What do we mean by a “network”? Is it a metaphor, concept, cultural construction, heuristic device, model, structural entity, or relationship between people and things defined socially or mathematically? Is the network a revolutionary technology of social organization, or merely a means to enhance what we already do? Or is it all the above? The elasticity and mutability of the term (and of networks themselves) have made them both an attractive tool and a compelling object of interest to many academic disciplines, not only the hard sciences, but psychology, social anthropology, sociology, economics, and, more recently, history. In practice, networks tend to be viewed in several overlapping ways: as a metaphor—just as the image of a web or net was used in the eighteenth and nineteenth centuries to characterize the integuments of human society; descriptively—as a particular account of sets of social relations; analytically—as a means of explaining their strengths and weaknesses; and prescriptively, as a superior means of social organization. As we shall see, it has been a common feature of discussions about networks that they have elided the network as phenomenon and the network as analytic tool, and have been far more concerned with the supposed virtues of networks in both senses than with the problems they pose.
Networks and networking are everywhere. The terms have increased exponentially in everyday speech, in N-diagrams of usage in Google Books, in newspapers and magazines (nearly a fourfold increase in their appearance in the pages of the New York Times), and in academic publications. A major social theorist, Manuel Castells, argues that we have entered a new epoch, that of “network society.” Figures 1 and 2 plot the increased use of the terms “networking” and “social + network” by decade in the Web of Science, the combined citation indexes of the major academic disciplines. The trend, especially in the new millennium, is stark. If in Adam Smith’s commercial society of The Wealth of Nations (1776) he viewed everyone as potentially a trader or merchant engaged in “truck, barter and exchange,” in contemporary information society we are all in some way seen or see ourselves as networkers.
The rise of the network, and particularly of social networks—first identified as such by Elizabeth Bott in the 1950s, as opposed to those systems of trains, ships, telegraphs and cables, radios, and televisions that were described as the key networks of the nineteenth and twentieth centuries—has, of course, been enormously facilitated by *digital information technology. The central thesis of Castells’s work is that information and communications technologies have enabled networks, that have had a long history, to overcome their limitations (chiefly of scale), and thereby to become the dominant form of economic, political, and social organization.
Digital networks began as a series of systems designed to facilitate communication and data sharing within the scientific community but quickly spread into the public sphere. The building of the ARPANET in 1968–69 by the Advanced Research Projects Agency in the United States, the invention of email programs, and, some years later, the development by Tim Berners-Lee and Robert Cailliau at CERN, the European Laboratory for Particle Physics in Geneva, of a system of *hyperlinks and hypertext that underpinned the *World Wide Web, created the platform on which a commercialized *internet was built in the 1990s. The dissemination of personal computers and smart phones, along with a raft of technologies, means that today individuals are constantly and inexorably reminded of their networked being, just as the very technologies that enmesh us also provide us with the capacity to access, analyze, and build networks for ourselves.
Figure 1. Use of “networking.” Web of Science, 1960–2018.
Figure 2. Use of “social + network.” Web of Science, 1960–2018.
It became a largely unquestioned assumption, especially among organization theorists, that networks were good: both a good way to understand the workings of the world, and an effective means of acting within it. This view predated the internet—it became current in the late 1970s and 1980s when experts argued that the best way to deal with the economic ills of stagnant growth and inflation, and with the shifting nature of the economy, away from commodities to information, from mass production to niche markets, was to replace the traditional and inflexible hierarchies of business organizations with informal networked enterprises. These were seen as “flat” (relatively egalitarian), flexible, adaptable to changing tastes and circumstance, and (a favorite term) nimble. Such attributes were a wish list, not an accurate historical account of how all networks functioned. Nevertheless networks were crucially seen as more receptive to information, able to spread it faster and react to it more quickly, to be both nimble and creative—an insight taken up alike by political activists and the new social movements, and by critics of such cumbersome entities as the nation-state. Even before the internet, speed and mutability were of the essence.
