Social network

• 1. Introduction

  ›

  Social networks are invisible architectures that shape nearly everything about how human beings find work, share ideas, and change one another's minds. At the heart of this field sits a deceptively simple question: what matters more, who you are or who you know? The answer, researchers discovered over more than a century of inquiry, keeps pointing toward the connections themselves. How do rumors spread through a population? Why do some artists get remembered and others are forgotten? How does violence ripple outward from a single act between two gangs? These are not separate puzzles. They are different faces of the same underlying structure. The field that emerged to answer them began not in computer science or Silicon Valley but in the notebooks of social psychologists in the 1930s, in the ethnographic fieldwork of British anthropologists in southern Africa, and in the mathematical abstractions of graph theorists. What drew all these traditions together was the conviction that social phenomena could not be explained by studying individuals alone. The unit of analysis had to be the relationship between them.

• 2. Pioneers and Foundations

  ›

  Georg Simmel, writing at the turn of the twentieth century, was among the first to treat the form of social connection as an object worthy of study in itself. Simmel pointed to the nature of networks and examined how network size shaped interaction, paying particular attention to what happened in loosely knit groupings rather than tightly bound groups. Around the same time, Emile Durkheim and Ferdinand Tonnies were reaching toward similar ideas from different directions. Tonnies drew a distinction between two kinds of social tie: Gemeinschaft, the German word often translated as community, describing ties built on shared values and direct personal bonds; and Gesellschaft, the German word for society, describing more formal and instrumental connections. Durkheim, for his part, argued that social facts arise from interaction between people and cannot be reduced to the properties of any single individual. These were the conceptual seeds from which the field grew.

  Jacob L. Moreno is credited with a more concrete breakthrough. In the 1930s, Moreno began systematically recording and analyzing social interaction in small groups, especially classrooms and work groups. He developed what he called sociograms: visual diagrams mapping who related to whom. These were the first tools designed specifically to study interpersonal structure, and they gave researchers a way to see patterns that had previously been invisible. Simultaneously, in anthropology, a group associated with Max Gluckman and the Manchester School, including John A. Barnes, J. Clyde Mitchell, and Elizabeth Bott Spillius, carried out some of the first fieldwork using network analysis in communities in southern Africa, India, and the United Kingdom. British anthropologist S. F. Nadel codified a theory of social structure that proved influential in later network analysis.

  By the 1950s, the mathematical foundations were being formalized. By the 1980s, theories and methods of social networks had become pervasive across the social and behavioral sciences. Sociologist Harrison White and his students at the Harvard University Department of Social Relations became a particularly generative cluster. Stanley Milgram was also independently active in the Harvard Social Relations department during this period, and it was Milgram who developed the "six degrees of separation" thesis. Mark Granovetter and Barry Wellman were among White's former students who went on to champion network analysis. Granovetter became known especially for his work on what he called "the strength of weak ties," a phrase that described why loose acquaintances often deliver more useful information than close friends.

• 3. Levels of Analysis

  ›

  At the smallest scale of analysis, social network researchers begin with a single individual and trace outward. This micro-level work considers the dyad first: a social relationship between two people, examined for its structure, its strength, and the degree to which each party reciprocates the other's investment. Add one person to a dyad and you have a triad, which Fritz Heider placed at the center of his balance theory of social dynamics. Heider argued that certain configurations of three people are stable and others are not. A rivalrous love triangle is a classic example of an unbalanced triad: the discord tends to resolve itself by changing one of the three relationships. The dynamics of social friendship in groups have since been modeled using this balancing principle, formalized through the theory of signed graphs.

  Meso-level analysis sits between the individual and the whole society. Formal organizations are one key subject here: social groups that distribute tasks toward a collective goal. Research at this level examines both the internal structure of an organization and the ties between separate organizations. Exponential random graph models became state-of-the-art tools for this kind of analysis in the 1980s. These models can represent reciprocity, transitivity, and homophily, which is the tendency of people to connect with others who resemble them.

  At the macro level, researchers step back further still. Rather than tracing relationships between specific individuals, macro-level analysis looks at aggregate outcomes across large populations, including the transfer of economic resources or the diffusion of ideas. One concept that links meso and macro levels is the scale-free network, a type of network whose degree distribution follows a power law. In a scale-free network, a small number of highly connected hubs hold outsized influence. Albert-Laszlo Barabasi developed a model of network evolution that provides a well-known illustration of this structure.

• 4. Structural Holes and Social Capital

  ›

  Ronald Burt's study of 673 managers at one of America's largest electronics companies, completed in 2004, produced findings that gave the abstract concept of structural holes a concrete and practical face. Burt found that managers who regularly discussed issues with people in other groups were better paid, received more positive job evaluations, and were more likely to be promoted. What made their positions valuable was not simply who they knew but where they sat in the network: at the junction between clusters that would otherwise have had no contact with each other.

