🎓 Why Successful Empires Preserve Divisions Instead of Erasing Them

How empires from Assyria to the Soviets maintained local differences as a tool of control, and what happened when they stopped.

I’ve always tended to imagine empires as steamrollers, imposing their cultural preferences (language, law, food, clothes) on everyone they conquer. The Romans made everyone speak Latin, the Brits got everyone playing cricket, that sort of thing. But the more I dig into it the more I think Josephine Quinn, author of In Search of the Phoenicians, is right and that many successful empires often preserved local differences, sometimes even created new ones, as a deliberate strategy of control.

She wrote “the maintenance of local differences is, after all, a traditional tool of imperial hegemony.” It’s nestled in with a longer argument about Carthaginian power networks, but I came across this quote in one of my Readwise reviews and was startled because it’s one of those things that is trivially true when I think about it — the whole point of an empire is that it encompasses many cultures, that’s what distinguishes it from a kingdom — but also somehow not how I ever thought of empires functioning before.

Put another way: encouraging diversity was a deliberate strategy for maintaining control.

If anything I thought the opposite, because in Against the Grain, James C. Scott argued that small polities with “flimsy political links and weak hierarchies” could resist imperial absorption through their fragmentation — “divide that ye be not ruled.”

How the Neo-Assyrians Ran an Empire

The Neo-Assyrian Empire (roughly 900–600 BCE) is the earliest imperial system with enough surviving records for us to really grasp anything about. There were two zones of Assyrian control: māt Aššur (“the Land of Assur,” directly governed territory) and nīr Aššur (“the Yoke of Assur,” the lighter burden of submission imposed on client states).

Conquered territory could become either a province under a royally appointed governor, or a vassal state with its own local ruler intact. In provinces, governors had no hereditary claim — they served at the king’s pleasure, and eunuchs were preferred because they could not found dynasties. In vassal states, local rulers kept their thrones, passed their offices by inheritance, and ran their own internal affairs. They just had to accept the presence of a royal delegate who represented Assyrian interests in the local government.

At least that’s how I think it worked? Remarkably little material culture of Assyrian origin survives in the provincial archaeological record. We’re going mostly off of written records, and there’s at least one bilingual record where the same official is called “governor” in the Akkadian text and “king” in the Aramaic version. Even so, Babylonia, though dominated by Assyria for most of the eighth and seventh centuries, was never treated as part of Assyria proper. Assyrian kings were crowned separately as “kings of Babylonia” and participated in Babylonian religious festivals.

The Assyrian reputation for terror was well-documented by their own victory monuments and their victims’ accounts, but tho people along the imperial periphery had to pay taxes and supply labor, they were still allowed and expected to just continue worshiping their own gods and speaking their own languages. This helped create a sort of wheel-and-spoke arrangement, because it let them keep conquered regions connected vertically to the Assyrian center, but never horizontally to one another.

A compliant vassal who collected your tribute was cheaper than a garrison... and you don’t need as big a garrison if your enemies don’t have a shared context from which to team up on you!

It was sort of a “mosaic model” in which each region was characterized by different imperial policies. To make it harder for resistance to cohere, they also resettled people. Millions of people were relocated over a couple of centuries. The Assyrians used gardening metaphors in their royal rhetoric, describing deportees as valuable plants transplanted to new soil. Deportees were dispersed and mixed with other populations, which is why the ten tribes of Israel “disappeared” while Judahites deported later by the Babylonians to a single location were able to preserve their collective identity.

Assyrian royal rhetoric used depictions of cultural diversity to demonstrate the truly universal nature of the empire, which is a weird sort of echo of the “diversity is our strength” and “a nation built on immigrants” rhetoric that we have today.

Phoenician Cosmopolitanism as Defense

Meanwhile, the Levantine cities that later became known as “Phoenician” (Tyre, Sidon, Byblos, Arados) were living out the opposite side of this strategy. Quinn argues that these cities never formed a political unit and rarely cooperated with each other. They kept themselves small and cosmopolitan, uncommitted to any particular way of being. It made them difficult to categorize, and difficult to conquer as a bloc.

