I. The Question Stated Properly
The familiar phrase “graveyard of empires” gets the problem backward. Afghanistan has been conquered often and successfully — by Cyrus, Alexander, the Mauryas, the Kushans, the Sasanians, the Hephthalites, the Arabs, the Samanids, the Ghaznavids, the Mongols, the Timurids, the Mughals and Safavids in partition, and Nader Shah — and it has been governed under most of those regimes for periods measured in generations. The genuine puzzle is narrower and harder: the region has almost never functioned as a self-supporting state within its own borders, drawing its revenue from its own population, defending its own frontiers on its own account, and holding together without either an outward campaign or an outside paymaster.
The distinction matters because it changes the object of study. The question is not “why can Afghanistan not be conquered” (it can), nor “why can Afghanistan not be governed” (it has been), but rather: what has to be true for a state centered in Afghanistan to sustain itself out of Afghanistan?
Framed that way, the historical record yields a short list of enabling conditions, an explanation of why they seldom co-occur, and a shorter list of disabling conditions that have proven far more reliable than the enabling ones.
A terminological caution: “Afghanistan” as a bounded territory is a product of the 1873–1895 Anglo-Russian frontier settlements and the 1893 Durand agreement. Before that the relevant units were Bactria/Balkh, Arachosia/Zabulistan, Gandhara/Nangarhar, Ghor, Khorasan, Kabulistan, and Sistan — regions that faced outward in different directions and were rarely governed as a set. Applying the modern container to the ancient record is an anachronism, but a useful one, because it isolates precisely the thing that turns out to be difficult: making those basins into one polity.
II. What the Long Record Actually Shows
Three political forms recur, and only three.
The imperial core projecting outward. The Kushans (roughly first through third centuries), the Ghaznavids (977–1186), the Ghurids (1150s–1215), the Timurids at Herat (1405–1507), the Hotaki interlude (1709–1738), and the Durrani empire (1747–1818) all had their seat in what is now Afghanistan and drew their wealth from beyond it — from Indian plunder, from transit tolls on the trans-Asian trade, from tribute in Khorasan and Sind. In every case the state’s cohesion was purchased with revenue that did not come from Afghan farmers.
The contested periphery. For far longer stretches, the same territory was divided among empires seated elsewhere: Achaemenid satrapies at Bactria and Arachosia; the Seleucid–Mauryan partition; the Sasanian–Hephthalite frontier; the Mughal–Safavid–Uzbek triple division of Kabul, Kandahar, and Herat in the sixteenth and seventeenth centuries. This is arguably the region’s default condition. It reflects a geographic fact: the settled basins each have a natural external orientation. Herat looks to Mashhad and the Iranian plateau; Balkh to Bukhara and the Oxus lands; Nangarhar and Kandahar to the Indus. The mountain core between them connects nothing to nothing.
The self-contained state. This appears twice with any durability. Abdur Rahman Khan’s amirate (1880–1901) and its Musahiban successor arrangement (1929–1973, most stably under Zahir Shah) are the two intervals in which a government seated at Kabul held the whole territory, faced no serious foreign occupation, and reproduced itself across time. Together they account for roughly 115 of the 279 years since Ahmad Shah Durrani’s accession — and, as will be argued, both rested on international conditions the Afghan state itself did not control.
The Greco-Bactrian kingdom (roughly 250–125 BC) deserves mention as a possible third case, but it was a Hellenistic settler polity oriented toward Indian expansion, and its collapse under nomadic pressure from the north illustrates rather than contradicts the pattern.
III. The Typology of Enabling Conditions
1. An outward valve for extraction. The most reliable historical mechanism of Afghan cohesion is a state that functions as a distribution system for wealth captured elsewhere. Mahmud of Ghazni’s seventeen Indian campaigns funded a court, a slave army, and a building program at Ghazni without taxing Zabulistan into revolt. Ahmad Shah Durrani’s nine invasions of India, culminating at Panipat in 1761, financed the Durrani confederation and gave the Abdali/Durrani tribes a positive reason to accept a paramount chief. The Ghilzai Hotakis took Isfahan in 1722 on the same logic.
The mechanism is straightforward: tribal confederation is expensive, and the price is paid in shares. So long as the shares come from outside, the confederacy has no reason to fracture. The condition is fragile in a specific way — it requires a wealthy and militarily weaker neighbor. British consolidation of the Punjab by 1849 closed the valve permanently, and the Durrani state’s nineteenth-century crisis dates precisely from that closure.