At the same time the insights of graph theory and topology (and their popularization by authors like Malcolm Gladwell and in such games as Six Degrees of Kevin Bacon) reinforced the notion that the world was a small place in which everyone is connected, and whose structures—virtual, biological, economic, and social—were self-organized networks that followed certain mathematical rules. Social network sites became the preferred portals through which to connect to this world. Network euphoria, a new species of technophilia fueled by the tech industries, focused, understandably, on the enormous potential unleashed by the availability of information exchange on an unprecedented scale and at astonishing speed. Who would not prefer, it was assumed, to be “information rich” rather than “information poor.” This was, of course, the latest iteration of the myth—found earlier in the promotion of the telegraph and the telephone—that technology is inherently emancipatory, that it would create a more democratic world, expand freedom, dispel ignorance, and enhance human understanding. (The assumption here is that social conflict derives from isolation and insufficient communication, but that technology can annihilate space and time.) This vision also manifested a disturbing tendency to elide the distinction between information and knowledge or wisdom. The fact that the content, context, history, and purpose of networks were every bit as important as their configuration and scale tended to be overlooked. These concerns reemerged in public discourse only as it became apparent that virtual networks suffered from the same ills as their material precursors: the need to maintain their integrity, avoid disintegration, secure trust, avoid malevolent and covert manipulation, and identify misinformation and mendacity.
Absorption in networks and fascination with their contemporary capabilities have transformed the sense of how things worked in the past, revealing networks and relationships that had hitherto been obscured. Different or various sorts of entity—cliques, circles, coteries, cells, cartels, friends, patrons and clients, and correspondents—have all come to be seen as having common features as networks. So the recent sense of their importance has spawned scholarly networks to study historical networks, among them the Harvard University Visualizing Social Networks project; Oxford University’s Cultures of Knowledge: Networking the Republic of Letters 1550–1750; Stanford University’s Mapping the Republic of Letters; Six Degrees of Francis Bacon at the Carnegie Mellon University; and the Dutch universities’ and academy’s consortium the Circulation of Knowledge and Learned Practices in the 17th-Century Dutch Republic. Some of these networks are substantial. The network News Networks in *Early Modern Europe has involved more than thirty scholars from nine different countries. Such historical initiatives are recent. The journal Social Networks was founded in 1979, but the Journal of Historical Network Research only in 2018.
What are the problems that networks can and have addressed? First, networks can help solve problems of information over distances; they can be used to address problems of scale. Thus when the nineteenth-century Parisian naturalist Georges Cuvier needed fossil skeletons for his work on comparative anatomy he used a (preexisting) network of geologists to which he belonged and that included English, Scots, Irish, German, Scandinavian, Swiss, and Italian savants to find him specimens from all over Europe. Second, they can provide collective assurance and insurance. They can offer a sense of belonging or identity, secured by mutually supportive actions, and shared narratives and values (as, for example, in the US civil rights movement of the 1960s). This has often entailed collective obligations to mutual protection, as in the case of political and social movements like the revolutionary European Carbonari or the American women’s suffrage movement of the nineteenth century, and among religious sects and minorities, whether sixteenth-century Protestants or twentieth-century Jews. Third, whatever their purpose—the pursuit and dissemination of learning, the creation or enjoyment of art, the consolidation of kinship or friendship networks, the hunt for profit, the provision of credit, or the acquisition of different sorts of power—networks can reduce uncertainty and risk by pooling sources of information and establishing conventions of acceptable conduct. The information embedded in the dense and interconnected neighborhood, familial, political, and economic networks of merchants and traders in fourteenth-century Florence made lending money less risky and borrowing more readily open to scrutiny. Fourth, in reducing risk and uncertainty, networks can enhance trust—in other members of the network, in the network itself, and in the reliability of information. So, in a way, a network historically has often been a form of information management.
