, their children grow up there, running a customs service and a foreign policy turns out to be more complicated than us-and-them, and the founders' certainty wears down. Taliban fighters from Helmand are living in a city of five million. Their kids are growing up next to kids whose parents fought against them, with phones in their hands. The Taliban of 2026 are not the Taliban of 2000, and the ones running things in 2050 won't be these.

The objection

The strongest argument against all this is Acemoglu's: institutions are chosen. North and South Korea started with the same geography, culture, and kinship, and one is a prison and the other is a $1.7 trillion economy. Institutions decided that, not cousins.

I think that's half right, and the half that's right supports the argument. Both Koreas built institutions that fit the substrate. The North built a totalitarian state, which maps onto strong family hierarchy uncomfortably well, and froze the radius. The South built a dictatorship first, under Park, which also fit, and then over forty years let it open. South Korea didn't succeed because it installed democracy in 1948. It succeeded because the sequence was paced to what the society could carry. Botswana did something similar by building its democracy on top of the tribal kgotla councils instead of in place of them. And northern and southern Italy, same country, same constitution, same currency, are one of the richest and one of the poorest regions in Europe because the north spent a thousand years in civic city-states and the south spent them under absentee landlords.

Institutions matter. They also run on a substrate, and the ones that take are the ones that match the trust the society already has, or widen it at a pace the society can bear.

Some readers of the first version of this essay called it deterministic. I understand why, and I'd answer this way: my parents were second cousins, and I'm writing this from San Diego, in English, having built companies with people from six countries I'm not related to. The radius moves. It moved in me in one generation. What it won't do is move on a foreign government's schedule.

What's changing

Three things are widening the Afghan radius right now, and nobody is directing any of them.

The internet. A fifteen-year-old in Kabul has a phone, and on it she watches people in Seoul and Lagos and Istanbul, learns English from YouTube, and follows Iranian and Saudi women getting out from under the same lock she lives under. Her grandfather didn't know any of that existed. Her radius starts wider than his ever got.

The cities. Kabul had a million people in 1979 and has more than five million now, and you cannot run five million people on tribal logic. They share buses, hospitals, jobs, and neighborhoods with people they aren't related to, and they extend small acts of trust to strangers every day whether they mean to or not. Cities have always done this. Afghanistan has never had them at this scale.

The diaspora. Millions of us are abroad. We call home, we send money, we visit, we marry across families and borders. I've been one of those phone calls for nineteen years. It's slow and it's invisible to every government and it's the largest sustained pressure on Afghan norms there has ever been.

None of this is fast, and none of it produces a parliament next year. But it works in the only way that has ever worked, from the bottom, at the speed the society can absorb, and it puts pressure on the regime and on the society at the same time. The phone that widens a girl's world in Kabul is the same phone that makes a Taliban minister's teenager argue with him at dinner.

What to do

Stop trying to install countries. You can't install one. A country is a trust network that grows at the speed of generations, not congressional cycles.

But doing nothing doesn't work either. North Korea has been isolated since 1953 and Eritrea since 1993 and neither has widened. So the question is what kind of engagement works, and I think the answer is one thing: keep the information channels open.

Not aid, not arms, not constitutions, not sanctions packages tied to conditions the current regime will never meet. Internet infrastructure that survives disruption. Online education in Dari and Pashto delivered to the phones of girls who can't go to school. Making it easier, not harder, for Afghans to work on remote platforms even under sanctions. Diaspora journalism in Afghan languages that the regime can't silence. And "the channel stays open" as the default position in every trade and sanctions conversation.

It's cheap, it needs no one's permission, and it works on the society and the regime at once. It's the closest thing to a Catholic Church for Afghanistan I can imagine: slow, accidental, and operating below the level any government can reach.

Two generations

My three youngest children were born in the United States. Their radius of trust started where mine ended. They'll never need to know which cousin to call to get a document stamped, and if I described the auditor who split the payroll with the commander because they were from the same village, they would find it as strange as I find Norway.

That's one generation. In Kabul it will take three, maybe more. The children of the Taliban fighters now living in the city will be part of it, whether their fathers intend it or not. The cities will do what cities do, and the phones will do what phones do.

Keep the channels open. Then wait.

Sources. Joseph Henrich, Jonathan Schulz, Duman Bahrami-Rad, and Jonathan Beauchamp, "The Church, intensive kinship, and global psychological variation," Science, 2019; Henrich, The WEIRDest People in the World, 2020; Jack Goody, The Development of the Family and Marriage in Europe, 1983. Cousin-marriage rates: Saify and Saadat, Journal of Biosocial Science, 2012 (46.2 percent consanguineous, 27.8 percent first-cousin, n = 7,484); UNICEF MICS. Republic aid dependence and ghost soldiers: SIGAR quarterly and lessons-learned reports. Soviet casualties: Russian Defense Ministry figures. Amanullah and Zahir Shah: Vartan Gregorian, The Emergence of Modern Afghanistan, 1969; Thomas Barfield, Afghanistan: A Cultural and Political History, 2010. PDPA decrees: Afghanistan Analysts Network archive. Taliban revenue: World Bank Afghanistan Development Updates. Saudi deal: Arab News and Amu TV, September 7, 2026. Institutions: Acemoglu and Robinson, Why Nations Fail, 2012; Putnam, Making Democracy Work, 1993. Trust: Fukuyama, Trust, 1995. Asabiyyah: Ibn Khaldun, The Muqaddimah, Rosenthal translation. A first version of this essay appeared on X in May 2026.