So the Assyrians dealt with them individually, demanding tribute from “kings of the seacoast” one by one. The cities’ mutual rivalry did the work of imperial control without even needing much in the way of conquest.

Tyre probably benefited from Assyria’s protection and preferential treatment, receiving privileged access to all harbors under Assyrian control.

Quinn observed that “economic and political connectivity is often more important to the ruling class than ethnic difference.”

The Persians who succeeded Assyria as overlords of the Phoenician cities designed their satrapy system so that a satrap, a general, and a secretary each reported separately to the center, preventing any one official from consolidating local power. Local elites were happy to play along as long as the new boss paid lip service to local religious preferences. Under Persian domination, Arados borrowed Syrian visual styles, Byblos used Egyptian ones.

Rome: The Long Experiment

Rome governed the largest empire in Mediterranean history with strikingly few imperial officials. When Rome formed each province, the Senate drew up a lex provinciae, a charter defining territorial limits, the number of towns, and their rights and duties. But the actual business of administration was, wherever possible, left to people who were already doing it. As Bret Devereaux has detailed in his series on Roman provincial administration, “Romans by and large did not interfere with the functioning of local courts or local laws or local customs or local religion.” Mary Beard makes a complementary point in SPQR: “pre-existing local hierarchies were transformed into hierarchies that served Rome.”

Religion worked the same way. The practice of interpretatio romana (identifying foreign gods with Roman equivalents, like Sulis-Minerva at Bath) allowed subject peoples to maintain their religions while linking them loosely to Roman frameworks. In Roman Egypt, emperors appeared in traditional pharaonic regalia on temple reliefs, maintained the nome system of administrative divisions, and permitted Egyptian priesthoods to inscribe them with pharaonic epithets. Rome’s religious flexibility had one consistent limit: monotheists, specifically Jews and Christians who refused to participate in the additive framework.

David Mattingly’s concept of “discrepant identities” captures the full picture: different groups within the empire experienced Roman rule in radically different ways, and any single narrative of “Romanization” obscures this variation.

The Mongol Empire: Thin Layer, Local Everything

The universal layer the Mongol Empire imposed was comparatively thin. Less than a million Mongols ended up governing an empire of perhaps 100 million people across Eurasia.

So like the Romans (and many other governments), they set up a census to enable taxation. The famous postal relay system enabled communication, and imperially appointed overseers called darughachi monitored local rulers without replacing them, much like the Assyrian qepu discussed above). The yasa legal framework was a sort of “federal overlay” rather than replacement. In the Ilkhanate, yarghu tribunals administered Mongol customary law alongside sharia courts, sometimes operating in the same building.

Below the thin crust of the imperial culture, governance varied wildly by region:

Genghis Khan sent administrators from one end of the empire to govern the other. Foreigners on both ends meant neither could easily consolidate local power, although it seems like such cross-posted administrators were the exception, not the rule.

In the Yuan Dynasty, Genghis’ grandson Kublai Khan restored the Chinese Imperial Secretariat but imposed a four-tier ethnic hierarchy that reminded me quite a lot of the colonial-era European powers in Rwanda and India (which we’ll get to later).

Over in Russia, the Golden Horde founded by a different one of Ghenghis’ grandsons (Batu Khan) let local princes kept their thrones after traveling to the capital to receive a patent confirming their right to rule. They collected taxes on the khan’s behalf, and the khans played princes against each other.

It’s a tale as old as empire.

Ottoman Millets: The Most Confusing System

The Ottoman Empire had a formal millet system. This tripped me up but it’s not “millet” as in the grain, it’s from Arabic milla, meaning “religious community” or “confessional group.” They organized non-Muslim subjects into self-governing religious communities: Orthodox Christians under the Ecumenical Patriarch, Armenian Christians under their own hierarchy, Jews under a chief rabbi. Each millet set its own laws, collected its own taxes, and ran its own courts.

