Access this issue
Kent State University
The first thing that stands out when reading these papers is their variety: (1) Ormerod and Wiltshire use a series of simulated social networks to explore binge drinking in the UK; (2) Campennì, Andrighetto, Cecconi and Conte develop the study of norm compliance by increasing the cognitive sophistication of the agents in their model; (3) Abdou and Gilbert construct an artificial (network-based) community to explore the recursive relationship between workplace segregation and social segregation; (4) Chattoe-Brown creates a simulated society - filled with partial communication, hegemonic groups, sub-optimal payoffs, ignored minority viewpoints and gossip - that challenges the rational choice model and its idea of complete information; (5) Kramer, Costello and Griffith create an artificial world of hostiles, friendly citizens and soldiers to explore citizen event reporting; (6) Mollona and Marcozzi's model explores how a weak (decentralized) management style allows skilled employees to better control, change and coordinate their team-work to better respond to the complex challenges of today's knowledge economy; and (7) Yildirim and Yolum's hybrid model advances the study of cooperation (made famous by Axelrod) by demonstrating the important role that reciprocity plays in symbiotic groups, especially for the elimination of defectors.
While these studies differ dramatically in substantive focus, three themes stand out. The first is that agent-based modeling has gotten very sophisticated. One sees this sophistication, for example, in the tremendous facility of such models as Mollona and Marcozzi's or Chattoe-Brown. These models do a lot and they do it well. Other examples include the increasing "real world" nature of the models: even when simple, the rules used in these models reflect the complexities of social interaction, as in the case of Ormerod and Wiltshire or Kramer, Costello and Griffith. The number and types of factors modeled is also impressive, as in the case of Yildirim and Yolum. Equally important is the cognitive complexity of some of the models. Modeling agent cognition is a difficult and controversial issue - how much is necessary and when do things become unnecessarily complex? Campennì, Andrighetto, Cecconi and Conte, for example, tackle this issue well. Another issue is using simulation to advance theory, as in the case of Abdou and Gilbert. Finally, there is the manner in which the articles are written. It is clear that a style for writing simulation papers has become codified, as demonstrated in the papers by Yildirim and Yolum and Abdou and Gilbert.
The second theme that stands out is the integration of simulation with real-world empirical inquiry - an important development that Gilbert and Troitzsch (2005), for example, have been pushing. As Epstein points out in his address Why Model? (2008), simulation serves a variety of purposes, many of which are to help us better understand, bracket, explore and collect real-world data. It is very encouraging, therefore, to see top scholars in the field demonstrating how to connect empirical data and simulation.
The final theme that stands out is the intersection of agent-based modeling and network analysis. While both are part of the methodological backbone of complexity science, much work remains to be done connecting the two, particularly on the side of network researchers. Four of the papers in this edition use or discuss the relevancy of network analysis to their research -(1) Ormerod and Wiltshire, (2) Abdou and Gilbert, (3) Mollona and Macozzi, and (4) Yildirim and Yolum. There is much to learn from reading these papers.
As with most edited works, some papers shine brighter than others (but not by much); and a few authors overstate the importance of their model. Also, while most of the writing was very good, in some places it is very poor, making it difficult to understand the important arguments authors are making. But, these are minor criticisms.
Overall, given the rigor of the review process, and the quality and diversity of the contributing authors, this collection of papers stands as a strong statement about the current place of simulation - both as a method and as a field of study within the social sciences. Flaminio Squazzoni has put together an excellent collection. I highly recommend this special edition.
GILBERT, N and Troitzsch, KG (2005) Simulation for the Social Scientist. 2nd Edition. Open University Press: London, UK
Return to Contents of this issue
© Copyright Journal of Artificial Societies and Social Simulation, 2010