All these benefits and access to them are described in the technical literature as “social capital,” a resource that network members do not own but can use to their advantage. A network is not just a connection; it entails exchange over time: reciprocal, preferential, and mutually supportive action and conversation; in short, friendship (admittedly a slippery concept), a connection that is informal rather than official. Social capital can be accumulated, but it can also be depleted, or even lost. Networks offer not just the promise of inclusion, but also the possibility of exclusion—hence the contemporary fear of not being connected, often seen as something akin to social death. The fear is not new. As the fourteenth-century Tuscan author and merchant Paolo da Certaldo wrote, “a man without a friend is like a body without a soul,” and “a man who loses his friends is worse than dead.” Access to social capital is not distributed evenly. It is easier for the privileged to acquire it, and when they do so, they gain further advantages over others.
Not least because they are not institutions (though they may depend on them), networks are also very much in the business of conserving and sustaining themselves. Communication among members of a network has often included extended discussions about how best to keep the network intact, how to maintain contact and to ensure the circulation of information and goods among its members—by internet, email, encryption, post/mail, private courier, friends and business associates, the diplomatic bag—and how to avoid obstacles to its transmission, such as censors, police, spies and customs officials, marauding warriors (cyber or real), or clumsy and incompetent carriers.
Equally important to networks is the question of how best to conserve, secure, and stabilize information, both when it is moving and when it needs storing. This might seem a much less significant obstacle in a virtual world than in one made up of networks of paper, print, and things. Transferring a series of electronic files from Shanghai to Rio de Janeiro by computer is clearly easier and swifter (though not necessarily safer) than shipping a container full of paper. But the networks of the virtual world, like almost all social networks, have a large and vulnerable material component; using an ineffable term like *“the cloud” brilliantly conceals the phone lines, fiber optics, satellites, thousands of miles of cables, and vast data warehouses (in 2015 data centers used 3 percent of the world’s electricity and produced 2 percent of global emissions, almost the same carbon footprint as the airline industry) that underpin virtual communication. All this requires maintenance. And virtual information is, as yet, more difficult to conserve over time than its traditional precursors.
But network maintenance and conservation have never been merely a matter of technical support. Above all they have involved interpersonal activity to sustain social bonds—acts of friendship and trust, gift giving and kindness, the sharing of knowledge and technique—that help sustain the network. One of the most important ways to extend this, as evident in the twenty-first as in the fourteenth century (though its style, form, and content have changed), has been the letter of introduction and recommendation.
What does the network do for the student of networks rather than for the networkers themselves? It makes visible and tangible what is often concealed. It is a cliché of network analysis that those inside a network, though they can see those who are closest to them, are often unable to envisage the network as a whole, or see its full extent and ramifications. But this emphasis on formal structure conceals the degree to which the network is often a powerful imaginary for networkers, made up of circulated stories, shared values, and activities, in which content matters to them more than form. Networkers can be perfectly well informed about what is necessary to make networks more or less effective without knowing their boundaries. These become important only when a network begins to fragment or fail.
There have been two main approaches to network analysis (though they rarely acknowledge one another): social network theory (SNT), sometimes and more accurately called social network analysis (SNA), and actor network theory (ANT). The former is chiefly concerned with the relationship between social behavior, network structure, and its outcomes; the latter with explaining how, through the use of instruments, tools, and particular work practices, certain scientific facts and insights come to be made, disseminated, and accepted. The two approaches have sometimes been rigorously applied in writing the history of information, but more usually their insights have been used selectively as helpful analytic tools.
