Based on geographer Peter Haggett's (1966) work, network configurations can be used to describe the characteristics of interaction that resulted in the distribution and circulation of goods. In describing Andean caravan transport, Nielsen (2000: 73-74, 91) uses the following terms: (1) distance that goods are transported; (2) segmentary vs. continuous - in segmentary networks a given node is connected to a small number of other nodes, while continuous networks each node is connected to all other nodes; (3) convergent (focalized) vs. divergent (non-focalized) - in convergent networks the individuals participating in exchange, and the goods they transport, tend to concentrate in a small number of central places or exchange locales.

Reciprocal exchange relationships that take the form of down-the-line trade may be described as continuous networks when these mechanisms serve to move goods between ethnic groups and across regions. Centralized political control by elites and true market mechanisms might result in network convergence at central places (Smith 1976). To Nielsen's third set of terms, convergent (focalized), divergent (non-focalized), one may add diffusive to describe the pattern of a single type of item radiating from the center to the surrounding region as occurs with obsidian.

Figure 2-3. Network configurations.

These network configurations serve to draw attention to the limitations of using raw material distributions as a proxy for all exchange behaviors. Obsidian exchange is sometimes used by archaeologists as gauge of the volume, frequency, and structure of prehistoric exchange relationships. As noted by Clark (2003), raw materials from geological sources diffuse continuously from a single point to the region, presumably following trade routes, until the materials are found deposited at archaeological consumption sites. In contrast, much exchange between complementary groups, such as between agriculturalists and pastoralists, is non-focalized and often segmentary, as it links producers and consumers through a variety of localized articulation methods. The structural differences between diffusive exchange networks and regular, household-level interaction are well demonstrated in mountain regions with distinct ecological zonation, such as the Andes.

Archaeological inferences that do not differentiate the expectations of one network configuration from another are problematic. Often, diffusive configurations will have evidence of the transport of goods in the opposite direction, as one might expect in system where obsidian is acquired through reciprocity-based relationships. However, the pattern where goods are reciprocated to the source area is a configuration model to be tested rather than one that can be assumed.