2. Transit revenue in place of plunder. Where the outward valve is closed, the corridor itself can be taxed. Kushan prosperity rested on the trans-Asian trade linking Han China, the Indian ports, and the Roman east; Timurid Herat’s cultural flowering rested on the same routes in their late form. This is the one enabling condition that is genuinely internal to the country’s geography — but it is hostage to developments elsewhere, and the maritime reorientation of Eurasian commerce after about 1500 removed it for four centuries. Modern attempts to restore it (the Ring Road, TAPI, Chabahar and Gwadar corridor schemes, Central Asian transmission projects) are recognizably efforts to rebuild this condition.
3. External subsidy paid without external competition. Abdur Rahman received a British subsidy, arms, and a guarantee against Russian encroachment, in exchange for surrendering foreign policy and accepting the frontier. The Musahiban dynasty later drew American and Soviet aid simultaneously, in a rivalry expressed in dam projects and highway contracts rather than in armed clients. Najibullah’s regime survived on Soviet transfers until 1991; the Islamic Republic on American and allied transfers until 2021.
The subsidized state is the rentier state, and its properties are consistent across cases: the ruler can build an army and a bureaucracy without taxing anyone, which removes the trigger for revolt but also removes any bargain between the state and its population. When the subsidy ends, the state has nothing to fall back on. Najibullah’s government collapsed within roughly four months of the termination of Soviet supply in January 1992; the Afghan National Defense and Security Forces dissolved in weeks once contractor maintenance, air support, and salary flows ceased in 2021. The enabling condition and the fatal condition are the same condition.
4. A monopoly of coercion established before the state attempts anything else. Abdur Rahman’s twenty-one years are the closest thing in the record to genuine state formation: systematic disarmament of tribal levies, a standing army paid from the center, forced population transfers (Ghilzai families resettled north of the Hindu Kush), the reduction of Hazarajat in 1891–1893 and of Kafiristan in 1895–1896, an internal intelligence apparatus, and the subordination of the religious hierarchy to a throne that claimed direct sanction. The Iron Amir understood that the sequence matters: coercion first, then administration, then — never, in his case — reform.
This condition carries its own poison. The violence that produced it produced grievances with a memory measured in generations, and the Hazara dispossession in particular has structured Afghan politics ever since.
5. Frontiers fixed and guaranteed from outside. The Anglo-Russian settlements of 1873 and 1895, the Durand line of 1893, and the 1907 convention converted a contested march into a buffer with a legal existence. Neither empire had an interest in occupying it, and both had an interest in nobody else doing so. Removing the external threat is what freed Abdur Rahman to spend two decades on internal conquest.
6. Mutual abstention by the neighbors. Related to the previous but distinct: Afghanistan is governable when surrounding powers agree not to compete inside it. Afghan society is inexpensive to penetrate and expensive to hold, which creates a standing temptation for neighbors to sponsor factions rather than invade. The intervals of self-government correspond exactly to the intervals when this temptation was suppressed by agreement (1880–1919) or by a rivalry channelled into development aid (1946–1973). The period since 1978 — with Pakistani, Soviet/Russian, Iranian, Indian, Gulf, and American sponsorship all operating on internal clients — is the negative case.
7. A legitimating formula that outranks tribal segmentation. Segmentary societies do not spontaneously produce paramount authority; something has to sit above the qawm. The Afghan repertoire is small and well-worn: dynastic claim (Sadozai, then Barakzai/Mohammadzai), the loya jirga as a ratifying assembly, custody of relics such as the Cloak at Kandahar, and the title of Commander of the Faithful — claimed by Abdur Rahman and again by Mullah Omar in 1996.
The strongest formula available is resistance to a non-Muslim invader, and it works: 1842, 1880, 1919, and the 1980s each produced mobilization no domestic appeal could have generated. But it has no successor. Legitimacy built on the presence of a foreigner expires when the foreigner leaves, which is why victory has repeatedly been followed by factional war — the 1842–43 restoration amid continued factionalism, the collapse of the Peshawar and Islamabad accords into the destruction of Kabul in 1992–1996, and the internal Kandahar–Haqqani strain since 2021.
8. A center that deliberately underreaches. The Musahiban compact is the clearest formulation: the state takes the cities, the army, customs, foreign policy, and the modern sector, and leaves village and family order alone. Four decades of relative quiet followed. This is not weakness but a bargain — the population accepts a center that does not touch it.
IV. Why the Conjunction Is Rare
The conditions above are not merely uncommon individually. They are, in several pairs, mutually antagonistic, and the antagonism explains the rarity better than any account of Afghan “ungovernability.”
Outward extraction requires a weak rich neighbor. Modern states are not weak. The condition that founded the Durrani empire cannot recur.
Subsidy requires patrons interested enough to pay but not interested enough to compete. This is a narrow band. A patron with sufficient stake to fund a government usually has sufficient stake to cultivate insurance policies among its rivals — which is what Pakistan did continuously after 1973 and what several powers did after 2001.
