What you see depends on the scale at which you look. For example, you can mark the place that you live on a map of the world or on a street directory. Both reflect reality but each gives very different information. Like maps, models for social insect communities can be created on different scales. For example, good mathematical models for how ants lay and follow trains exist at the equivalent of street directory and city-scale maps. (These are simulation-type models, neural-net type models and ODE models with a pre-determined trail network.) We are aiming to construct models on a national- and world-map scale using stochastic differential equations, partial differential equations and partial integro-differential equation. In general, spatial and temporal scale in social insect models is interesting and these systems often lie at intermediate scales. If you assume everything is smooth and use partial differential equations you often miss out on important effects of individual behaviour; if you use a cellular automata or individual-oriented model, you may not be able to make it large enough to capture the behaviour observed in the real colony.
Social insect modelling is a new and growing field, not only of biological interest but also with applications to information technology and is likely to develop significantly in the next ten to fifteen years.