Information Technology Reference
In-Depth Information
Table 1.
Summary of model comparison
Model &
key
references
Unit of
knowledge
or
technology
Unit repre-
sentation
Value:
pre-defined or
endogenous
Sources of
novelty
Emergent
patterns
L&F
[1]
Workers'
beliefs /
routines
Bit strings
NK fitness
landscape
Recombina-
tion;
Mutation
Convergence on
peak / optimum
S&V
[2, 3]
Technolo-
gies
Nodes in a
grid network
Height &
connection
to baseline
Chance
discovery
from nearby
nodes
Scale-free distri-
bution of
advances of
connected nodes
A&P
[4]
Technolo-
gies
Systems of
NAND logic
gates
List of desired
logic
functions
Recombina-
tion
Scale-free distri-
bution of obso-
lescence sizes
SKIN
[5, 27, 28]
Firms'
“kenes”:
capabilities,
abilities,
expertise
Ordered
triples of
integers
Endogenous
demand for
inputs
Recombina-
tion of
kenes; new
abilities
Social networks;
Scale-free
distribution of
production
network sizes
CJZ
[6]
Firms'
knowledge
Vector of
continuous
variables
Endogenous
demand for
collaboration
partners
Not
specified
Social networks
Hypercy-
cles
[7, 29, 34]
Firms'
production
skills (rules)
Algorithmic
chemistry
Role in hyper-
cycles system
Emergence
of self-
maintaining
system
Self-maintaining
system of firms
& rules
and each rule producing the input to the next rule in the cycle, with the final rule pro-
ducing the input to the first rule. This is clearly a very abstract view of industrial pro-
duction relations. The SKIN model uses modulo arithmetic (that is, its normalised
sum-products) to control the relationships between kenes. This is simple to specify
and compute, but remote from real-world applications. The CJZ model relies on its
decomposability parameter and its CES production function to control innovation
production, and no further description of the underlying knowledge or technological
world is needed, but whether knowledge resembles constant elasticity of substitution
remains to be demonstrated empirically.
