Browse Books

Reviewer: H. Van Dyke Parunak

Historically, the social sciences have been fragmented in two ways. First, as the plural in the name suggests, the study of human behavior embraces many different foci, including (among others) economics, political science, communication, anthropology, and management. Second, in recent decades, the social sciences have drawn on an increasing array of mathematical and conceptual tools, including graph theory, game theory, and nonlinear dynamics. This breadth of material poses a formidable challenge to new students and their professors. On one hand, a curriculum composed of specialized courses in each contributing discipline risks atomizing the student's view of the field, pitting (for example) social network analysts against game theorists, and defers engagement with the underlying social questions that motivate many students in the first place. On the other hand, introductory survey courses too often present the foundational disciplines at an informal level, leaving the student with a sense that the field lacks rigor and encouraging a "cookbook" approach that applies the tools in a rote manner without understanding their inner logic. This text offers an integrated, but not superficial, introduction to these new mathematical concepts and their application across a range of social problems. Each section provides rigorous proofs of key results and rich references to the literature, while remaining accessible to the undergraduate with only a high school mathematics background. References among the sections show how the concepts from different underlying sources contribute to a truly interdisciplinary approach to important social questions. As someone with a background in the conventional disciplines, I found the first few sections confusing. Is this a mathematics book__?__ A book about social networks__?__ A book about economics__?__ As each section built on those before, I realized that the authors are challenging the isolated status of those conventional disciplines. The shared use of new mathematical tools in a number of social disciplines is drawing those disciplines together into a new synthesis, which this book seeks to capture. After an introductory chapter, the book is organized into seven parts, each a mini-text on a specific tool or application area. Part 1 introduces basic concepts in graph theory and social networks. The notion of networks is a unifying thread throughout the book (though it would be overly constraining to describe the book as a "network book" along the lines of recent offerings by Newman [1] and Jackson [2]). Networks are interesting not only for their structure, but for the processes that they support. Part 2 provides an introduction to game theory as a formal system that can usefully describe interactions among parties whose relations are captured in a graph. Part 3 introduces markets, casting them in terms of the graph- and game-theoretic concepts in the first two sections. The result is a novel perspective that will yield new insights even for those with a classical economics background. Part 4 explores the World Wide Web as an example of an information network. After developing formal models of search, it brings the previous parts to bear in a discussion of sponsored search. Part 5 discusses network dynamics at the population level, using models that presume decision-makers have a fairly global picture of the network's structure, while Part 6 focuses on dynamics that result from local observations and actions by network nodes. Both sections consider both dynamics on networks and dynamical changes in the structure of the network itself. Part 7 draws on the rich array of themes in the previous six sections to analyze three kinds of systems encountered in the real world: commercial markets (more detailed and realistic than the idealized mechanisms in Part 3), voting, and property rights. The book is an ideal text for introductory classes. It also holds great promise for people with a strong background in another field who wish to understand some of the key questions addressed by the social sciences. Each chapter but the first has exercises, and many chapters have optional material that provides additional details on concepts introduced in the chapter. The abundance and accessibility of proofs of central concepts will be particularly helpful in helping students to understand the inner logic and limitations of standard results. A bibliography of over 400 entries extends through 2009, and includes pointers to many classic papers, as well as newer work. Online Computing Reviews Service

Reviewer: Fernando Berzal

As complex systems made of individual nodes connected by links, networks are behind many natural and artificial phenomena. So, it comes as no surprise that scientists from different disciplines try to understand how connected systems work. This book is a recent addition to the growing collection of recent textbooks that deal with networks [1,2,3], with substantial overlap among them. While these others try to understand networks by mainly focusing on the mathematical [1], algorithmic [2], or game-theoretic [3] aspects of existing models, Easley and Kleinberg survey the field with a strong emphasis on connecting models to real-world situations. This is partially due to the book's origin: an undergraduate course at Cornell University. Hence, mathematical details have been stripped away, whenever possible, and algorithmic considerations are almost missing. It is, however, an excellent introductory textbook. The 24 chapters are organized into seven blocks that correspond to its central themes. Most chapters include optional sections at the end, with advanced material that would not typically be taught in a standard undergraduate one-semester course. They also include lists of exercises, most of which are quite mechanical, intended to acquaint students with the material covered in each chapter (generally, you can find more demanding problems and open questions in Newman's book [1]). After the usual introductory overview chapter, the first block of chapters starts the study of network structure by introducing graph theory and some key topics related to social networks. An easy-to-understand nontechnical introduction to graph theory is followed by a nice treatment of strong and weak ties in social networks, key structural metrics (such as clustering coefficient and betweenness), external factors that affect link formation (such as homophily and affiliation), and the stability of positive and negative relationships (structural balance). The second part of the book introduces game theory as a key tool to model interdependence in the behavior of individuals. Like the rest of the book, it is written clearly and provides many intuitive examples that highlight the key issues discussed in each section. Modeling traffic through a network and auctions are discussed as two interesting applications of game theory that sometimes lead to counterintuitive conclusions (for example, Braess' paradox, which states that adding capacity to a network can sometimes slow down the traffic). A third block of chapters combines the material from the previous two, from an economic point of view. A preliminary chapter on matching markets sets the stage; without taking network structure into account, it focuses on optimal assignments and market-clearing prices. Later, trading networks-markets with intermediaries-and network exchange theory receive their deserved share of attention, as they model the natural convergence of network structure (graphs) and strategic interaction (games). The fourth part of the book changes its application domain. This time, the focus of study is information networks and the World Wide Web (WWW). In less than 100 pages, readers learn about the bow-tie structure of the Web, the link analysis algorithms used in Web searches (such as Kleinberg's HITS algorithms and Google's PageRank), and the sponsored search markets at the heart of keyword-based advertising. The fifth and sixth blocks of chapters turn their attention to network dynamics. They study situations in which a person's behavior and decisions depend on other individuals' choices. First, they are studied for their aggregate effects, such as information cascades and the rich-get-richer phenomena. For instance, readers will learn that the so-called "wisdom of crowds" does not always produce accurate results, due to network effects. The second set of chapters on network dynamics analyzes how individuals are influenced by their neighbors. Taking a closer look at the fine structure of the network, readers can study social contagion as a rational decision-making process (with the help of threshold models) and epidemics as a probabilistic random process (with the help of standard SIR/SIS models). The small-world phenomenon is also studied in its own chapter, due to its importance for finding short paths and performing decentralized searches. The final set of chapters deals with key societal institutions. They are not typical network topics, but they are often taught in multiagent systems courses [4]. However, their importance is clear when you take into account how their design can produce different forms of aggregate behavior in connected systems. In particular, the authors discuss the ways in which markets serve to aggregate individual beliefs (information asymmetry in markets), how to reach a single decision (voting mechanisms), and how to allocate resources (property rights). This unusual range of topics is what makes this book invaluable. Instead of just focusing on abstract mathematical models and their formal properties, it puts models in their proper place within a process that begins with empirical observations, leads to mathematical models, is followed by some predictions, and is then subject to experimental validation that starts the cycle anew. In the meantime, one reasons about the connected world around us, discovers some facts, gets some insight into the behavior of complex systems, and even enjoys some "aha!" moments. The book is a pleasure to read. Online Computing Reviews Service