Geography Reference
In-Depth Information
results before we can decide on the appropriate space to represent our compounds'
(Nightingale, 2000, p. 337), applies to the entire innovation process in clusters. Human
agency is important, through the purposeful projection of a vision of sorts and through
the actions intended to fuli ll it, but this does not imply predictability. Agency is an
indeterminate process in which no actor has dei nitive knowledge and in which clear
models are often lacking. People often make decisions based on cognitive distortions
and perceptual errors (Levinthal and March, 1993). They select initiatives because they
trust the information from small samples, they overestimate their analytical capabilities,
and they interpret events ex post as something they responded to strategically when, in
fact, they reacted haphazardly. This view of intentionality departs somewhat from that
followed by complexity theorists who interpret the emergent properties of clusters as the
spontaneous
outcome of individual actions (Lindsay, 2005). It also challenges the view of
evolutionary theorists who think of organizations as relatively stable bundles of routines.
Procedural organizational memory may help to institutionalize certain ideas, protecting
them from extraneous inl uences. But this does not imply that such ideas will (automati-
cally) structure actual behavior.
While variation in the cognitive capabilities and imaginativeness of agents may be
crucial to system evolution, existing institutional and social forces may reduce the diver-
sity of outcomes from such variation. The task for evolutionary analysis is to assess the
net balance between these opposing forces in particular instances. In dynamic systems
such as clusters, it may be useful to consider blind variation in agency, together with
chance events occurring in volatile environments, a baseline model with a null hypothesis
and then to explain the emergence of new variations beyond stochastic developments.
For this kind of analysis, all actions, interpretations, and decisions have
potential
rel-
evance, and any attempt to create new knowledge, develop a new strategy, or build new
relationships has
potential
value. The practical implication of this is that intervention
should not aim at compensating for variations with unknown adaptive value, but should
embrace these variations through, for example, supporting entrepreneurs located at the
periphery of a cluster (Belussi et al., 2008), building institutions that encourage continu-
ous experimentation (Grabher and Stark, 1997), or encouraging rel exive network man-
agement practices (Sydow, 2003).
Mechanisms of idea transmission
If the evolutionary approach is to deliver on its promise of explaining persistence and
change, it must specify the causal mechanisms that operate at various levels in a given
system. Theorists should not be content just with a process account of evolving socio-
economic systems. Understanding the causal mechanisms operative in such systems
is essential, both in itself and for understanding process. Cluster researchers often use
causal concepts when they refer to actors 'constituting' networks or 'enacting' beliefs
that then 'dif use' throughout the cluster, institutions 'enabling' or 'constraining' gov-
ernance, or historical legacies 'embedding' i rms in local social structures. In general,
however, they are vague about the mechanisms by which beliefs are enacted, i rms are
embedded, or institutions constrain behavior. The problem of dei ning mechanisms
arises because the various entities relevant to selection exist in a nested hierarchy. The
evolution of an entity may be viewed in isolation (by treating it as a unit with self-
contained mechanisms of replication), constitutively (by identifying the lower level