Karen Barkey has argued that this diversity management was central to Ottoman longevity. In her chapter “The Ottomans and Diversity” in Culture and Order in World Politics, she describes how the Ottomans managed what she calls “the mobile markers of difference that prevent the formation of horizontal connections that disrupt the imperial structure.” The state dealt only with leaders of each millet, never individual members. Even their taxes were lump sum and levied on entire communities, reinforcing the sense of each group as a separate corporate entity. Groups stayed separate, and their leaders were cool with the arrangement because it backed up their own authority.

That said, I’m not entirely sure she’s right. Apparently the formal millet system as traditionally described had “no grounding in historical reality” in the classical era and the term millet was not widely applied to non-Muslim communities until the nineteenth century. Relevant to my argument, tho, the Ottoman state retroactively organized its subjects into neat confessional boxes, also rather like how Belgium ended up hardening the Hutu and Tutsi Rwandan castes into fixed ethnic categories, or how Britain invented “martial races” in India.

Bear with me as I try to go in chronological order here, but basically, the Ottoman empire was invested in giving their subjects discrete identities, to help keep groups separate enough that they would have a hard time uniting against outside rule.

Then, during the Tanzimat reforms, the Ottomans tried to shift from maintaining differences to promoting “Ottomanism” — equality for all subjects. Sultan Mahmud II declared: “From now on, I do not want to recognize Muslims outside the mosque, Christians outside the church, Jews outside the synagogue.” But as Barkey argues in her Cambridge chapter, without its historical focus on diversity, “the empire lacked incentives to keep increasingly independent peripheries within its orbit, and the imperial galaxy became a cosmos of nations.”

The whole system depended on peripheral elites getting something out of the arrangement: local authority, legal autonomy, a share of the tax base. When the Ottomans switched from managing diversity to imposing uniformity, apparently they removed the incentive structure that kept those elites loyal.

Obviously the actual fall of the Ottomans was multicausal. For one thing, note that I have not mentioned the corruption of the bureaucracy at all, or the Janissaries shifting to be hereditary. For another, I’ve argued in the past that mismanaging locusts was a big part of what brought the Ottomans down, and certainly the Jews were not treated well by the Ottomans, but Aaron Aaronsohn wasn’t exactly an elite, either.

Regardless, I’m not (yet!) an expert on the Ottomans or anything, and I know some of my readers are, so I’m kind of just noticing patterns and hoping someone will push back if I’m misunderstanding the situation or reading the wrong sources.

The Inca Empire: Mandatory Distinctiveness

For an empire with fewer sources to get confused by, I present: the Inca! They also maintained local differences as a pretty extreme version of this deliberate administrative strategy.

Conquered peoples were obligated to maintain distinctive clothing, hairstyles, and head-deformation practices. Each group had to be visually identifiable at a glance. I suppose the idea was that if you can tell at a roadside checkpoint whether someone is group A or group B by what they’re wearing, you don’t need more complicated systems. Local lords were kept in place as intermediaries, collecting tribute and organizing labor in basically the same manner as Assyrian vassals, Persian satraps, and Roman client kings.

Quechua was imposed as the administrative lingua franca, but local languages were tolerated. Like the Assyrians, they had a mitmaq (resettlement) program, although they relocated entire communities to distant provinces, not individual families. Estimates suggest a little over a third of some provincial populations were relocated mitmaqkuna, who were placed above the natives of the region. The Inca ended up creating ethnic patchworks that prevented any single group from dominating a region. The transplanted communities kept their identities and answered to their original kurakas even in their new locations, so the empire could track them through two overlapping systems of control. As far as I can tell this was a deliberate imperial policy intended to reduce the likelihood of rebellion:

