Reviewed by
Francesca Giardini
University of Groningen, The Netherlands
When I first looked at the title, and I read the foreword I thought that this book was really timely. In 2018, the theme of the Social Simulation conference in Stockholm was “Looking in the mirror”, thus calling for self-reflection as a necessary step for agent-based modelling (ABM) and social simulation on the path towards “proper scientific adulthood” ( http://ssc2018.dsv.su.se/about). In a similar vein, the book by Silverman offers an introduction to the methodological perks of agent-based modelling and some reflections on the pros and cons of using it in the Social Sciences. Through the four parts of the book, the author develops a parallel between biology and artificial life (AL), discussing the multiple ways in which artificial agents can inform the scientific understanding of natural life, and then he extends it to the understanding of societies thanks to social simulation.After a general Introduction to the book content and aims (Chapter 1), the second chapter provides the reader with a basic explanation of simulation science by reconstructing its philosophical and methodological background. Silverman uses AL as an example of the different ways in which description and explanation, the two main goals of using modelling in science, were achieved by different kinds of simulation studies. The development of Artificial Life as a scientific field and the acceptance of artificially generated data by the scientific community is the topic of Chapter 3. By describing the struggles that AL-researchers had to face for making sense of and justifying artificial life, Silverman poses a fundamental question to all of those who use ABM: “Can a simulation create novel datasets which allow us to discover new things about natural systems, or are simulations based on the natural world destined to be mere facsimiles of the systems that inspire them?” (p.39). In Chapter 4 the case of modelling in population biology is used as an introduction to the traditional struggle every modeller knows, the one between realism and tractability. This chapter concludes the first part of the book, which was meant to present and discuss perspectives, debates and results of computational models in Artificial Life.
The second part of the book presents a collection of chapters focusing mostly on modelling in and for the Social Sciences, but unfortunately, this seems to be the least developed part of the book. Chapter 5 presents the work of Cederman and discusses the risks and benefits of ABM for political scientists, but the scope of the chapter is quite limited, as well as the picture of social simulation emerging from it. In the last 25 years many different ABM models have been developed in psychology, sociology, economics and anthropology, and limiting the discussion to the work of a single author in political science does not do justice to the quite ample body of modelling work already existing in the social sciences.
A broader and more accurate picture of the field is put forward in Chapter 6, in which classical views, from Doran’s to Axelrod’s and Tesfatsion’s works, are presented and discussed, together with a summary of the first part of the book and the parallel (again) with AL. Chapter 7 revolves around the famous Schelling’s residential segregation model, which is extensively praised for its simplicity and transparency. Silverman laments a lack of these elements in the majority of agent-based models developed by social scientists, arguing that simplicity makes it possible to understand, replicate and communicate simulations. The last chapter of this section (Chapter 8) offers a summary of the previous chapters. The author revises the arguments used so far in order to discuss the benefits and the limitations of ABM for the social sciences, quoting Schelling’s model again as a positive example, but also stressing the difficulty to proceed from such a specific model of micro-behaviour to what he calls a “deeper social explanation”.
The last part of the book (Chapters 9-12) is devoted to the introduction of model-based demography, starting with an history of the field of modelling in demography and its scientific challenges (Chapter 9), moving to the presentation of some influential models in the field (Chapter 10 and 11), and to their relationships to AL and the Social Sciences. In the last, concluding chapter Silverman summarises the importance of modelling for demography, and he goes back to some of his conclusions about the limits of ABM for understanding human societies.
Going back to the need for self-reflection felt by many in our scientific community, this book’s contribution to that is quite limited. However, it might be an interesting read for scholars and professionals coming from biology and artificial life who are also willing to know more about simulation in the social sciences.
Return to Contents
of this issue