Social network analysis seeks to map the ties, the constraining and enabling dimensions and distribution of patterned relationships, and to express them with varying degrees of mathematical precision. A key concept for this analysis is “homophily”—“birds of a feather flock together”—based on the observation that the same kinds of people are drawn to one another and often inhabit the same space: courtiers in a sixteenth-century ducal court, humanists in a university, philosophes in an eighteenth-century Paris *salon, business entrepreneurs in a nineteenth-century international exhibition, scriptwriters in a Hollywood coffee shop, and “friends” on a social media site. The combination of homophily and proximity (which media like the written letter and the internet free from geographical constraints) leads them to influence one another, so they become more and more alike. Such groups, made up of nodes (people or groups), and connected by ties (or edges), make up networks that configure in different ways, and which are often most easily understood through graphic visualization. There are three main types of such networks: ego networks attached to an individual person or node, like the remarkable correspondence networks of European savants such as the jurist Hugo Grotius (7,725 letters), the religious reformer Ignatius Loyola (six thousand letters), the humanist Desiderius Erasmus (three thousand letters), and the Enlightened historian, critic, and philosophe Voltaire (nineteen thousand letters to and from fourteen hundred correspondents); bounded networks within organizations, such as the employees at CERN that Berners-Lee helped connect; and open networks such as the *Republic of Letters whose boundaries are flexible and indeterminate.
For the social network analyst looking at the operations of a network as a whole, its distribution and density are vital—the number and type of connections between nodes, the extent to which they work reciprocally or involve more than one type of connection (say employment, kinship, religion), the distance between those that are furthest apart, the ease with which nodes can link through others—all these affect a network’s capacities and function. A densely connected network aids social cohesion and shared ideas, but within it there may be points (so-called structural holes) where two different parts are joined through one individual person, an essential go-between if the network is to survive. Another position of equal importance is occupied by those with weak ties to the network. They are sometimes bridges between tightly knit network segments or connect one network to another and are therefore more likely to bring new information, knowledge, and insight into the network. For example, the network of geologists that Cuvier relied on for his anatomical fossils included a cluster of mutually acquainted Italian savants scattered through the Italian peninsula, but the key figures in sparking new initiatives and new connections within the Italian network were two figures who connected the Italians to the rest of Europe: a Swiss watchmaker and amateur savant in Geneva, Stefano Moricand, and a young Irish geologist, Joseph Barclay Pentland, working for Cuvier, who shuttled between Paris, England, and Italy.
Even when historical source constraints have prevented full-blown SNT analysis, the method has confirmed the importance of go-betweens and intermediaries bridging boundaries, figures on the margins (in several different senses) who are anything but marginal: interpreters and translators, diplomats and spies, international bankers, and anyone on the cusp of different cultures, like the Cantonese compradors (agents) in nineteenth-century Hong Kong, whose combination of language skills in English and Cantonese, and local expertise in employment, insurance, shipping, and banking, made them indispensable to the British and American trading houses in China. The importance of such figures is one of the few areas shared by SNT and actor network theory.
The more intermediate skills a go-between was able to use, the greater his or her ability to connect different networks. Johann Georg Wille (1715–1808), a German engraver who lived and worked in Paris, linked three overlapping networks: an international community of engravers and artists from all over Europe, more than seventy of whom were his pupils; a wealthy community of art collectors and their dealers, notably German, Russian, and Danish aristocrats, in which Wille was a prominent marchand-amateur (gentleman dealer); and a body of critics and writers, including his friends Denis Diderot, the philosopher Johann Gottfried Herder, and the art critic Johann Joachim Winckelmann. His place at the junction of these networks enabled him to become one of the most important cultural brokers between the French and German *Enlightenments. Wille’s movement among networks was not geographical—after 1739 he never left Paris—but the result of juggling a congeries of identities, a stratagem often associated with the role of intermediary, whose opportunities for reinvention were great. He enjoyed, of course, working in a medium that was both reproductive and visual—not constrained by language. He was at once German and French, an artisan teacher of manual skills, the conveyor of tacit knowledge, and a connoisseur whose taste was expressed in his collections and his criticism (written in both French and German). He promoted French engraving outside France, and Flemish art and German criticism within it. An art dealer and factotum to the European aristocracy, he still retained his independence as a critic. His importance as a sedentary intermediary not just within but between networks was sustained by the constant movement of his pupils, fellow collectors, literary friends, objects, and texts, whose circulation spread and nurtured his reputation. Although he refused frequent invitations to travel, he was indefatigable in encouraging the movement of everyone and thing except himself.