The resettlement policy can be reduced to the pursuit of a set of goals. The first was to prevent rebellion in an empire too geographically large to be under military control. This was achieved by breaking up local community groups that may have been particularly likely to rebel and systematically relocating them to distant regions. Often they were resettled in recently conquered towns where the Indigenous peoples spoke a different language, to prevent the uprising of a coherent coalition. The mitmaqkuna were installed as the elite upper class (hanan) within their new communities, where the remaining Indigenous group became the lower class (hurin). The foreigners had the necessary political power and knowledge to dispense traditional Inka ceremonies and imperial order to the Indigenous people of that province, inevitably establishing a social disparity between the mitmaqkuna and the Natives.
[...]
both authors concur that the resettlement system created a healthy conflict (from the imperial perspective) that would focus the people’s attention on local competition, which served to increase working productivity and allowed imperial observation of work capability and diligence for later recruitment . Importantly, the people would be distracted from unifying into greater rebellion against the Inka state.

The same pattern cropping up in the largest pre-Columbian empire is a pretty interesting indicator that this is “convergent evolution” and not just one bad idea about empire spreading down through successor states trapped in a loop (sorry, sorry, I’m still annoyed at Christopher Beckworth’s The Scythian Empire claims that since the Chinese had governors they had to be Scythian, it was a pre-Mongol Scythian invention, or something).

Colonial Innovations in Division

Anyway to get us a bit closer to the modern day, European colonial powers spent a lot of time defining group identities to facilitate governance too.

For example, the system of “indirect rule” in Northern Nigeria preserved existing political institutions and incorporated them into the colonial administrative system. Local rulers kept their courts and their treasuries. In southeastern Nigeria, where the Igbo had no centralized authority, Lugard simply manufactured his own chiefs — the notorious “warrant chiefs” who lacked any traditional legitimacy. The bifurcated state helped keep resistance ethnically and locally defined, so it didn’t end up coalescing into a single movement.

After the 1857 Indian Revolt, during which Hindus and Muslims had fought side by side against British rule, Lord Elphinstone reportedly declared: “Divide et impera was an old Roman maxim, and it shall be ours.” The British introduced separate electorates where Muslims voted only for Muslims and Hindus for Hindus. Politicians now had to appeal to narrow religious identities, because the system rewarded communal loyalty.

The military was no different: the British restructured the Indian Army to recruit selectively from Sikhs, Gurkhas, and Punjabi Muslims while excluding communities that had rebelled. Nicholas Dirks argues in (the blurb of, I haven’t read it yet) Castes of Mind that “under British domination caste did become a single term capable of naming and above all subsuming India’s diverse forms of social identity and organization.” The census hardened what had been fluid social categories into fixed administrative ones, which is particularly ironic given how strongly “castes” are associated with India.

Tbh it reminded me of how pizza is an American invention but people associate it with mainland Italy, and now Italy is where people go to get the “best” pizza even though tomatoes come from ~Mexico.

The Spanish colonial casta system constructed an elaborate classification hierarchy combining religious, racial, and genealogical categories. Spain maintained hegemony in a multiethnic society “without overt force or coercion” through patron-client networks that “promoted divisions among the poor.” When plebeian classes briefly overcame racial divisions in the 1692 Mexico City riot, the colonial state exposed fissures among the ethnically mixed poor and survived. The system gave every group someone other than the government to dislike.

The Soviet Union tried yet another variation. The policy of korenizatsiya (indigenization) in the 1920s created territorial homelands for every officially recognized ethnic group, each with its own constitution, territory, and cultural institutions. Terry Martin has a whole book about this called “Affirmative Action Empire” that I have not read but want to. The thesis is (aka the blurb says) that the Soviet government “was the first of the old European multiethnic states to confront the rising tide of nationalism and respond by systematically promoting the national consciousness of its ethnic minorities.”

When the USSR collapsed in 1991, it for the most part dissolved along the exact national administrative lines the Bolsheviks had drawn… not the historic borders of the 1800s.

