Cities, Communities & Social Networks: Why Where You Live Is Who You Meet

Two people on the same street can have almost nothing in common, and two people on opposite sides of a continent can be four handshakes apart. Both facts are about the same thing: the shape of social ties, laid over a map.

You see it when you look for a job. The person who tells you about the opening is rarely your best friend, who knows the same people you do. It is a former colleague, a friend of a cousin, someone you last saw at a wedding. You see it when you notice that a neighbourhood has "turned", and nobody remembers deciding that it would. And you see it in the fact that more than half the world now lives in a city, most of them in places no architect drew.

This course asks how people sort themselves in space, how the places they end up in shape them back, and how the web of ties between them behaves. Its tools are old (a map of a city from the 1920s) and new (a network of hundreds of millions of accounts). Its best models are small enough to play with, and they are all below.

The Chicago school: a city as an ecosystem

In the 1910s and 1920s the University of Chicago had a city outside its window that was growing faster than anyone could count: a million immigrants and a great migration of Black families from the American South, all arriving in the same few decades. Robert Park, a former journalist, and Ernest Burgess asked their students to go out and map it. Park borrowed the vocabulary of plant ecology: groups invade an area, they compete for it, and one succeeds another. Their collected work appeared as The City (1925).

Burgess drew the picture that made the school famous: the concentric zone model. A city grows outward from its business centre in rings. Around the centre is a zone in transition of cheap rooms and industry, where the newest, poorest arrivals land. Next come the workers' homes, then better housing, then a commuter ring. As people prosper they move outward, and the next wave takes their place. The detail that mattered most: in Clifford Shaw and Henry McKay's studies (1942) the zone in transition had the highest delinquency rates decade after decade, whichever immigrant group lived there. The rate belonged to the place, not to the people. That is the founding claim of the sociology of neighbourhoods.

Two other ideas from Chicago stayed. Louis Wirth (1938) argued that urban life follows from three things: size, density and social variety. A big, dense, mixed place means many shallow contacts, and Wirth feared it left people anonymous. Herbert Gans's study of an Italian-American neighbourhood in Boston, The Urban Villagers (1962), answered that for many city people the village never went away, and Jane Jacobs (1961) described the "eyes on the street" that make a busy sidewalk safe. Nearly every urban theory since has been a reply to one of these.

Without JavaScript: Burgess drew five rings out from a business centre; Hoyt (1939) argued that cities grow in wedges along transport lines and by income; Harris and Ullman (1945) said cities grow around several centres at once.

The Chicago picture has serious critics. Homer Hoyt (1939) found that rich and poor areas form wedges along roads and rails rather than rings; Chauncy Harris and Edward Ullman (1945) said cities grow around many centres. The bigger problem is method. W. S. Robinson showed in 1950 that a pattern across areas can be quite different from the pattern across people (the ecological fallacy): a neighbourhood with a high crime rate does not mean its typical resident commits crimes. And the Chicago picture was built from one fast-growing American industrial city; W. E. B. Du Bois had already mapped Philadelphia's Black Seventh Ward in The Philadelphia Negro (1899), with far more attention to how racial exclusion, not "natural" competition, shaped where people could live.

Neighbourhoods that watch themselves: collective efficacy

If the place matters, what about the place does? Robert Sampson spent decades answering that in Chicago. In a 1997 study in Science, Sampson, Stephen Raudenbush and Felton Earls surveyed 8,782 residents in 343 neighbourhood clusters and measured collective efficacy: neighbours' mutual trust together with their willingness to step in (to stop kids skipping school, say, or to break up a fight). Neighbourhoods with higher collective efficacy had less violence, even after allowing for poverty and instability. It was a shift from "slums breed crime" to something more precise and less insulting: some poor neighbourhoods have the glue and some do not.

The honest limits: the study is a snapshot, so it cannot fully separate glue causing safety from safety allowing glue; and trust measured by survey is not the same everywhere. What it contributed is a way of thinking: a neighbourhood is more than the sum of its residents' incomes. This idea leads straight into Deviance, Crime & Social Control (35/83).

Cities of the Global South: the settlements that build themselves

"Global South" is a loose, contested label for lower- and middle-income countries, many of them in the southern hemisphere; some writers prefer other terms. Whatever it is called, this is where urban growth is. In 2018 the United Nations projected that the world would go from 55% urban to 68% by 2050, with most of the increase in Asia and Africa. And much of that growth is informal: homes built without permission on land the builder does not formally own, connected to services the builder arranges.

