Skip to Content

Workshop Two, Session Two

Complex-Systems Agent-based Modelling

Presenter: Hume Winzar

Abstract

A landmark paper in Marketing Letters by Goldenberg, Libai and Muller (2001) provides evidence for the validity of Granovetter's (1973) theory of the strength of weak ties.  This theory posits that occasional, seemingly unimportant, contacts frequently make information available to otherwise isolated groups and are thus necessary for the dissemination of knowledge.  The greater value in the paper is that it introduces business scholars to the application of Complexity Theory and the tools of Agent-Based Modelling (Holbrook, 2003).  This research calls into question some of the findings and conclusions of the original Goldenberg et al. paper.  Subsequent analysis suggests a power law relationship between weak ties and strong ties in terms of their relative strengths and relative numbers. We find that the strength of weak ties is unimportant and that the mere number weak ties is the important issue in the diffusion process.

Background: What are Agent-Based Models?

Agent-based models are generally computer-based simulations of complex systems.  Typically independent "agents" operate concurrently and interact with each other in space and time.  This makes it possible to explore the connection between the micro-level behaviour of individuals and the macro-level patterns that emerge from the interaction of many individuals (Wilensky, 1999).  Where economists and financial analysts in the past have regarded a marketplace as a single object to be analysed with a mathematical equation, Complexity Theory recognises that a marketplace is simply the aggregation of individual behaviours.  Cellular automata are the tool for which we can model or simulate behaviour at the level of the individual and observe their aggregation.