Why Ancient Empires Collapsed at Similar Scales

In the year 9 CE, the Roman general Publius Quinctilius Varus led three legions — roughly 20,000 men — into the Teutoburg Forest in what is now northwestern Germany, and lost them all. The Germanic chieftain Arminius had arranged an ambush so complete that the legions ceased to exist as coherent units within hours. When news reached Augustus in Rome, the emperor reportedly wandered his palace at night crying out “Varus, give me back my legions!” The defeat at Teutoburg is conventionally cited as the moment Rome accepted the Rhine as its northeastern boundary — as if the legions themselves decided where the empire ended.

But Augustus’s grief, vivid as it is, misidentifies the cause. Rome was not defeated by Germanic warriors at Teutoburg. Rome was defeated by its own information system. The supply chains, the communication networks, the administrative capacity required to project power reliably beyond the Rhine — through forest terrain, across hundreds of miles of hostile territory, in a region with no existing urban infrastructure to tax or co-opt — would have strained Roman administrative capacity beyond its operational limits even if Arminius had never existed. The river was not primarily a military boundary. It was the boundary of legible, administrable territory. Augustus did not lose three legions because Germania was too dangerous. He lost them because governing Germania would have cost more than it could possibly yield.

The Consistent Ceiling

The coincidence across ancient empires is striking enough to demand explanation. The Roman Empire at its Antonine peak governed somewhere between 50 and 70 million people. The Han Dynasty at its height, roughly contemporaneous, governed between 50 and 60 million people across a territory of similar geographic scale. The Maurya Empire under Ashoka, a century earlier, is estimated at 50 to 60 million subjects. The Abbasid Caliphate at its ninth-century peak encompassed a population in the same range. The Ottoman Empire at its sixteenth-century maximum was larger — perhaps 25 to 35 million, but from a smaller pre-modern base — and began its slow administrative dissolution at roughly comparable administrative ratios.

These are rough estimates, subject to significant scholarly debate. But the clustering is too consistent to be coincidental. Every major agrarian empire that attempted to govern more than approximately 60 to 70 million people from a single administrative centre ran into structural difficulties that, over decades or centuries, produced fragmentation, regional autonomy, or outright collapse. The specific trigger for collapse varied — military invasion, succession crises, fiscal bankruptcy, climate disruption. But the underlying fragility was consistent, and it preceded the trigger in every case.

The explanation that best fits the evidence is administrative bandwidth, a concept borrowed from information theory but applicable to any organisation that depends on transmitting accurate information over large distances in real time. An empire is, at its core, a system for collecting taxes, maintaining order, resolving disputes, and projecting military force. All of these functions require accurate information: about harvests, population, property, crime, military threats, and the reliability of local officials. That information has to flow from the periphery to the centre, instructions have to flow back from the centre to the periphery, and the whole system has to be fast enough that the instructions remain relevant when they arrive.

In a world without telecommunications, this information transmission happened at the speed of a horse — roughly 50 kilometres per day on good roads. The Roman cursus publicus, the imperial postal system, was a remarkable feat of infrastructure that could move messages from London to Rome in roughly a month under ideal conditions. That was genuinely fast by the standards of the ancient world. It was also too slow to govern a continent in real time. By the time a provincial governor’s report reached Rome, was considered, and a response was sent back, months had elapsed. The local situation had changed. The instructions were stale. The governor was effectively governing autonomously whether the central administration wanted him to or not.

The Information Cost of Scale

This creates a specific and unavoidable scaling problem that every large pre-modern state encountered. As an empire expands, the number of peripheral administrative units increases roughly proportionally with territory. But the communication cost of coordinating those units increases faster than proportionally, because each additional unit must be coordinated not just with the centre but with its neighbours, and the chain of dependencies that must be maintained to keep the system coherent grows combinatorially. Rome could govern Italy with modest administrative overhead. Governing Italy plus Gaul plus the Balkans plus Asia Minor plus Egypt plus North Africa required an administrative apparatus whose complexity grew faster than its productive base.

The Roman response to this problem was delegation — the creation of provincial governors with broad local authority. But delegation is not a solution; it is a restatement of the problem. A governor who must be trusted with broad local authority is a governor who has the capacity to defect. Roman history is largely a history of governors who did exactly that, either slowly through the quiet accumulation of local loyalty and private wealth, or suddenly through outright revolt. Every mechanism the central government devised to monitor and constrain governors created additional administrative overhead and additional opportunities for the monitoring layer itself to defect.

