Agent Models

What is agent-based modelling,and why does it matter?
As archaeologists, all we ever recover are the traces of individual interactions. We don’t find the evidence of the grand sweep of history; it is but rarely that we find the echoes of famous events. For every Pompeii, there are hundreds of thousands of anonymous hovels.

Yet all the same, the archaeologist interprets all of these traces and is able to pronounce on events of grand durée. There is an enormous gap between the artefact and the interpretation. How we bridge this gap depends on how we think individual interactions combine to form society.

Agent-based modelling on the other hand doesn’t necessarily depend on us to do this explicitly because it is predicated on the idea that society emerges from individual interactions through self-organisation. Using a computer, we instruct autonomous agents in the rules by which to behave. These rules are the behaviours that we observe in the archaeology, in the traces of individual interactions that we find. Then, we set the agents loose, and let them interact with each other as per the rules. From all of these countless interactions, iterated over and over again, larger-scale behaviour (society) begins to emerge.

Thanks to the computer, we can study these large scale behaviours (what in human terms is social science) and compare our artificial society with what we know or believe about the ancient society. If we have deduced correctly the behaviours behind the individual interactions found archaeologically, then our artificial society should be an excellent analogue to the ancient society.

History runs only once. But in the computer, it can run over and over again. We can explore (by altering the variables) the entire possible range of outcomes for different behaviours. We can compare what did happen with the parameters of its closest silicon analogue, and know the ancient society better than ever before.

Purges & Proscriptions – An Agent Based Modelling Project

One of the central issues in the social sciences concerns individual agency and its relationship to wider society, of how individual actions (free will) give rise to social reproduction and collective behaviour. Turning the question around we can ask what happens to collective behaviour when individuals are lost from society? My research examines this problem in the light of an archaeological case study. Using archaeological evidence for social relationships in Roman Central Italy (the interconnections, evident in stamped brick from the Tiber Valley, between different levels of society), Purges & Proscriptions is an investigation into the self-organisation and robustness (the ability to withstand stress) of ancient Roman Society.

Episodes of elite self-extermination mark many passages of Roman History, yet seemingly without any fundamental consequences for society at large. What is it about the structure of society in the Roman world which allowed it to weather the loss of major figures in social and economic life? In a culture where the economy is embedded in social and political networks, the changing pattern of those networks has important ramifications for understanding historical change. The initial aim of this project therefore is to understand and explain the structural dynamics of Roman society from the point of view of individuals.

Watch this space

As time goes by, I will be placing copies of my models on-line. I use the Netlogo modelling environment. This environment can be learned easily by people without any great background in computer programming. The code for the models is written in a ‘near-english’ syntax, which makes the models fairly intelligible to the non-specialist. It is my hope that others will play with the models, offering their critiques, evaluations, and suggestions for improvements.

If you are interested in learning more about Agent-Based Models and Netlogo, may I recommend the Institute for Modeling Complexity’s Netlogo Workshop at Mesa State College, Grand Junction Colorado! I found the workshop to be enormously helpful and well worth the trip!

update Jan 15 2008: unfortunately, the link above no longer seems to work, so perhaps the workshop is no longer happening. Fortunately, there are other places you can go to jump into this fascinating field - check out this list of resources maintained by the Netlogo people.