New book uses network analysis to explain complex archaeological data

For over a decade, Matthew Peeples has researched and studied how network science can be used for archaeology. What can look like constellations of stars when graphed, network science has been used to describe complex relationships among people and things across a broad range of scientific disciplines. 

Peeples, an associate professor and director of the Center for Archaeology and Society at Arizona State University's School of Human Evolution and Social Change, and his colleague, Associate Professor Tom Brughmans at Aarhus University in Denmark, are publishing the first comprehensive guide to network analysis using archaeological data. 

“This book is a comprehensive overview of network methods, theories and models designed specifically for an audience of archaeologists and those in related fields,” Peeples said. “It is both a ‘how to’ of sorts but also provides a lot of guidance on the importance of building network theories and using network methods to address different kinds of archaeological questions.”

The book, “Network Science in Archaeology,” is available for preorder now with a publication date of next month. ASU News spoke with Peeples about the book and the topic of network science for archaeology.

Question: What is a network in the sense you use the term in this book?

Answer: A network is simply some set of nodes, which could be people, organizations, locations or other entities, and a formal definition of the connections among them. Connections in a network can be based on relationships like friendship and frequent contact or formal exchange relations, or even physical features like roads. 

Network methods have been used to study a broad range of phenomena including human social organization, computer networks and even things like flights between airports. This work has revealed some interesting commonalities in network properties across a broad range of settings, which allow researchers to make formal comparisons and predictions about the outcomes associated with different kinds of network processes.

Many people are probably used to seeing network diagrams showing nodes and connections these days as they make their way into news stories and popular media. Network graphs are powerful visuals that help communicate complex patterns in network data.

Q: As an archaeologist, you deal with artifacts and archaeological sites. How do you go from that kind of information to networks?

A: There are lots of different ways that archaeologists have generated formal networks using archaeological data. This includes things like building a network from a set of archaeologically documented settlements where the connections between them are defined by the presence of roads. Much of my own work has focused on material culture, or the objects that people make, use and discard that we archaeologists find. 

For example, I’ve recently been working on a project focused on defining networks among sites in the U.S. Southwest and Mexican Northwest using obsidian or volcanic glass. Obsidian was used to make tools like arrow points and can be traced to specific geographic sources using geochemical methods. Since we know where the sites are and where the obsidian originated, we can use this information to build a network connecting those locations to explore the structure of exchange and distribution relations. I’ve done similar work exploring connections in terms of similarities in materials like ceramics and even architectural details of houses. There are really lots of options, and in this book, we try to outline many of those in the literature and areas with future potential. 

Q: Have networks played a role in archaeology for a long time?

A: Archaeologists have been interested in the structure of relationships for things like long-distance trade and exchange for a long time, but the use of network models and formal analyses only gained popularity recently. We track this history in the book, and the earliest use of formal network methods in archaeology we were able to find was in 1967. Things really don’t pick up to more than a publication or so per year until the last 10–15 years. There have been nearly twice as many network publications in archaeology in the last 10 years than in the previous 50, so we’ve seen really rapid growth.

A simple example of a network graph. In this graph you can clearly see clusters of nodes in red and blue that have lots of connections among them, as well as the node in yellow, which is intermediate between these clusters. Graphic courtesy Matthew Peeples

Q: What is something that you are very excited for people to see with this book project?

A: One of the things I’m most excited for with this book is that it isn’t just a book. We’ve also created a very detailed online companion, which provides data and code to replicate all of the analyses and even lots of things we didn’t have room for in the book. By providing the analytical code in commonly used programming languages like R and Python, I’m really hoping that anyone who wants to try these methods, but has perhaps been intimidated by the programming side, will feel able to jump right in. The online resources were developed in part in relation to a graduate course on network methods I taught here at ASU so that class provided a great test run of the resource. I’m really excited we were able to do this.

Q: Has there been anything surprising you’ve discovered as an archaeologist using network analysis that you would otherwise not have known?

A: I conducted an interesting analysis looking at settlements in the U.S. Southwest that fell within important intermediate positions that network analysts often call brokerage positions. In a network, a broker is a node that connects other pairs of nodes that would otherwise be disconnected. It’s a sort of intermediate or middle-person position. What I found particularly interesting is that outcomes for settlements that fell in broker positions changed through time. During a period of economic growth and expansion across the Southwest in the 10th–12th centuries, sites in brokerage positions tended to be large, long-lived and stable, and found in some of the most densely populated portions of the region. During later centuries, as populations began to decline and settlements were retracting, sites in brokerage positions were instead among the smallest, shortest-lived and isolated settlements. 

This was very interesting to me because it shows how the risks and rewards of particular kinds of network positions can change through time in relation to the broader social, political and economic circumstances. This long-term perspective on network dynamics is something I think archaeology could add to perspectives on networks in the social sciences broadly.