Start with the counterargument, because it's the obvious one and it has to be answered rather than skipped. The Homestead Act of 1862 opened essentially the entire unsettled Great Plains to any settler on the same terms — Kansas, Nebraska, Montana, and both Dakotas were all comparably empty and comparably available at the same moment. If cheap open land were the whole explanation for where an immigrant group ended up, Norwegian settlement should have spread roughly evenly across all of it. It didn't. By 1910, Norwegians made up 21.3 percent of North Dakota's population; by 2009, 30.8 percent of the state claimed Norwegian ancestry, still the highest share of any group in any U.S. state.^[1] Minnesota, the state that actually carries the Scandinavian stereotype in the national imagination, sat at 16.5 percent Norwegian ancestry that same year, and its own Norwegian share has been falling since at least 1980, when it was 19.1 percent.^[2] Empty land was available in a dozen states at once. Only one of them filled up with Norwegians at nearly double the rate of the state next door that everyone associates with them.

Land explains the opportunity. It cannot explain the unevenness — and the unevenness is the actual finding. Once one Norwegian family homesteaded successfully, letters home named the specific county, not the region; later arrivals followed a name, not a map. That is a network effect operating on top of a land-availability fact, and the two get told as one story so often that the network half disappears entirely.

The same mechanism runs through Washington, DC, on a completely different population and a completely different century. The Ethiopian–American diplomatic relationship is more than a hundred years old, and Howard University's standing as a leading historically Black university made DC a natural draw for Ethiopian students well before any crisis forced anyone to leave home.^[3] That gave the city a real foothold — diplomats, students, professionals — years before the 1974 revolution that deposed Haile Selassie and installed the Derg military government. When that revolution, and the repression and famine that followed it, drove a much larger wave of Ethiopians out of the country, DC is where a large share of them went, because the foothold from the diplomatic era was already there to go to.^[3] The city's Ethiopian population grew from a small community of students and diplomats to what is now the largest Ethiopian community in the United States, concentrated for decades around 9th and U Streets NW, an area the DC Council formally recognized as Little Ethiopia in 2020.^[3] Diplomacy built the first foothold. Revolution created the refugees. The concentration in one specific city is neither of those forces alone — it's what a network does once a foothold exists for a crisis to land on.

Minnesota's version of the same mechanism runs through policy rather than diplomacy, and it is recent enough to trace almost year by year. Minnesota had no meaningful Somali population before the country's civil war began in 1991. The state started accepting Somali refugees directly from African refugee camps in 1993, largely people from Mogadishu, placed there by federal resettlement policy and agencies like Lutheran Social Service and the International Institute of Minnesota — not because Somalis chose Minnesota, but because that is where the placement system sent the first arrivals.^[4] What happened next was not policy. Minnesota's Department of Human Services recorded 13,582 additional Somali refugees arriving in the state between 2005 and 2018 alone, well after the initial placements — later arrivals choosing the same state their community was already in.^[4] Minnesota now holds roughly 108,000 Somali residents, the largest Somali population of any state.^[4] The same pattern repeats inside the same state for a second population entirely: Ethiopian refugees, most of them ethnically Oromo, first arrived in Minnesota in 1981, founding the Oromo Relief Association of North America in the Twin Cities that same year. Minnesota is now home to roughly 30,000 Ethiopians, most of them Oromo, and after Kenya itself, Minnesota holds the largest Oromo population anywhere outside Ethiopia.^[5] A federal placement decision put the first few dozen people in Minneapolis and St. Paul. Everyone who has arrived since chose to go to the place their own people already were.

Lay the three cases next to each other and the mechanism is the only thing they share. Norwegian settlement in North Dakota started with land and a letter home. Ethiopian settlement in DC started with an embassy and a university. Somali and Oromo settlement in Minnesota started with a caseworker's placement list. Nothing about a farm, a diplomat, and a refugee resettlement agency should produce the same shape of outcome — three unrelated starting mechanisms, in three different centuries, all converging on the identical downstream pattern: whatever got the first handful of people somewhere, the concentration that followed was never that original cause repeating itself. It was the first arrivals becoming the reason the next ones came.

The largest-scale version of this mechanism in American history didn't need a diplomat, a caseworker, or a homestead claim at all — it ran on nothing but the network itself. Roughly six million Black Americans left the rural South for Northern, Midwestern, and Western cities between 1910 and 1970, and where any single family landed was rarely a guess: a relative, a church, a hometown network already in Chicago or Detroit or Harlem was usually the reason that city and not another one. This site has already traced how that exact mechanism built Newark, Detroit, and Harlem as one connected story rather than three separate local ones — the same network logic as North Dakota's Norwegians and Minnesota's Somalis, running here at a scale of millions rather than thousands, and needing no government agency or foreign embassy to get started. If land, diplomacy, and policy can each independently trigger the identical downstream pattern, and pure kinship can trigger it with no external cause at all, the network isn't a side effect of any of those origins. It's the actual mechanism, and everything else on this list is just how it got switched on.

And that is exactly why integration is hard, not just why concentration happens. A law can remove the barrier that kept a neighborhood or a school segregated. It cannot remove the network that made living near your own people the default in the first place, and the network doesn't stop pulling the moment the barrier does. That is the honest reason Brown v. Board and the Fair Housing Act ended the legal enforcement of separation and still didn't produce integration on their own — removing a rule and reversing a five-word instinct are two different jobs, and only one of them was ever done by statute. The mechanism that built North Dakota's Norwegians, DC's Ethiopians, Minnesota's Somalis, and Newark's and Detroit's Great Migration communities is the identical mechanism a country runs into every time it tries to integrate any of them, because it was never a story about walls coming down. It was always a story about which people other people wanted to live near, and no law ever fully reaches that.

The pull doesn't need you to stay in place to keep charging you, either. Someone who leaves the network for an unrelated reason -- a job, not a rejection of the people they left -- still carries it on the trip back. A 2019 Harvard Business Review study by Courtney McCluney and colleagues on the psychological cost of code-switching found that the tension between suppressing a home identity and maintaining professional standing persists regardless of how diverse the destination workplace actually is -- evidence the cost is systemic, not a function of any one office's makeup.^[6] This site has already traced what that costs at work, where the same research found more education increases the switching burden rather than reducing it. What it costs on the trip home runs the identical mechanism in the other direction: moving away for a reason that has nothing to do with the network doesn't loosen its hold. It just adds a second register on top of the first, and holding both, for decades rather than the length of one job, is the actual price of having left.