The Han Dynasty faced the same problem through its system of commanderies and counties, and responded similarly. The Abbasid Caliphate, which governed the most geographically dispersed empire in history to that point, developed one of the most sophisticated bureaucratic systems of the ancient world — the diwan system for tracking tax revenues, the barid postal network for communication, the elaborate protocols for judicial and administrative appointment — and still fractured progressively from the ninth century as governors in Khorasan, Egypt, and eventually Anatolia achieved de facto independence while maintaining formal allegiance to a caliph who could do nothing to reverse it.

The pattern is so consistent that it suggests a near-universal law of pre-modern state formation: the administrative capacity of an agrarian empire scales roughly logarithmically with territory and population, while the administrative demand scales linearly or super-linearly. At some point, demand exceeds capacity, and the state either contracts, fragments, or develops compensating institutions — local autonomy, aristocratic sub-contracting, religious administration — that effectively decentralise governance while maintaining a nominal central authority.

The Military Illusion

Popular accounts of imperial collapse typically focus on military explanations: the legions were overwhelmed by barbarians; the Tang were defeated by the An Lushan rebellion; the Abbasids were crushed by the Mongols. These events happened and they matter. But they are almost always proximate causes rather than fundamental ones. Military defeat is a symptom of administrative fragility, not its cause.

The Roman legions in the third century were not individually inferior to their predecessors in the first century. They were arguably better equipped and better trained. What had changed was the administrative system behind them: the fiscal infrastructure was undermined by monetary debasement, the supply chains were strained by the increasing difficulty of extracting and forwarding taxes from distant provinces, the officer class was increasingly recruited from the very frontier populations they were meant to control, and the central government had lost the capacity to maintain the loyalty of commanders who discovered that loyalty to a distant and unstable emperor was a worse career strategy than loyalty to a successful general who paid in the same vicinity as where they fought.

When the Visigoths sacked Rome in 410 CE, the event was shocking to contemporaries as a symbol, but the Roman state had effectively ceased to function as a coherent administrative entity in large parts of its western territory decades earlier. The sack was the punctuation on a sentence that had been written over the previous century of administrative decomposition. Similarly, when the Mongols destroyed Baghdad in 1258 and killed the last Abbasid caliph, the event ended a dynasty that had not actually governed most of the territory nominally under its authority for at least four hundred years.

This sequencing matters enormously for how we interpret collapse. If military defeat were the fundamental cause, we would expect empires to collapse suddenly when faced with superior military force. What we observe instead is a decades-long or centuries-long process of administrative erosion, followed eventually by a military event that finalises what the erosion had already accomplished. The barbarians did not destroy the Roman Empire. The Roman Empire dissolved, and the barbarians walked into the space left behind.

The Modern Echo

The administrative bandwidth constraint did not disappear with the invention of the telegraph, the telephone, or digital communication. It transformed. Modern states can transmit information instantly across any distance, which eliminates the communication lag that killed ancient empires. But they have replaced that constraint with a different one: the cognitive and political capacity to process information and generate coherent responses.

A modern government receives vastly more information than any Roman emperor ever did, from vastly more sources, with vastly less latency. What it does not have is a correspondingly larger capacity to make good decisions with that information. The pathologies of modern large states — regulatory capture, bureaucratic inertia, inability to reform entrenched interests, the gap between formal policy and actual implementation — are the contemporary equivalents of Rome’s provincial governor problem. The information arrives. The response is still slow, still subject to corruption and defection, still incapable of maintaining coherent central direction over vast and diverse territories.

The specific ceiling has shifted. Modern states can govern populations far larger than 70 million without fragmenting — the United States, China, India, and the European Union all maintain some form of political coherence at scales the Romans would have found incomprehensible. But the administrative problems have not disappeared. They have been deferred by technology and partially compensated by democratic mechanisms that align local and central interests better than any ancient empire managed. The fundamental tension between the information demands of large-scale governance and the cognitive limits of any administrative system remains. Ancient empires hit their ceiling at 50 to 70 million people. Modern states hit theirs at different points and manage differently. But the ceiling exists, and no empire in history has found a permanent way around it. The question is never whether a political system will encounter its administrative limits. It is only when, and what form the resulting fragmentation will take.