  A structural hole, in the vocabulary developed from this line of research, is the gap between two clusters of people who hold different and non-overlapping information. Information within any single cluster tends toward homogeneity because members of a tight group share experiences and see the world through similar lenses. The person who bridges two such clusters can access insights from both sides, then act as a broker by passing information in either direction. British philosopher John Stuart Mill put a version of this intuition into words long before the formal theory existed. Mill wrote that it is hardly possible to overrate the value of placing human beings in contact with persons dissimilar to themselves, calling such communication one of the primary sources of progress.

  Mark Granovetter's theory of weak ties runs parallel to the structural holes concept. Granovetter's key observation was that weak ties, the loose connections to acquaintances rather than close friends, are paradoxically powerful for finding information and innovation. Close friends tend to know the same things you already know. The consulting firm Eden McCallum offers a real-world illustration of how bridging structural holes can advance careers: its founders built their practice by connecting former big-three consulting firm professionals with mid-size industry firms, each side possessing knowledge the other lacked. Not every organizational culture rewards this kind of brokerage, however. A study of high-technology Chinese firms by Zhixing Xiao found that the control benefits of structural holes were dissonant with those firms' dominant spirit of cooperation and communal sharing values, suggesting that the payoff from bridging gaps depends on the surrounding social context.

• 5. Networks in Practice

  ›

  Nicholas Christakis and his collaborators ran field experiments in settings as varied as villages in Honduras and urban slums in India, finding that cascades of desirable behaviors could be deliberately induced in social groups. The same experimental methods have documented the social contagion of voting behavior, emotions, risk perception, and the uptake of commercial products. These findings moved social network analysis from an observational science toward an interventional one.

  In criminology, researchers have mapped the network structure of gang violence. Murders, in this framework, are not isolated events but exchanges between criminal actors: violence diffuses outward from a source because weaker gangs cannot afford to retaliate against stronger ones directly, and instead commit other acts to maintain their reputations for strength. In public health, network analysis has moved into epidemiology, models of patient communication, mental health diagnosis, and the study of health care organizations. Respondent-driven sampling, a technique developed from network principles, allows researchers to estimate and reach populations that are otherwise hard to count, including homeless people and intravenous drug users: survey respondents recommend further respondents, tracing connections through a hidden population.

  In the arts, network analysis has revealed that the company an artist keeps in museum exhibitions affects how that artist is remembered in historical narratives, even after controlling for individual accomplishment. Auction performance follows similar patterns. Research has also mapped literary systems by integrating polysystem theory, which dates back to the writings of Even-Zohar, with network theory, allowing scholars to visualize relationships among writers, critics, publishers, and literary histories. Social network analysis has even reached into the study of segregation: because homophily, the tendency to form ties with similar people, shapes where people live and who they connect with, network data can be used to measure the degree to which different groups are exposed to one another within a given area.

• 6. The Interdisciplinary Field Today

  ›

  Social network analysis draws on social psychology, sociology, statistics, and graph theory simultaneously, and that combination gives it unusual reach. Beginning in the late 1990s, a new wave of researchers joined the conversation, including physicists such as Duncan J. Watts and Albert-Laszlo Barabasi, political scientists, and others who brought new models and methods to the growing body of data generated by online social networks and by what researchers call the digital traces of face-to-face interaction.

  The theoretical toolkit that practitioners bring to this field includes both imported frameworks and a few theories that originated within social network analysis itself. Graph theory and balance theory were imported from other disciplines. Heterophily theory and structural role theory were generated from within the field. The core axiom that anchors all of them is relational: social phenomena should be understood through the properties of the ties between units, not through the properties of the units themselves. One persistent criticism of this approach is that individual agency can get lost in a purely structural account, though proponents note that agent-based modeling offers tools for bringing individual decision-making back into the picture.

  Social capital is the concept that has perhaps done the most to carry these ideas beyond academic sociology. Social capital refers to the value that flows from social ties: the job opportunities that newly arrived immigrants can access through connections to established migrants, the creative ideas that flow across structural holes in an organization, or the trust that enables cooperation in communities. A positive relationship exists between social capital and the intensity of social network use, and in a dynamic framework, higher activity in a network builds higher social capital, which in turn encourages more activity still. The field that Jacob Moreno helped launch with hand-drawn sociograms in the 1930s now forms part of what researchers call network science, a nascent discipline that studies social, biological, and technological networks as instances of the same underlying mathematical structures.