The scale is large. The UN's SDG tracking (indicator 11.1.1) estimated that about 1.12 billion people lived in slums or informal settlements in 2022, about a quarter (24.8%) of the world's urban population, and 130 million more than in 2015. Regional totals vary from one edition of the report to the next, a reminder that these are model-based estimates.

Number	Source	Caution
About 1.12 billion in slums or informal settlements (2022)	UN SDG report 2025, indicator 11.1.1	Definition covers inadequate housing as well as informal land; regional splits change between editions
Kibera, Nairobi: 170,070 counted (2009)	Kenya national census 2009, as usually cited	Popular claims of a million or more were common before it. Recheck the figure before you repeat it
55% urban in 2018; 68% projected for 2050	UN World Urbanization Prospects 2018	A projection; countries define "urban" differently

Three views compete. Janice Perlman's The Myth of Marginality (1976), on Rio's favelas, found residents were workers and organisers fully part of the city's economy, not drop-outs from it. Hernando de Soto's The Mystery of Capital (2000) argued that the poor already own enormous assets, trapped because they hold no legal title; give them title and they gain credit. Studies of Peru's titling programme (Erica Field, 2007) found that titles changed how households used their time, and did less to open bank credit than the theory predicted. A third view, Mike Davis's Planet of Slums (2006), emphasised exploitation, and AbdouMaliq Simone's writing on Johannesburg described "people as infrastructure": where roads and pipes fail, people's relationships do the work of carrying goods, news and credit.

The same self-building appears in organised form. In Karachi, the Orangi Pilot Project (from 1980, led by Akhtar Hameed Khan) helped residents lay their own sewers, at a fraction of the cost of public contractors. And since Jennifer Robinson's Ordinary Cities (2006), many urbanists argue that theories built on Chicago and London should be tested against Lagos, Mexico City and Mumbai too, not exported to them.

Schelling's model: mild preferences, sharp segregation

In 1971 the economist Thomas Schelling (later a Nobel laureate, 2005) published a model small enough to run with coins on a chessboard. Two kinds of people live on a grid with some empty squares. Each person is happy if at least some fraction of their neighbours are like them. An unhappy person moves to a spot where they would be happy. Nobody wants to live in a segregated city. Nobody is asked to. Watch what happens.

Without JavaScript: in this model, if everyone wants at least a third of neighbours like them, the average share of like neighbours rises from about 50% to about 70%; at 60% it goes above 95%; at 75% or more, people get stuck and cannot find anywhere to go.

Here is the lesson, and the limit. Slide the threshold to 35%: people are content to be in the minority, two thirds different from them. Run it. Average like-for-like neighbours moves from about half (the random start) to about 70%. At 60% you get near-perfect separation, in a handful of rounds. The segregation is much stronger than anyone wanted, because each move changes the neighbourhood for everyone around, which triggers the next move. Schelling called this the problem of micromotives and macrobehavior (his 1978 book title).

What it does not show is as important. The model shows that preferences could produce segregation; it does not show that they did. Real cities add everything the grid leaves out: unequal incomes, prices, lenders, landlords and police. In the United States, segregation was also built by law and policy (restrictive covenants, redlining, the siting of public housing), as Richard Rothstein documents in The Color of Law (2017); apartheid South Africa's Group Areas Act (1950) enforced it outright. Survey work by William Clark and others also found that people's stated preferences are a mixture, and that what households say they want does not map neatly onto Schelling's threshold. Treat the model as a warning that "nobody chose this" does not mean "nobody did anything", and not as an explanation of any real city. For a second playable version, see Nicky Case's Parable of the Polygons (below).

Weak ties: why acquaintances find you the job

Mark Granovetter's 1973 paper "The Strength of Weak Ties" is one of the most cited in sociology. His starting point was a survey of professional workers in a Boston suburb for his book Getting a Job (1974). Most who found their job through a personal contact had got the lead from someone they saw only occasionally or rarely. Why? Your close friends know the same people you do and have heard the same news. An acquaintance sits in another circle and hears things you cannot. A weak tie is a bridge: the only short path between two clusters.

The tool below makes the geometry visible. Sixty people form six tight groups of ten (everyone in a group knows everyone else). Person 0, highlighted, wants to hear news from outside their own group. Add random acquaintances between groups and watch how many people lie within three steps.

Without JavaScript: with no weak ties, person 0 can reach only their own group of ten (9 other people). With 2 weak ties the figure is 29 people within three steps; with 16 it is 49; with 24 almost everyone can be reached.

