Page - (000096) - in Knowledge and Networks

88 innovation has taken place, especially with respect to place accessibility and multi- level analysis (Mathis, 2003), dialogue with other social sciences remains the exception rather than the rule. However, these studies can provide valuable input for political geography because the structure and characteristics of technical networks can reveal social, economic, and spatial patterns. In what became for decades the reference handbook of network analysis in geography (Haggett & Chorley, 1969), one small example puts into perspective the physical aspects, socioeconomic vari- ables (e.g., urbanization, hierarchy of cities), and indices of road and rail networks (pp. 88–89). The density and connectivity of technical networks are factors that reveal the level of regional or national wealth that can explain given patterns of rela- tions among these actors. Another trend in geographical research is relational by nature because it involves the study of flows between places. The use of a model inspired by laws of gravita- tion became a convenient tool in the 1960s and has been common since then to study and model patterns of interaction (Nystuen & Dacey, 1961; Tobler, 1970). Based on valued flows (goods, communication, migration), the techniques devel- oped did not involve any dialogue with social network analysis. Note also that this literature barely makes explicit reference to the term network. However, several recent works by physicists draw on these interaction models mixed with complex network approaches and generally demonstrate that the two methods produce com- plementary results (Gorman et al., 2007). Interest in such approaches for political geography cannot be underestimated; they are able to reveal, for instance, preferen- tial relations as well as barriers between pairs of actors. Mixing tools from flow studies and network analysis clearly appears promising for investigating globaliza- tion processes at multiscalar levels (Van Hamme & Grasland, 2011). Some Recent and Welcome Changes The soaring number of physicists and computer scientists in the network field since the 1990s has changed the landscape on two fronts. First, some geographers were quick to adopt new measurements from related works (clustering coefficient, power law degree distribution). Rozenblat’s papers on economic and transport relations between world cities are good examples of this imitation process (Rozenblat & Melançon, 2007). This early acceptance can be explained by the not-so-new charac- ter of the scale-free network as a power law distribution, which is a common and well-known process in several social sciences (e.g., rank distribution of cities in urban geography, Zipf’s word distribution in textual analysis). It must also be men- tioned that the conclusions offered in the works on world cities remain classic yet are somehow deceptive: Using so-called innovative methods to show that Paris, London, New York, and Tokyo are the main world cities appears to contribute little to geographical knowledge (Rozenblat & Melançon, 2007). More noteworthy is the interest some physicists are taking in spatial networks and models and the methods proposed to analyze them (Barthelemy, 2011; Gastner & Newman, 2006). Methods L. Beauguitte