File-file yang bisa di- | Fakultas Peternakan UNSOED Carl Smith1
19/11/2016
SYSTEMS THINKING – AN
ESSENTIAL SKILL FOR
UNDERSTANDING AND
MANAGING COMPLEXITY
Carl Smith
Macquarie Island
• Situated about 1500 km south-south-east of Tasmania,
about half way between Tasmania and Antarctica
• The main island is approximately 34 km long and 5.5
km wide at its broadest point
• Around 3.5 million seabirds arrive on Macquarie Island
each year to breed and moult
School of Agriculture and Food Science, University of
Queensland (UQ)
Macquarie Island
• Many types of feral animals were introduced to
Macquarie Island in the 19th century, including cats,
rabbits, rats and mice.
• Feral cats had a devastating effect on the native
seabird population, with an estimated 60,000 seabird
deaths per year. From 1985, efforts were undertaken
to remove the cats. In June 2000, the last of the nearly
2500 cats were culled in an effort to save the seabirds.
Macquarie
Island
1
19/11/2016
Macquarie Island
With Cats
• So what do you think happened when the feral cats
were removed? The seabirds were saved, right?
Without Cats
Macquarie Island
• What the management plan of Macquarie Island failed
to realise is that cats don’t just eat seabirds, they also
eat rabbits, rats and mice
• The eradication of cats caused an explosion in the
rabbit, rat and mice populations on Macquarie Island,
the impact of which was just as bad for seabirds as
having cats on the island
• This is a classic example of policy resistance (a fix that
failed), which resulted from the failure of managers to
understand the system
Seabird Habitat
Eaten
+
Rabbit Birth Rate
Rabbit Death Rate
+
+
Rabbit Births
R1
+
Rabbit Population
-
+
B1
+
Rabbit Deaths
+
Cat eradication effectively
neutralised the link between
rabbit population and cat
deaths
Cat Death Rate
Cat Birth Rate
B3
+
+
Cat Births
R2
Cat Population
-
+
+
B2
Cat Deaths +
+
Cat Cull
+
Seabirds Eaten
2
19/11/2016
Fixes that Fail – Macquarie
Island
+
Seabirds Lost
+
B
Remove Cats
Delay
The Moral to this Story
(Sterman, 2000, Chapter 1)
• “When you are confronted with any complex system
that has things about it that you want to fix, you cannot
just step in and set about fixing them. You cannot
meddle with one part of a complex system without the
almost certain risk of setting off disastrous events that
you hadn’t counted on in other parts. If you want to fix
something, you are first obliged to understand the
whole system” (biologist Lewis Thomas, 1974)
R
Habitat
Destruction
+
Rabbit Population
+
The Moral to this Story
The Relationship between
System Structure and Behaviour
(Sterman, 2000, Chapter 1)
• To understand the whole system you have to:
1. Integrate knowledge from different disciplines, i.e. be
multidisciplinary
2. Understand how the interactions among system
components (system structure) influence system
behaviour
3. By understanding the relationship between system
structure and behaviour, identify points where you can
intervene to influence system behaviour whilst
minimising unintended consequences
• Complex system dynamics is determined by
interactions and feedbacks among system
components, not by the number of components
• The dynamics of all systems arises from the
interaction of just 2 feedback loops – positive (or
reinforcing) and negative (or balancing)
• Positive feedback loops reinforce or amplify change
while negative feedback loops balance or counteract
change
3
19/11/2016
Positive or reinforcing feedback loops
Negative or balancing feedback loops
Dynamics of Multiple-Loop Systems
Common modes of behaviour in dynamic systems (Sterman,
Chapter 4)
+
Eggs
+
R
+
Chickens
B
Road
Crossings
-
Dynamics arise from the interaction of multiple loops
The fundamental modes of system behaviour
4
19/11/2016
Out of these common modes of behaviour, which one would best
represent the number of people with the flu over the flu season?