Johann Georg Wille, as the steady but protean center of a constantly moving network of people—pupils, critics, and collectors—and objects—books, journals, paintings, reproductive engravings—makes a suitable object not just for social network analysis but for actor network theory. This theory is best known for two controversial claims: first, that *facts and knowledge are not found but made (constructivism), and that their status therefore depends less on their truth than on the ways in which they secure acceptability within a community or network; and, second, that in networks inanimate and nonhuman objects—natural or artificial—enjoy the same status as persons as “actants” making facts. Rather than adjudicating these claims, it is more constructive to consider how these sorts of assertion contribute to an understanding of networks more generally. (This seems apt given that many proponents of ANT have argued that it is not a theory, but rather a changing “repertoire,” and that this is precisely its strength.) If SNA points to the importance of network structure in understanding how networks work, then ANT argues for examining their processes—whether of creation or discovery—in order to comprehend their role in knowledge production, dissemination, and legitimation, what is described as “the work of ordering.” The view is dynamic and sees travel and circulation—like the movement of Wille’s friends, pupils, and artifacts—as vital to the spread of information and the creation of knowledge, while also grappling with the vexed question of the stability of knowledge that circulates. Hence the interest in “immutable mobiles,” texts, images, and objects whose combination in “centers of calculation” makes possible the creation of scientific knowledge, fabricated from materials from a variety of peripheries. (The view has rightly been criticized as somewhat “imperial” in its perspective.)
Given the rapid advance of *artificial intelligence, and that “bots” now send us birthday greetings, and tell us what we need (after a recent accident I was advised of the need for burial insurance), it is not hard in the world of AI to see nonhumans as actants. But networks have invariably depended on a complex relationship between technologies, material culture, and human agency. The technical and material environment, broadly conceived, can at the very least be seen as a necessary condition or as an affordance of certain sorts of network: writing, paper, and posts for correspondence networks; printing and graphic media for a public sphere; precious metals, coin, and paper money, or certain sorts of commodity for mercantile and business networks. More specifically, materials like Wille’s designs and engravings, which circulated with his former pupils through Europe, helped constitute networks of knowledge and taste, shaping, for instance, the designs to be found on Meissen porcelain. Examining these interrelations (while avoiding technological determinism) is essential in understanding the historically specific nature of networks—their operations, powers, and constraints.
Clearly networks are a powerful mechanism of social action and for the dissemination of information, and an illuminating way to analyze them. The proponents of networks (those who take a prescriptive view) lavish on them an abundance of positive qualities: efficiency, flatness, speed, adaptiveness, and the revelation that knowledge often depends less on individual genius than on flexible cooperation or vigorous rivalry. Questions of cause and effect, of how networks come into being rather than what they do when they exist, tend to take a back seat; too often networks just “are.” And network thinking has rightly been criticized as too insular, formal, and abstract, paying insufficient attention to issues of difference or reducing them to a question of access to or exclusion from the network. Networks may appear (or be represented as) flat, but as only the most superficial glance at Wille’s networks demonstrates, actors are not equal; they are teachers and pupils, aristocrats and commoners, savants and artisans, all mindful of their standing. Networks are not a class-free zone, nor are they just made up of individuals. And it is not clear why or where networks begin or end. Yet the boundaries we draw—who is in, who is out—significantly affect how we understand such processes as knowledge creation. Writing a history of botanical knowledge, do we include the findings of Indigenous collectors and informants, or confine our network to better-known (to us) European collectors? The shape and extent of the network affects the story that we tell. Or, perhaps more accurately, the story we tell affects the network we build. In short networks and network analysis are powerful tools but need to be exercised with caution, a prudence best exercised by treating them less as ideals and more by attending to their specific histories.
John Brewer
See also computers; digitization; globalization; knowledge; letters; money; newspapers; public sphere; social media; travel