This fourth blog post in the series discusses process-related issues in archaeological network studies. As I mentioned before, I recently published a review of formal network methods in archaeology in Archaeological Review from Cambridge. I want to share the key problems I raise in this review here on my blog, because in many ways they are the outcomes of working with networks as an archaeologist the last six years. And yes, I encountered more problems than I was able to solve, which is a good thing because I do not want to be bored the next few years 🙂 In a series of four blog posts I draw on this review to introduce four groups of problems that archaeologists are faced with when using networks: method, data, space, and process. The full paper can be found on Academia.
Many archaeological network studies treat networks as static snapshots. This is at least in part a result of the nature of archaeological data and our inability to observe past processes directly. Graph visualisations and many network analysis techniques further enforce this idea of a static network by exploring structural features of particular networks in isolation. However, the past systems we study were dynamic phenomena and the network approach used to understand these phenomena should reflect their changeable nature. In fact, one could argue that no network is truly static since our assumptions underlying the creation of ties imply flows of resources, which are dynamic processes taking place in a changing network.
Archaeological data often does not have the chronological accuracy to reconstruct an exact sequence of events: which ties and nodes appeared and disappeared in what order? A number of network modelling approaches exist that can help one deal with this issue, including agent based modelling (e.g. Graham 2006), algebraic modelling (e.g. Menze and Ur 2012), and statistical modelling (e.g. Lusher et al. 2013). Underlying all of these modelling approaches are clearly formulated assumptions of what a relationship means and what types of flows it allows for. They therefore require one to explicitly acknowledge the dynamic nature of past processes and the dynamic assumptions underlying the definition of ties.
But which model is best? Many models, representing different hypothetical processes, can be created that could all give rise to the same observed network. Since archaeologists cannot directly observe past processes, and given that our data are incomplete and are merely indirect proxies, how then can we ever claim that one model is more probable than any other? The problem of equifinality (the idea that multiple processes can have the same end result) is equally critical for network analysis as for any other technique in the archaeologist’s toolbox. There are a few ways in which formal network methods can help us address this issue. Firstly, archaeological data (however flawed) used in statistical models can help us to identify very general processes that are more probable than others. Secondly, these models can help us to formally express otherwise ill-defined hypotheses and evaluate their likeliness given certain archaeological assumptions. Thirdly, they might not be able to prove certain processes, but models can definitely be used to negatively test or falsify certain hypotheses, or at least identify which processes are less likely than others (given our current knowledge). In this way, such models serve as experimental laboratories (Premo 2006). One has to acknowledge, however, that some past processes are unknowable given our current techniques and datasets. All archaeological approaches suffer from this disadvantage and network analysis is no exception.
Graham, S. (2006). Networks, agent-based models and the Antonine Itineraries: implications for Roman archaeology. Journal of Mediterranean Archaeology, 19(1), 45–64.
Lusher, D., Koskinen, J., & Robins, G. (2013). Exponential Random Graph Models for Social Networks. Cambridge: Cambridge university press.
Menze, B. H., & Ur, J. a. (2012). Mapping patterns of long-term settlement in Northern Mesopotamia at a large scale. Proceedings of the National Academy of Sciences of the United States of America, 109(14), E778–87. doi:10.1073/pnas.1115472109
Premo, L. S. (2006). Agent-based models as behavioral laboratories for evolutionary anthropological research. Arizona Anthropologist, 91–113.