Basic Overshoot and Collapse CLD for
the Flu
+
Infection Rate
+
+
Infected
Population
R
+
Recovered or Dead
Population
B
B
Susceptible
Population
-
Immunisation and hygiene
Delay time for
symptomatic people to
recover or die
Delay time for
symptoms to develop
Disease transfer
probability
Infected people starting
to show symptoms
+
Susceptible people
becoming infected
Infected people
recovering or dying
+
+
+
Infected
Population with
Symptoms
Number of contacts
between infected and
susceptible people
+
+
B1
-
R1
Probability of meeting a
susceptible person
Total Population
Total contacts between
infected and other people +
per day
+
+
Total infected
population
Recovered/Dead
Population
+
Infected
Population with
Symptoms
+
Susceptible
Population
Number of contacts
between infected and
susceptible people
+
+
B1
R1
Probability of meeting a
susceptible person
-
+
Recovered/Dead
Population
R2
+
+
Total Population
Contacts an infected person
has with other people per
day
+
B3
B2
Infected
Population
without
Symptoms
-
+
-
R2
+
+
Infected people
recovering or dying
+
+
B3
B2
+
Susceptible
Population
Infected people starting
to show symptoms
+
Susceptible people
becoming infected
+
Infected
Population
without
Symptoms
-
Delay time for
symptomatic people to
recover or die
Delay time for
symptoms to develop
Disease transfer
probability
Total contacts between
infected and other people +
per day
+
Total infected
population
+
Contacts an infected person
has with other people per
day
Keep sick people away from healthy people
B4
B4
5
19/11/2016
Stock and Flow Structure derived
from the CLD
Simulated Behaviour of the System
Reduce contacts per infected person from
2 to 1 other person per day
Reduce chance of disease transfer due to
contact from 50% to 25%
(Infected population peaks at 14,500 at day 22)
(Infected population peaks at 14,500 at day 22)
(Infected population peaks at 17,500 at day 12)
6
19/11/2016
Reduce contacts to 1 person per day and
chance of disease transfer to 25%
(Infected population peaks at 9,500 at day 43)
What has Systems Thinking got
to do with Livestock Production?
• Lots!
• The livestock production system contains many
feedback loops that control its dynamics
• We are currently building a simple livestock production
model for Selayar as part of a World Bank and UQ
funded project called Capturing Coral Reef and
Related Ecosystem Services (CCRES)
Competition with
substitutes and imports
What has Systems Thinking got
to do with Livestock Production?
+
+
+
-
Demand for meat and
livestock products
Price of meat and
livestock products
B1
-
R1
• The system models we are building for CCRES aim to
simulate the interactions between society and coastal
ecosystems to understand problems such as fish
catch decline, mangrove loss, water pollution and food
security
• Some of these interactions are caused by land use
change and land use change is influenced by crop and
livestock production, as well as population growth
• The next slide represents some of the feedback loops
within the livestock production system that we are
trying to model in Selayar
R2
+ Population growth
Household income
B2
Supply of meat and
livestock products
+
+
+
+
+
Livestock
Production
+
B5
B7
+
Consumption of land and
water by population
+
B3
Land and water resource
available for livestock
production
-
Consumption of land and
water for livestock
production
B6
-
B4
Land and water resources
available for population
-
Land and water
+
degradation
+
B8
7
19/11/2016
Livestock Production Model
Livestock Supply Model
Livestock Demand Model
Livestock Price Model
8
19/11/2016
What do these Models do?
