My colleagues and friends doubt my professionalism when I show Caribbean views like this, taken whilst “working”. The last two years I’ve actually been working quite hard, but it’s just difficult to make Caribbean archaeology look like hard work. It was an absolute privilege to meet and learn from colleagues and friends in a number of Caribbean islands, about their knowledge of the landscapes that surround them and their relationship to the peoples who lived there before them. The current natural disasters befalling many of these islands, and their fall-out, are terrible. I wish all my friends there much strength and hope their respective governments (and aid workers around the world) do their duties in taking care of their residents.
A first paper, co-authored with Maaike de Waal, Corinne Hofman and Ulrik Brandes, explores what we can learn about Amerindian social networks by examining Caribbean views: how places are connected based on what can be seen from them. It applies a wide range of computational methods (visibility networks, total viewsheds, visual neighbourhood configurations), but it should not be seen as a methodological exercise. The paper aimed to express some of the ideas that Caribbean archaeologists have formulated about how views could have mattered to past peoples, because they could be used for navigation, to share information through smoke or fire signalling, or to determine suitable settlement locations. Doing so led to some unique insights into the connectivity of landscapes in Eastern Guadeloupe (the paper’s research area), that led us to formulate a theory about the structuring role played by views in the pre-colonial Lesser Antilles as a whole: short-distance views at which people or smoke signals could be seen structured placement of settlements and community interactions locally, within regions on landmasses; whereas long-distance views at which only huge landmasses could be seen would structure navigation between communities on different landmasses. We see the Lesser Antilles as consisting of thousands of local connectivity clusters, all connected through the long-distance visibility of landmasses (see figure below).
A second paper, co-authored with Ulrik Brandes, vastly expands the methodological toolbox for visibility network methods. Having reviewed the archaeological use of formal methods for studying visibility phenomena (i.e. what people in the past could see), we noticed that there was a discrepancy between the theories formulated and the methods used to explore them. The theories were often very complex, involving many different ways in which visibility could have structured past human behaviour and could have affected past human decision-making. Few of these theories have been explored using formal methods, often because of their share complexity, and those that have been treated formally were explored with a very limited range of formal methods: mainly binary viewsheds and simple visibility network representation. So we thought there was some fun methodological work to be done here, that could benefit future archaeological (and other) research. We approached visibility as a purely relational phenomenon, connecting the observer’s eyes with the observed feature. Doing so allowed us to represent any kind of visibility study to be represented as networks, which led to some really cool new network representation. For example (see figure below), a cumulative view shed can be represented as a two-mode network where observation points like site locations are connected to the landscape locations that can be observed from them. This two-network can be split up into two one-mode networks: a network where sites are connected if they have landscape locations in common that can be seen from both, and a network where landscape locations are connected if they have sites in common from which both can be observed. In addition, we also explored how complex theories of visibility can be teased apart into their constituent parts, where each part is represented by a small network data representation. We can count the frequency of these patterns and even simulate the preferential creation of these patterns, to explore how probable our complex theories are.
I will write more about these studies at a later time. All of this work was funded by EU HERA and ERC Synergy funding. Don’t hesitate to get in touch if you like this kind of thing!