The Hungarian-appointed governor of Croatia in the 1880s and 1890s did basically the same thing when he granted the Serbian minority within Croatia preferential treatment. He offered them posts, patronage, and a degree of cultural recognition that Croats were denied. Rebecca West talks about this a lot in her doorstopper (and very interesting, Annie Normal and I have been working our way through it for like a year now!) travelogue Black Lamb and Grey Falcon, which I’m given to understand is remarkably accurate given her background as, well, a lady who visited a couple of times and read some books. Croat resentment focused on the Serbs, not the capital, which frustrated West to no end as she was friendly with Croats and Serbs and really wanted the Balkans to do better. Instead, neither group organized against the empire because each was too busy watching what the other one got.

To hear her tell it, it’s an insane miracle that a couple of low-level guys managed to kill Franz Ferdinand.

On the End of Hegemony

When this works, peripheries interact almost exclusively through the core, never directly with each other. The center’s power is protected by maintaining “structural holes” between governed groups. Again, picture a wheel with spokes.

Maintaining or imposing differences seems to work because it is cheap and leverages existing administrative capacity. It throws a spike in the wheel of the most dangerous thing an empire can face: unified resistance. If your A subjects are focused on being mad at your B subjects, neither group is teaming against you. If each province has its own laws and customs and identity, there is no “provincial” identity to rally around. Identity matters!

I’m not entirely sure if this works until it doesn’t (and then the empire tries to switch strategies), or if it works until someone decides to try to consolidate (and then the empire falls because united, the periphery is stronger than the center).

After conquering the Warring States in 221 BCE, Qin Shi Huang imposed comprehensive standardization: unified weights, measures, currency, writing, legal codes, even road widths and axle gauges. The result was the shortest dynasty in imperial Chinese history, lasting only 14 years. The Han ended up moderating Qin uniformity, though the resulting hybrid system brought its own rebellions.

Under Akbar, the Mughal Empire’s mansabdari system organized diverse elites into a unified hierarchy based on rank alone, regardless of origin. Then, Aurangzeb reimposed the jizya tax on non-Muslims and discontinued syncretic court rituals, Jat, Sikh, Maratha, and Rajput rebellions escalated and the empire began its long disintegration. Would they have been better off not unifying? I don’t know. Maybe the whole thing was inevitable.

Pre-revolutionary France was so fragmented that Voltaire joked you could “change legal systems as often as you change horses.” I’m still working my way slowly through Peasants into Frenchmen about the modernization of rural France, but my friends ( Evan Þ and Annie Normal ) swear by it as a fascinating account of how the French did not have anything resembling a unified identity despite how Napoleon came in with the Code Civil, uniform departments, and standardized administration. I get the sense that Napoleon built an “inner empire” where uniformity built on existing traditions, but the “outer empire” never held precisely because they didn’t do the tried-and-true method of imposing a thin rulership layer on diverse local rule. Not sure how true that is, though, almost everything I know about the French comes from the fictional Honor Harrington series modeled on a British admiral.

On Legacy

I almost hesitated to write this because of how fraught this issue is, but I do think it’s genuinely interesting to contrast the different perspectives that people like James C. Scott and Josephine Quinn bring to the question of diversity. On the one hand, there’s the old Aesop “united we stand, divided we fall” and Matthew 12:25 “And Jesus knew their thoughts, and said unto them, Every kingdom divided against itself is brought to desolation; and every city or house divided against itself shall not stand.”

Abraham Lincoln once gave a speech along those lines, too.

But if I get away from the mythmaking and the just-so stories for a moment, I think Quinn is right that diversity can be a tool of control just as easily as it can be a source of protection, and that a house divided itself does, in fact, stand.

Whether that’s good or not depends on whether or not you’re the one trying to maintain the empire and the peace it brings, or the one trying to break it to have more freedom.