Social networks and simulations

In standard introductions to agent-based modeling, agents have attributes, use decision rules, and occupy a space. Although the interactions between agents drive the model, the actual pattern of local interactions (random, grid or torus based or in a circle) are often an afterthought. Recent work on the small-world effect in interaction networks (Watts and Strogatz, 1998; Watts, 1999) led to the introduction of new agent network topologies; typically circles with a few long-distance ties. Do these modifications accurately capture the properties of the social networks of interest?