Coercive monopoly requires violence that manufactures the next generation’s fracture lines. Abdur Rahman achieved the state at the cost of the Hazara and Nuristani grievances that have furnished mobilizing material ever since.
Fixed frontiers required an imperial cartography that split the core ethnic group. The Durand line stabilized the state and simultaneously created its permanent foreign-policy wound. No Afghan government has ever accepted it; Pakistan has therefore never wanted a strong Afghan state; and the Pashtun belt straddling it has furnished every insurgency since with rear areas the Afghan state cannot reach.
Underreach forecloses the modernization that would end dependence on subsidy. A state that does not tax, conscript, educate, or register cannot generate domestic revenue, and so remains a rentier — but the attempt to do those things is precisely what has destroyed Afghan governments. Amanullah’s reforms in family law, dress, girls’ education, conscription, and taxation produced the Khost rising of 1924 and the collapse of 1928–1929. The People’s Democratic Party’s decrees of 1978 on land redistribution, bride price, and compulsory literacy produced a countrywide rising before Soviet troops arrived. Both governments were destroyed by the very program that would have made them solvent.
This is the heart of the matter. Afghan governability is not a puzzle about culture or terrain. It is a trap in which each available road to a self-sustaining state runs through a condition that destroys the state.
V. The Disabling Conditions, Ranked
If the enabling conditions are rare and mutually obstructive, the disabling ones are common and mutually reinforcing. In descending order of destructive reliability:
1. Internal extraction. Every Afghan government that has attempted to fund itself by taxing its own population at scale has faced armed revolt. This is the sharpest single predictor in the record.
2. Reform that reaches into the household. Family law, women’s status and schooling, land tenure, and conscription. Every modern collapse is preceded by the center penetrating the qawm — 1928–1929 and 1978–1979 most plainly, and one may read the post-2001 program and the 2021 counter-program as opposite-signed instances of the same pattern.
3. Competitive external sponsorship. This converts ordinary segmentary rivalry — which is self-limiting, because feuds run out of resources — into resupplied factional warfare that does not.
4. Asymmetry between the ethnic composition of the state and of the population. From 1747 to 1978, and again from 1996, the throne was held by Pashtuns and, with the nine-month exception of Habibullah Kalakani in 1929, by Durrani Pashtuns of a narrow set of lineages. Tajik, Hazara, Uzbek, Turkmen, Aimaq, Nuristani, and Baluch populations have never held a durable share of central power. The consequence is structural: every weakening of the center produces regionally based mobilization along ethnic lines, because those are the only lines along which excluded populations have ever been able to organize. The Hazara case adds a sectarian dimension — a Shia population in a state whose legal identity is Sunni Hanafi, carrying the memory of the 1891–1893 conquest and enslavement, the massacres of 1998, and continued targeted violence since 2021.
5. The Durand line and cross-border sanctuary. Structurally the hardest of all to remedy, because it requires a concession by two states whose incentives are opposed. Afghan recognition would cost any Kabul government its nationalist standing; Pakistani accommodation would cost it the strategic depth doctrine it has pursued for fifty years.
6. The jihad-legitimacy trap. Cohesion generated against an outsider is real but non-transferable, and its dissolution is not a failure of the victors’ intentions but a property of the formula.
7. Rentier dependence as such. Listed among the enabling conditions above and again here, because it belongs in both places. It is the only condition in the typology that reliably produces both the interval of order and the manner of its ending.
VI. What Follows
Three conclusions seem warranted.
First, the periods of Afghan self-government are best understood as intervals granted by the international system rather than achievements of Afghan state capacity. Both 1880–1901 and 1929–1973 depended on arrangements among outside powers, and both ended when those arrangements did — in 1919 by Afghan initiative but within a collapsing imperial order, and in 1973–1978 as Cold War rivalry moved from dams to clients.
Second, the standard prescriptions run against the record. Building central administrative capacity, extending the state’s reach to the village, raising domestic revenue, and reforming customary law are the recognized components of state-building, and in Afghanistan they are also the recognized precipitants of collapse. Any program that treats Afghan governance as a capacity problem is proposing the mechanism of failure as the remedy.
Third, and least comfortably: the one enabling condition that is internal to the country’s own geography rather than borrowed from outside is transit. The corridor is the only asset Afghanistan holds that does not depend on a weak neighbor, a generous patron, or a level of internal violence that mortgages the next generation. Whether transit revenue can be reconstituted in an era of maritime and air freight, and whether the required regional cooperation is any less rare than the conditions catalogued above, is the question on which the next interval — if there is one — most likely turns.
A note on method: this paper treats “governability” as a property of a configuration rather than of a people, and therefore counts a condition as enabling only where its absence is followed by breakdown and its presence by durability across at least two separated instances. Cases with a single instance — the Greco-Bactrian kingdom, the Hotaki interlude — are used illustratively rather than evidentially.
