Transcript Slide 1

Thinking Small and Long:
Service-Dominant Logic &
Agent Based Modeling
S-D
Logic
Robert F. Lusch
Lisle & Roslyn Payne Professor of Marketing
University of Arizona
University of Hawaii
March 10, 2006
Small and Long Thinking
S-D
Logic
S-D Logic
Agent Based
Modeling
Thinking
Small
All agents
exchange service
or competences.
Agent microscopic
actions and
interactions.
Thinking
Long
All economies are Evolution of
service
complex adaptive
economies.
systems.
S-D Logic & ABM as a Paradigm Shift:
From Constructs to Actors
S-D
Logic



Virtually all social science theory models
relations between constructs.
S-D logic views marketing as interactions
between entities and ABM provides the
method to model and research these
interactions.
What emerges from interactions?




Macro structures
Relations between variables
Rules (institutions and norms)
Co-creation
Building Markets from Ground Up
S-D
Logic
Digital Organisms
Genetic algorithms
Object
Oriented
Programming
Fuzzy Logic
Data Capturing &
Aggregation
Object Oriented Programming
S-D
Logic



OOP Integrates Data and Functions.
Every digital organism is an object with
its own information and functions it
uses to operate.
Every digital organism has receptors,
memory, decision system, and effectors.
Creation of Digital Life
S-D
Logic
Object Oriented
Software Program
Environment
Sensory
Capability
Memory
Capability
Learning & Decision
Capability
Effector
Capability
Environment
Genetic Algorithms & Digital Learning
S-D
Logic
Learning Mode
Genetic Mechanism
Imitation
Selection &
Reproduction
Communication
Crossover
Experimentation
Mutation
Decision-Making: From Substantive
Rationality to Procedural Rationality
S-D
Logic




Simon (1978) argues the concept of rationality is
“economics” main export to other social sciences.
In complex environments actors evolve and their
actions and anticipations are unknown from each
other; the relevant rationality is procedural rationality.
These environments are the “permanent and
ineradicable scandal of economic theory” (Simon
1976).
Mind is the scarce resource; how the actor finds
efficient and effective search algorithms is the key.
Procedural Rationality: How do
Individuals Reason & Learn?
S-D
Logic




Inductive reasoning—ampliative method
of reasoning (gap filling)
Extinguish rules or actions that are
unsuccessful and adopt rules or actions
that are successful—market hypotheses
Information processing and actions not
fine-grained but are fuzzy
Memory lingers; little is completely
forgotten
Fuzzy Logic
S-D
Logic
Weekend Days



Saturday
Friday
Sunday


Lack of crisp, welldefined boundaries
Membership in two or
more sets
Imprecise linguistic
concepts
Everything a matter of
degree
Speed of perception
and information
processing
A Pair of Interesting Observations
S-D
Logic

What used to work no longer works?



Competitive dynamics
Competition is a disequilibrating process
If it works don’t fool with it.



Learning via exploitation
Learning via exploration
The ambidextrous organization
Real Competitive Markets
S-D
Logic








Competition is an evolutionary & disequilibrating process
(Schumpeter 1934; Alchian 1950; Nelson & Winter 1982)
Competition occurs in uncertain world and competition is
a knowledge discovery process (Hayek 1935)
Demand and supply are heterogeneous (Chamberlain
1933; Alderson 1957, 1965)
Competition involves a struggle for advantage (Clark
1954; Alderson 1957, 1965)
History counts (North 1981; Chander 1990)
Entities constantly strive to do better (Bain 1954, 1956)
Resources are tangible and intangible and imperfectly
mobile (Penrose 1959; Lippman & Rumelt 1982).
Knowledge is the fundamental source of competitive
advantage (Vargo & Lusch 2004).
Competitive Dynamics:
Simple Rules
S-D
Logic





