Geography Reference
In-Depth Information
that relatively simple rules of agglomeration and centrifugal forces can produce emer-
gent spatial orders that resemble edge cities and clusters or show statistical regularities
(Arthur, 1994; Batty et al., 2004; Krugman, 1994, 1996). At best, they may generate
selected 'stylised facts' in economic geography and 'space' appears in these models as
a patterned outcome, but the models tell us little that is new about complex economic
change and system adaptation over space. And the nature of geographic space in these
modelling exercises is either wholly abstract or empirically crude. As noted earlier,
some of the characteristics of non-linear, path-dependent system change have also been
applied to geographical entities such as clusters, regions and nation-states (see Arthur,
1994; Garnsey, 1998; Krugman, 1996; Martin and Sunley, 2006). However, we should
start by carefully considering whether such spatial entities can legitimately be described
as complex adaptive, self-organising systems.
The dynamics of complex adaptive systems depend on their coni guration and how
this responds to innovation and shocks. Such systems are constituted by interconnections
across multiple scales between diverse and heterogeneous agents. Many of them display
spatial behaviour that bears crucially on their robustness and adaptability. Moreover,
local co-evolutionary ef ects within such systems typically produce innovation and
novelty (Markose, 2005; Metcalfe et al., 2006). Indeed, it has recently been argued that
complex systems are preternaturally 'spatial' (Thrift, 1999). O'Sullivan et al. (2006), for
example, argue that spatial variability is central to complexity, as where elements are
located relative to other elements is critical to their behaviour (Manson and O'Sullivan,
2006). As Bullock and Clif (2004) have argued, however, the role that spatiality plays
in underpinning complex adaptive behaviour is poorly understood. While many of the
leading accounts of complexity and complexity economics discuss system movements in
'state-spaces' and their adaptive walks on 'i tness landscapes', they say little about geo-
graphic space and its relation to the adaptive behaviour of individuals and businesses.
If there is a spatial metaphor essential to complexity economics then it is the 'network'.
Complexity approaches represent 'the economy' as made up of innumerable l ows and
connections, and they are predicated on the notion of incomplete and selective net-
works. The starting point is that everything is not connected to everything else so that
the 'force i eld' metaphor underlying neoclassical economics is inappropriate (Potts,
2000). Instead, the bounded rationality and imperfect knowledge of economic agents
mean that we should address imperfect and incomplete networks that are irreducibly
broken and partial, and in this sense they are spatially distributed, with a bias towards
some degree of localisation. Complexity thinking tends to start with the assumption that
components interact most strongly with their nearest neighbours, and in some physical
systems this means a form of distance decay. The idea of complex territorial economies
is then most aptly applied where interactions between agents are geographically local-
ised. But, of course, where distance is as much or more 'relational' rather than spatial,
'localised' interactions do not always translate into spatial proximity. As Cilliers (2005a,
2005b) has argued, we can not always assume that socio-economic subsystems are spa-
tially contiguous. Parts of socio-economic systems may exist in totally dif erent spatial
locations and systems may interpenetrate each other and so be part of dif erent systems
simultaneously. Localised interactions can then simply mean interaction with selected
other components. Ultimately, it is plausible to argue that spatial entities such as regions
and cities become self-organising complex systems when they are strongly interactive