There is a modern test. Karthik Rajkumar and colleagues (2022, Science) ran experiments on LinkedIn's "People You May Know" for over 20 million users, and found that moderately weak ties produced the most job moves, better than both very close contacts and very distant ones. Granovetter's idea held, and became more specific. A caution: the strength of a weak tie depends on culture. Fei Xiaotong's From the Soil (1947) described Chinese rural society not as groups but as ripples spreading from a stone: relationships measured by distance from the self, with obligations shrinking outward. Networks work in every culture; what a tie is and what it owes differs.

Small worlds: few steps, tight clusters

In 1967 Stanley Milgram asked people in the American Midwest and in Boston to forward a folder to a named stockbroker near Boston, through acquaintances only. Of the 296 chains started in the follow-up study (Travers and Milgram, 1969), 64 reached the target, taking on average about 5.2 intermediaries. "Six degrees of separation" took its name from John Guare's 1990 play. Critics, notably Judith Kleinfeld (2002), pointed out that most chains never finished, so the true figure could be larger. Peter Dodds, Roby Muhamad and Duncan Watts re-ran it by email in 2003 across 13 target countries, and again most chains were lost, but the estimates came out at roughly five to seven steps. In 2011 a Facebook study (Backstrom and colleagues) put the average distance at 4.74 across about 721 million accounts.

The puzzle is why. Real networks are clustered: your friends tend to know each other. That alone would make worlds large, with long chains of overlapping circles. Duncan Watts and Steven Strogatz (1998) showed that rewiring only a few links to random distant places collapses the distances while the clusters stay intact. A few weak ties do the work of a thousand. Try it.

Without JavaScript: with no rewiring the average path between two people is about 17 steps; with 1% of links rewired it drops to about 7.5 (44% of the original) while clustering is still about 96% of its starting value; with every link rewired it is about 3.2 but clustering is nearly gone.

The same idea sits behind how disease, rumours and fashions spread, and it is the reason a few well-placed links can help or hurt a whole society. It is also a hint at a limit: a small world says that short paths exist, not that people can find them without a map or that the path carries anything useful. Being four steps from a CEO does not mean you can reach one.

What a graduate just sees

• A place has properties its residents do not: Shaw and McKay's high-delinquency zone stayed high as whole populations came and went.
• A map of areas is not a map of people: before you say "this neighbourhood is X", ask whether it is the ecological fallacy.
• "Nobody wanted this" does not mean no one did anything: Schelling's moves produce more separation than anyone asked for, and real segregation also has laws, banks and landlords in it.
• About a quarter of city-dwellers worldwide live in informal settlements, and the builders of those homes are often the most active people in the city.
• For news and opportunity, your acquaintances matter more than your best friends, and the middle of the range of tie strength may beat both ends.
• A handful of random links turns a big world small, so short paths are cheap and finding them is hard.
• Treat a "huge slum" headline number as an estimate. Kibera's census count was a fraction of the popular figure.

Where it shows up next

• Race, Ethnicity & Migration (35/79): residential segregation, ethnic enclaves and the policy history behind them.
• Stratification & Inequality (35/78): neighbourhood effects on children's outcomes, and how networks carry advantage.
• Deviance, Crime & Social Control (35/83): social disorganisation and collective efficacy are theories of crime as much as of neighbourhoods.
• Public Policy (35/16): zoning, slum upgrading versus titling, transport, and housing vouchers.
• In the world: how a recommendation system finds "people you may know", how a public-health team traces contacts, how a planner decides between upgrading and relocating an informal settlement.

If you want to go deeper

• 📘 Textbook. Richard LeGates and Frederic Stout (eds), The City Reader (Routledge, many editions), is the standard anthology of Park, Burgess, Wirth, Jacobs and later writers. For networks, David Easley and Jon Kleinberg, Networks, Crowds, and Markets (Cambridge University Press, 2010).
• 📘 Gentler. Duncan Watts, Six Degrees (2003), and Jane Jacobs, The Death and Life of Great American Cities (1961). For the Global South, Mike Davis, Planet of Slums (2006), and Jennifer Robinson, Ordinary Cities (2006).
• 🌐 Free play. Nicky Case's Parable of the Polygons is a free interactive Schelling model; the Easley and Kleinberg book is free to download as a draft.
• 🌐 Free course. Matthew Jackson's Social and Economic Networks has been offered free to audit on Coursera; check that it is currently running. I could not confirm a single standard free urban-sociology course.