Sellers must independently decide on price,
advertising, product attributes, inventory
level.
Seller has four fuzzy states (low, moderately
low, moderately high, high) for each of four
decisions. 44 = 256 rules
These 256 rules form a “market hypothesis”
Ten rule bases characterize 10 market
hypotheses each seller uses.
Utilization of which market hypothesis to use
is based on their fitness.
Simple Setting: Complex Market
S-D
Logic
Buyers are
heterogeneous with
preferences in ndimensional product
space.
Sellers have cost functions
and decision alternatives.
Decisions include price,
advertising, product
attributes, inventory.
Buyer demand is a
function of price,
advertising, product
offering, social capital.
Buyer demand function
is homogeneous and
non-changing.
Sellers have four fuzzy states
for each of four decisions.
Thus each seller has 256
rules which for a market
hypothesis. Each seller has 10
market hypotheses. Each
market hypothesis has a
fitness function
How Fuzzy Inputs Interact to
Affect Price Decision
S-D
Logic
Price Decision
70
60
50
40
30
2
1.5
2
1.5
1
1
0.5
Relative Social Capital
0.5
0
0
Relative Price
Evolution of Profit Payoff from Price: Seller-1
Seller-1: profit impact of price
S-D
Logic
0.35
0.3
price coefficient
0.25
0.2
0.15
0.1
0.05
0
1
2
3
4
5
6
7
8
9
10
11
time
12
13
14
15
16
17
18
19
20
Evolution of Profit Impact from Price Across Sellers
Dynamic Price Impacts Across Sellers
S-D
Logic
0.35
0.3
Profit Impact of Price
0.25
S1:1 b(2)
S2:2 b(2)
S3:3 b(2)
S4:4 b(2)
Poly. (S1:1 b(2))
Poly. (S2:2 b(2))
Poly. (S3:3 b(2))
Poly. (S4:4 b(2) )
0.2
0.15
0.1
0.05
0
1
2
3
4
5
6
7
8
9
10
11
Time
12
13
14
15
16
17
18
19
20
Evolution of Cross Profit Impact from Price: Sellers
1 &2
S-D
Logic
Seller 1&2 Cross Price Impacts
0.3
Profit Impact of Competitor Price
0.25
0.2
S1:2 b(2)
S2:1 b(2)
Poly. (S1:2 b(2))
Poly. (S2:1 b(2))
0.15
0.1
0.05
0
1
2
3
4
5
6
7
8
9
10
11
Time
12
13
14
15
16
17
18
19
20
The Ambidextrous Organization &
Evolutionary Biology
S-D
Logic


When the environment changes slowly then
mechanisms of exploitation that work on
variation, selection and retention work
well.We learn by communicating and do this
primarily by crossover.
When there is dramatic shift in the
environment or a punctuated equilibria then
relying purely on exploitation will not allow
the organism to survive. It must explore to
innovate or face extinction.
The Ambidextrous Organization:
Modeling Exploitation with Crossover
S-D
Logic
Moderate Crossover (moderate
exploitation) is represented by 50%
probability of crossover every 30 periods.
High Crossover (high exploitation) is
represented by 100% probability of
crossover every 30 periods. In this
situation the seller takes advantage of
every opportunity to investigate the space
for a good solution.
The Ambidextrous Organization:
Modeling Exploration with Mutation
S-D
Logic
High Mutation (high exploration) is represented
by 50% probability of mutation every 30 periods.
Moderate Mutation (moderate exploration) is
represented by 25% probability of mutation every
30 periods.
Low Mutation (low exploration) is represented
by 5% probability of mutation every 30 periods.
Simple Setting: Complex Market
S-D
Logic
Buyers are homogeneous.
Buyers in market-A are
stable and do not change
their preferences but in
market-B change their
preferences randomly
every 1500 periods.
Buyer preference is a
function of price and
product offering.
Sellers have cost functions
and decision alternatives.
Decisions include price,
product attribute,
production level.
Sellers have four fuzzy
states for each of three
decisions. Each market
hypothesis has 64 rules.
Sellers vary in the
exploration & exploitation.
Organizational Learning Strategies
S-D
Logic
Low
Exploration
Moderate
Exploitation
High
Exploitation
Seller-Four
Moderate
Exploration
High
Exploration
Seller-Two
Crossover = .5
Mutation = .25
Seller-One
Seller-Three
Crossover = 1.0 Crossover = 1.0
Mutation = .05 Mutation = .25
Crossover = .5
Mutation = .5
Market-A: Stable World
S-D
Logic


Buyer preferences are fixed or
unchanging.
In this situation we would expect the
organization that focuses heavily on
exploitation as a learning mechanism
and seldom uses exploration to learn to
perform best (seller four). On the other
hand an organization with high
exploration would do poorly (seller
one).
S-D
Logic
S-D
Logic
S-D
Logic
Stable World
Market B: Turbulent World
S-D
Logic


Buyer preferences are randomly
changed every 1500 periods
(50*crossover frequency).
In this situation we would expect
ambidextrous organizations to do best.
The organizations that both, to a good
degree, exploit and explore. This would
be sellers 2 or 3. Seller four who hardly
ever explores should perform the
poorest.
S-D
Logic
Turbulent World
Profit Payoffs
S-D
Logic
Stable
Turbulent
Environment Environment
Seller-1 (low exploit;
high explore)
($256,372)
$185,182
Seller-2 (low exploit;
mod explore)
($247,593)
$105,849
Seller-3 (high exploit;
mod explore)
($ 52,813)
$307,339
Seller-4 (high exploit;
low explore
$417,781
($46,703)
($138,997)
$551,667
TOTAL MARKET
Moderating Effect:
Market Environment (average profit)
S-D
Logic
$500,000
$400,000
$300,000
$200,000
Ambidextrous
Non-Ambidextrous
$100,000
$0
($100,000)
($200,000)
Stable Turbo
Concluding Observations
S-D
Logic