We are all interested in relationships. Archaeologists and historians constantly talk about how entities (whatever they are interested in) are related to each other, and they sometimes even use network concepts to describe these relationships. But is this network science? Should we see a network every time we think about past relational phenomena? Of course not! In this post I will discuss how it is crucial for archaeologists and historians to clearly define what they mean by “network”. Otherwise we see networks everywhere and don’t understand what advantages this offers over other approaches. In which case we might as well see networks nowhere at all.
This is the second in a series of posts about our The Connected Past book that was recently published. Have a look at the book on the Oxford University Press website, and use this discount voucher if you wish to order it.
In the introduction to our book we aim to make an argument for the need to clearly define what makes network science special. How does it allow archaeologists and historians to do something they could not do before? We are interested in many past relational phenomena, but when does a network science approach really add something?
To answer these questions we need to first illustrate that there is indeed a need to define network science. Isn’t it good enough to just say that every study that is interested in past relational phenomena and abstracts these using concepts to describe entities and relationships vaguely qualify as network approaches? And that in those research contexts it’s probably worth our time to apply network science?
No! This is just not good enough, because network approaches defined like that are everywhere. Almost any interpretative statement ever made by any archaeologists would then qualify as a network approach.
To make our point, we decided to pick up three books from the library and quote randomly selected sections, gambling that they would fit in this vague definition of network approaches I just wrote. Here’s the first:
In the process of diffusion and creation the isle of Crete played a foremost rôle. Its geographical position enabled the Cretans to take advantage of advances made in the South and East without becoming dependent either on Egypt or on Sumer. At the same time the limited resources of their homeland obliged islanders to turn to maritime trade and thereby to diffuse their civilization along the coasts of the Mediterranean and the Black Sea. (Childe 1925: 24).
And now a second random excerpt from a different book:
The practice of fortification can give a more reliable indication of inter-village or inter-regional hostilities. Nea Nikomedeia may conceivably have had some defensive works in early neolithic times (Rodden 1965, 84), while during the middle neolithic period two sites in Thessaly, Sesklo (Tsountas 1908, 75) and Magula Hadzimissiotiki, had enclosing walls possibly defensive in purpose, although this is not certain. Servia in Macedonia at about this time was surrounded by a ditch, possibly defensive in purpose (Heurtley 1939, 45). The outstanding example of a site apparently fortified during the neolithic period, although in the late phase, is afforded by Dhimini (dig. 18.13, 1). While less impressive in appearance on the ground than it appears in plan, the site has a central area enclosed by several lines of concentric circuit walls of stone. It is clearly the precursor of such early bronze age citadels as Troy. The concentric arrangement emphasising the significance of the central buildings contrasts markedly with the altogether democratic arrangement of the earlier neolithic villages. (Renfrew 1972: 392).
And a third:
A model of the interior of a house from Ovcarovo provides an exact representation of the interior of the houses excavated at that site (Todorova 1978). I will discuss the nature of these models later in this chapter. For the moment, it is of interest to note that the “oven” in the house model has a sloping gabled roof and openings in the front and side walls. Indeed this “oven” looks like a “house” model. “The shape of the ovens is so much like that of the houses, that during excavations models of ovens have sometimes been mistaken for models of dwellings” (ibid., 52). Certainly many house models do have sloping roofs, but it is unclear whether they should be reinterpreted as ovens, or whether the ovens mimic houses, or whether the Ovcarovo house model oven should be reinterpreted as a “shrine”. Whatever the answer to this problem, it seems likely that the centrality of the oven within the house was sometimes reaffirmed by drawing symbolic parallels between houses and ovens. (Hodder 1990: 57–59).
OK, I have to admit these books were not picked up at random. The archaeologists among you will have spotted the names of three well-known archaeologists who are commonly associated with very different ways of interpreting the past, and these are some of their more well-known books. And the excerpts are also not entirely random but were drawn from sections of the books where the archaeological record is described and interpreted, rather than sections setting out the theoretical or methodological frameworks employed.
But that really does not matter, almost any other excerpt from these sections that includes some form of interpretative statement (or from books written by other archaeologists for that matter) would have served our purpose, since the point we wished to make with this exercise is very simple: archaeologists (like historians) are interested in past phenomena that can often be abstracted using relational concepts, and they make assumptions about what entities and relationships mean, what kind of behaviour they allowed for and the implications for understanding the past phenomena they are interested in.
Childe aims to understand the role played by the island of Crete and its inhabitants in the broader Mediterranean/Black Sea region and assumes that the geographical structuring of entities allowed for and determined the spread of innovations, the creation of power dependencies, and the flow of material resources. Renfrew is interested in inter-village and inter-regional hostilities and how this is expressed in architecture and settlement planning. He assumes that changes in the nature of hostilities gave rise to fortifications, which in turn changed the nature of those hostilities themselves. Hodder is interested in the role of house and oven models, and the symbolic parallels between them, and assumes that conceptual analogy is expressed through morphological similarities, allowing him to make interpretations regarding the centrality of the oven within the house.
All three examples explore complex past phenomena defined by multiple interacting entities (e.g. the geographical arrangements, trade routes, and cultural diffusions between Crete, Egypt, and Sumer; the spatial arrangements of regions, villages, and individual buildings; and the symbolic parallels and analogies between houses, ovens, and people) through a more abstract framework of concepts. Entities may or may not be conceptualized as clearly bounded and separable physical things, but they are regarded as analytically distinct for the purposes of the specific arguments being forwarded in each case.
Network concepts can be intuitively appealing and particularly suitable for describing many of the past phenomena we aim to understand. However, to conflate ‘relational’ studies like this with ‘network perspectives’ in the more formal sense risks devaluing both approaches: if almost every past phenomenon can be described using loosely defined ‘network’ concepts we end up seeing networks everywhere, but they will rarely lead to critical insights since it is not clear what advantages they offer over other approaches.
The more formal network perspectives explored in The Connected Past book come with additional criteria in terms of data requirements and/or method that make them more suitable for addressing certain research questions over others. These criteria need to be understood if formal network perspectives are to fulfil their potential of contributing new perspectives on the past.
Next time I will continue this argument by introducing a definition of network science that clearly positions it within the archaeologist and historian’s toolbox: it makes clear what’s so special about network science that it deserves writing a book about.