• They essentially model balancing loops within the
livestock production system that seek a dynamic
equilibrium between consumption, production and
price
• They also obey limits to growth, that is, livestock
production cannot exceed the carrying capacity of land
and water
• Lets look at a scenario
Population growth
1: Population
1:
2:
3:
4000
450
25
1:
2:
3:
2500
300
15
2: Houses
1
2
3
Total demand for meat
1:
600
1:
350
1
2
3
2
1
3
1
1
Population grows and reaches
limits to growth
3
2
1:
2:
3:
1000
150
5
1
1
1:
0.00
250.00
500.00
Weeks
Page 1
750.00
1000.00
12:30 PM Wed, 16 Nov 2016
100
0.00
250.00
Page 1
Untitled
6
1
2
1:
2:
750.00
1000.00
2:31 PM Wed, 16 Nov 2016
The desired household demand for
meat doubles at week 300 and
goes back to normal at week 700
Meat price
1: total desired liv estoc…y by liv estock ty pe[Cattle] 2: total f attening liv estock by liv estock ty pe[Cattle]
1
500.00
Weeks
Untitled
Total livestock production
1:
2:
There is no change in the desired
household demand for meat
1: total demand f or meat by liv estock ty pe[Cattle]
3: Urban land area
1
Population grows and reaches
limits to growth
1: actual meat price[Cattle]
1:
1.05
1:
0.8
1:
0.55
1
2
2
4
2
1
1
1:
2:
2
0.00
Page 1
250.00
500.00
Weeks
Untitled
750.00
1000.00
2:31 PM Wed, 16 Nov 2016
0.00
Page 1
1
250.00
1
500.00
Weeks
1
750.00
1000.00
12:30 PM Wed, 16 Nov 2016
Untitled
9
19/11/2016
Population growth
1: Population
1:
2:
3:
2: Houses
Total demand for meat
1
1
2
3
Key Lessons
1: total demand f or meat by liv estock ty pe[Cattle]
3: Urban land area
4000
450
25
1:
1100
1:
600
1
2
3
2
3
3
1:
2:
3:
2500
300
15
1
1
1
2
1
1:
2:
3:
1000
150
5
1
1:
0.00
250.00
500.00
Weeks
Page 1
750.00
1000.00
12:30 PM Wed, 16 Nov 2016
100
0.00
250.00
Page 1
Untitled
Total livestock production
Meat price
1: actual meat price[Cattle]
1:
8
7
2
1
750.00
1000.00
2:42 PM Wed, 16 Nov 2016
Untitled
1: total desired liv estoc…y by liv estock ty pe[Cattle] 2: total f attening liv estock by liv estock ty pe[Cattle]
1:
2:
500.00
Weeks
1.05
1
2
2
1
1:
2:
1
5
4.5
1:
0.8
1:
0.55
• When managing any system, understand the
relationship between system structure and system
behaviour
• Use multiple sources of knowledge to develop your
understanding, i.e. be multidisciplinary
• Use your understanding of system structure to
carefully target your interventions, otherwise your cure
may end up being worse than the disease
• When managing systems there is rarely a silver bullet
solution to problems, therefore multiple interventions
may be needed
2
1
1:
2:
1
2
2
0.00
Page 1
250.00
500.00
Weeks
Untitled
750.00
1000.00
2:42 PM Wed, 16 Nov 2016
0.00
Page 1
1
250.00
1
500.00
Weeks
750.00
1000.00
2:42 PM Wed, 16 Nov 2016
Untitled
10
SYSTEMS THINKING – AN
ESSENTIAL SKILL FOR
UNDERSTANDING AND
MANAGING COMPLEXITY
Carl Smith
Macquarie Island
• Situated about 1500 km south-south-east of Tasmania,
about half way between Tasmania and Antarctica
• The main island is approximately 34 km long and 5.5
km wide at its broadest point
• Around 3.5 million seabirds arrive on Macquarie Island
each year to breed and moult
School of Agriculture and Food Science, University of
Queensland (UQ)
Macquarie Island
• Many types of feral animals were introduced to
Macquarie Island in the 19th century, including cats,
rabbits, rats and mice.
• Feral cats had a devastating effect on the native
seabird population, with an estimated 60,000 seabird
deaths per year. From 1985, efforts were undertaken
to remove the cats. In June 2000, the last of the nearly
2500 cats were culled in an effort to save the seabirds.
Macquarie
Island
1
19/11/2016
Macquarie Island
With Cats
• So what do you think happened when the feral cats
were removed? The seabirds were saved, right?
