INTRODUCTION

During the past decades considerable progress has been made in developing research methods that are particularly suited to examine network relations in the Internet and their consequences for outcomes such as the diffusion of applications and services, the winner-take-all dynamics of digital markets, and the spread of malware.

The Internet profoundly affects economic and social interactions while enabling novel forms of coordination. Weaving individuals, organizations and devices into new, wider and denser networks of interaction, it also creates new interdependencies. For example, contacts on social media platforms and Web 2.0 services such as Facebook, Twitter, YouTube, Snapchat or Digg likely affect individuals’ perceptions and preferences. Thus, in an economy with dense connectivity, the preferences of agents may evolve in response to other agents’ choices and revealed preferences. Such endogenous preference formation may require a rethinking of the notion of rational, independent decision-makers that continues to be at the heart of much economic analysis (Gintis, 2009). Another example is the complex value network of the Internet economy, in which complementarity and platform relations between firms abound. These developments change the nature of com­petition in ambiguous ways, often requiring companies to compete and cooperate with their rivals (Brandenburger and Nalebuff, 1996; Christensen, 1997; Farrell and Weiser, 2003; Anderson, 2009).

As communication flows between individuals and devices are increasingly computer- mediated, the Internet allows the collection of highly granular and detailed data on information flows between agents, networked relations, agent behavior, and on Internet- mediated economic transactions. This is a significant departure from the past, when such information was difficult and expensive to collect in a reliable way (Newman et al., 2006, p. 5). Consequently, new and innovative forms of inquiry have evolved that take advantage of this massive empirical base. Given the early state of the field’s development, different trajectories of investigation can be distinguished, although they may eventually

72

become more integrated. Analytical network science often relies on game theory to model the behavior of individual nodes and combines it with insights from graph theory. New agent-based and other computational approaches allow applying game theory to situations for which no analytical solution can be devised (Tesfatsion, 2003; Judd and Tesfatsion, 2005). Much of empirical network science uses computational techniques that are spreading rapidly in the social sciences (Lazer et al., 2009). There is enthusiasm about the opportunities offered by ‘big data’ analytics (Mayer-Schonberger and Cukier, 2013; Siegel, 2013; Pentland, 2014), although the early experience has revealed the need to develop careful theoretical and empirical models to take advantage of the abundant information (Monge and Contractor, 2003; Easley and Kleinberg, 2010; boyd and Crawford, 2012; Doornik and Hendry, 2015).

These developments have led to an invigoration of research applying network theory perspectives to better understand the Internet and other networks. This chapter briefly discusses these approaches with a focus on network science and its relevance for studying the Internet and its economic effects. It discusses the uses and limits of network theory models and whether they offer new answers to the questions and challenges raised by the Internet economy. The following section describes the emergence of network science and discusses its precursors in mathematical graph theory and the social sciences. Section 4.3 develops basic concepts of network models and section 4.4 applies them to explain the Internet’s global structures. Section 4.5 discusses several selected areas in which network science enriches the analysis of Internet economics, including the use of network models to develop a better understanding of dynamic processes in the Internet. Section 4.6 sum­marizes key points.

4.2