Without Cats
Macquarie Island
• What the management plan of Macquarie Island failed
to realise is that cats don’t just eat seabirds, they also
eat rabbits, rats and mice
• The eradication of cats caused an explosion in the
rabbit, rat and mice populations on Macquarie Island,
the impact of which was just as bad for seabirds as
having cats on the island
• This is a classic example of policy resistance (a fix that
failed), which resulted from the failure of managers to
understand the system
Seabird Habitat
Eaten
+
Rabbit Birth Rate
Rabbit Death Rate
+
+
Rabbit Births
R1
+
Rabbit Population
-
+
B1
+
Rabbit Deaths
+
Cat eradication effectively
neutralised the link between
rabbit population and cat
deaths
Cat Death Rate
Cat Birth Rate
B3
+
+
Cat Births
R2
Cat Population
-
+
+
B2
Cat Deaths +
+
Cat Cull
+
Seabirds Eaten
2
19/11/2016
Fixes that Fail – Macquarie
Island
+
Seabirds Lost
+
B
Remove Cats
Delay
The Moral to this Story
(Sterman, 2000, Chapter 1)
• “When you are confronted with any complex system
that has things about it that you want to fix, you cannot
just step in and set about fixing them. You cannot
meddle with one part of a complex system without the
almost certain risk of setting off disastrous events that
you hadn’t counted on in other parts. If you want to fix
something, you are first obliged to understand the
whole system” (biologist Lewis Thomas, 1974)
R
Habitat
Destruction
+
Rabbit Population
+
The Moral to this Story
The Relationship between
System Structure and Behaviour
(Sterman, 2000, Chapter 1)
• To understand the whole system you have to:
1. Integrate knowledge from different disciplines, i.e. be
multidisciplinary
2. Understand how the interactions among system
components (system structure) influence system
behaviour
3. By understanding the relationship between system
structure and behaviour, identify points where you can
intervene to influence system behaviour whilst
minimising unintended consequences
• Complex system dynamics is determined by
interactions and feedbacks among system
components, not by the number of components
• The dynamics of all systems arises from the
interaction of just 2 feedback loops – positive (or
reinforcing) and negative (or balancing)
• Positive feedback loops reinforce or amplify change
while negative feedback loops balance or counteract
change
3
19/11/2016
Positive or reinforcing feedback loops
Negative or balancing feedback loops
Dynamics of Multiple-Loop Systems
Common modes of behaviour in dynamic systems (Sterman,
Chapter 4)
+
Eggs
+
R
+
Chickens
B
Road
Crossings
-
Dynamics arise from the interaction of multiple loops
The fundamental modes of system behaviour
4
19/11/2016
Out of these common modes of behaviour, which one would best
represent the number of people with the flu over the flu season?
Basic Overshoot and Collapse CLD for
the Flu
+
Infection Rate
+
+
Infected
Population
R
+
Recovered or Dead
Population
B
B
Susceptible
Population
-
Immunisation and hygiene
Delay time for
symptomatic people to
recover or die
Delay time for
symptoms to develop
Disease transfer
probability
Infected people starting
to show symptoms
+
Susceptible people
becoming infected
Infected people
recovering or dying
+
+
+
Infected
Population with
Symptoms
Number of contacts
between infected and
susceptible people
+
+
B1
-
R1
Probability of meeting a
susceptible person
Total Population
Total contacts between
infected and other people +
per day
+
+
Total infected
population
Recovered/Dead
Population
+
Infected
Population with
Symptoms
+
Susceptible
Population
Number of contacts
between infected and
susceptible people
+
+
B1
R1
Probability of meeting a
susceptible person
-
+
Recovered/Dead
Population
R2
+
+
Total Population
Contacts an infected person
has with other people per
day
+
B3
B2
Infected
Population
without
Symptoms
-
+
-
R2
+
+
Infected people
recovering or dying
+
+
B3
B2
+
Susceptible
Population
Infected people starting
to show symptoms
+
Susceptible people
becoming infected
+
Infected
Population
without
Symptoms
-
Delay time for
symptomatic people to
recover or die
Delay time for
symptoms to develop
Disease transfer
probability
Total contacts between
infected and other people +
per day
+
Total infected
population
+
Contacts an infected person
has with other people per
day
Keep sick people away from healthy people
B4
B4
5
19/11/2016
Stock and Flow Structure derived
from the CLD
Simulated Behaviour of the System
Reduce contacts per infected person from
2 to 1 other person per day
Reduce chance of disease transfer due to
contact from 50% to 25%
(Infected population peaks at 14,500 at day 22)
(Infected population peaks at 14,500 at day 22)
(Infected population peaks at 17,500 at day 12)
6
19/11/2016
Reduce contacts to 1 person per day and
chance of disease transfer to 25%
(Infected population peaks at 9,500 at day 43)
What has Systems Thinking got
to do with Livestock Production?
• Lots!
• The livestock production system contains many
feedback loops that control its dynamics
• We are currently building a simple livestock production
model for Selayar as part of a World Bank and UQ
funded project called Capturing Coral Reef and
Related Ecosystem Services (CCRES)
Competition with
substitutes and imports
What has Systems Thinking got
to do with Livestock Production?
+
+
+
-
Demand for meat and
livestock products
Price of meat and
livestock products
B1
-
R1
• The system models we are building for CCRES aim to
simulate the interactions between society and coastal
ecosystems to understand problems such as fish
catch decline, mangrove loss, water pollution and food
security
• Some of these interactions are caused by land use
change and land use change is influenced by crop and
livestock production, as well as population growth
• The next slide represents some of the feedback loops
within the livestock production system that we are
trying to model in Selayar
R2
+ Population growth
Household income
B2
Supply of meat and
livestock products
+
+
+
+
+
Livestock
Production
+
B5
B7
+
Consumption of land and
water by population
+
B3
Land and water resource
available for livestock
production
-
Consumption of land and
water for livestock
production
B6
-
B4
Land and water resources
available for population
-
Land and water
+
degradation
+
B8
7
19/11/2016
Livestock Production Model
Livestock Supply Model
Livestock Demand Model
Livestock Price Model
8
19/11/2016
What do these Models do?
• They essentially model balancing loops within the
livestock production system that seek a dynamic
equilibrium between consumption, production and
price
• They also obey limits to growth, that is, livestock
production cannot exceed the carrying capacity of land
and water
• Lets look at a scenario
Population growth
1: Population
1:
2:
3:
4000
450
25
1:
2:
3:
2500
300
15
2: Houses
1
2
3
Total demand for meat
1:
600
1:
350
1
2
3
2
1
3
1
1
Population grows and reaches
limits to growth
3
2
1:
2:
3:
1000
150
5
1
1
1:
0.00
250.00
500.00
Weeks
Page 1
750.00
1000.00
12:30 PM Wed, 16 Nov 2016
100
0.00
250.00
Page 1
Untitled
6
1
2
1:
2:
750.00
1000.00
2:31 PM Wed, 16 Nov 2016
The desired household demand for
meat doubles at week 300 and
goes back to normal at week 700
Meat price
1: total desired liv estoc…y by liv estock ty pe[Cattle] 2: total f attening liv estock by liv estock ty pe[Cattle]
1
500.00
Weeks
Untitled
Total livestock production
1:
2:
There is no change in the desired
household demand for meat
1: total demand f or meat by liv estock ty pe[Cattle]
3: Urban land area
1
Population grows and reaches
limits to growth
1: actual meat price[Cattle]
1:
1.05
1:
0.8
1:
0.55
1
2
2
4
2
1
1
1:
2:
2
0.00
Page 1
250.00
500.00
Weeks
Untitled
750.00
1000.00
2:31 PM Wed, 16 Nov 2016
0.00
Page 1
1
250.00
1
500.00
Weeks
1
750.00
1000.00
12:30 PM Wed, 16 Nov 2016
Untitled
9
19/11/2016
Population growth
1: Population
1:
2:
3:
2: Houses
Total demand for meat
1
1
2
3
Key Lessons
1: total demand f or meat by liv estock ty pe[Cattle]
3: Urban land area
4000
450
25
1:
1100
1:
600
1
2
3
2
3
3
1:
2:
3:
2500
300
15
1
1
1
2
1
1:
2:
3:
1000
150
5
1
1:
0.00
250.00
500.00
Weeks
Page 1
750.00
1000.00
12:30 PM Wed, 16 Nov 2016
100
0.00
250.00
Page 1
Untitled
Total livestock production
Meat price
1: actual meat price[Cattle]
1:
8
7
2
1
750.00
1000.00
2:42 PM Wed, 16 Nov 2016
Untitled
1: total desired liv estoc…y by liv estock ty pe[Cattle] 2: total f attening liv estock by liv estock ty pe[Cattle]
1:
2:
500.00
Weeks
1.05
1
2
2
1
1:
2:
1
5
4.5
1:
0.8
1:
0.55
• When managing any system, understand the
relationship between system structure and system
behaviour
• Use multiple sources of knowledge to develop your
understanding, i.e. be multidisciplinary
• Use your understanding of system structure to
carefully target your interventions, otherwise your cure
may end up being worse than the disease
• When managing systems there is rarely a silver bullet
solution to problems, therefore multiple interventions
may be needed
2
1
1:
2:
1
2
2
0.00
Page 1
250.00
500.00
Weeks
Untitled
750.00
1000.00
2:42 PM Wed, 16 Nov 2016
0.00
Page 1
1
250.00
1
500.00
Weeks
750.00
1000.00
2:42 PM Wed, 16 Nov 2016
Untitled
10