thinking systems class 10 matt cohen, phd

18
Thinking Systems Class 10 Matt Cohen, PhD

Upload: rich

Post on 22-Feb-2016

35 views

Category:

Documents


0 download

DESCRIPTION

Thinking Systems Class 10 Matt Cohen, PhD. -. -. +. A Rat Infestation. Gainesville home built in 1928 No rats when we moved in Lived there for just under 2 years “Massive” control efforts by the end. Owners of 2 large dogs Exceedingly poor hunters Neighborhood of cat owners - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Thinking Systems Class 10 Matt Cohen, PhD

Thinking SystemsClass 10

Matt Cohen, PhD

Page 2: Thinking Systems Class 10 Matt Cohen, PhD

A Rat Infestation• Gainesville home built in 1928

– No rats when we moved in• Lived there for just under 2

years– “Massive” control efforts by the

end

• Owners of 2 large dogs– Exceedingly poor hunters

• Neighborhood of cat owners– Every direction (E, W, N, S) had

one or more felines– Drove the dogs crazy…ever-

vigilant border patrols

+

- -

Page 3: Thinking Systems Class 10 Matt Cohen, PhD

Elements of Systems

• Boundary (the yard, canine patrolled)• Inputs and outputs (cats, dead rats)• Internal components (rats, dogs, cats)• Interactions

– Positive interactions (rats breeding)– Negative interactions (cats on rats, dogs on cats)

Page 4: Thinking Systems Class 10 Matt Cohen, PhD

Why Systems?• Interactions create complexity

– Emergent behavior• Water is “wet”• Traffic snarls (even without accidents)• The Rise of Fall of Pet Rocks

• Thresholds (tipping points) exist– Predicting these is enormously important

• Global climate change, business cycles, disease epidemics• Epileptic seizures, landslides, fisheries collapse

• Systems aren’t more complex than we think, they are more complex than we can think.– But…we have to try!

$3.95 each (!)

Page 5: Thinking Systems Class 10 Matt Cohen, PhD

Key Attributes of Systems I.• Mutual causality

– Components affect each other, obscuring linear cause-effect • Popularity → sales → popularity• Poverty → soil erosion → poverty• Chicken → Egg → Chicken

• Indirect effects– Component A exerts control over

Component B via its action on Component C

A B

A B

C

Page 6: Thinking Systems Class 10 Matt Cohen, PhD

Indirect Effects - Aleutian Islands• Nutrients are essential for plant

and animal production– Phosphorus (P) is often limiting

nutrient• Essential for ribosomes and

metabolism• Limited geologic source in the

region• Amount of P controls the

productivity of the ecosystem• Grassland production of Aleutian

islands is P limited• Sea bird guano is a rich P source

– Was mined for fertilizer for years Abu

ndan

t P

Dep

lete

d P

Croll et al. (2005) - Science

Page 7: Thinking Systems Class 10 Matt Cohen, PhD

Nutrients and Sea Birds

• Seabirds eat fish from the sea but poop on land

• Major flow of P from sea to land that supports productive grasslands

MarineBirds

GrasslandProduction

Fish

Soil P

+

+

+

Page 8: Thinking Systems Class 10 Matt Cohen, PhD

Predator Control of Ecosystems• Introduce Arctic Foxes

– Top-predator– Seabirds never had a

terrestrial predator– Decimated the sea-bird

populations

MarineBirds

GrasslandProduction

Fish

Soil P

+

+

+

Arctic Foxes

-

Roughly 300% more soil P AND biomass on fox-free islands than on fox-infested islands

Page 9: Thinking Systems Class 10 Matt Cohen, PhD

Key Attributes of Systems II.• Consist of processes at

different space/time scales– Fast and slow variables

• Humans and viruses• Evolution and extinction• Supply and demand

• Systems are historically contingent– Deep dependence on what

happened in the past• The Great Unfolding• Beta-max, Bacteria, Base 10

A B

A

B

C

Page 10: Thinking Systems Class 10 Matt Cohen, PhD

Fast and Slow: Beer and the Business Cycle

• There exists a cycle of boom (bull) and bust (bear) periods in economic systems…WHY?

Page 11: Thinking Systems Class 10 Matt Cohen, PhD

A Systems View of Boom and Bust1. The structure of a system influences

behavior. Systems cause their own problems, not external forces or individual errors.– Distribution chains (and economies) contain fast

and slow moving parts– Communication between parts is LAGGED

2. Human systems include the way in which people make decisions.

3. People tend to focus on local optimization NOT global optimization.

Page 12: Thinking Systems Class 10 Matt Cohen, PhD

Consider a Typical Supply Chain• Retailer: Sells products, varying consumer demand, orders to

wholesalers for next weeks delivery• Wholesalers/Distributors: Distribute beer to multiple

retailers, orders to brewery for two weeks in the future• Brewery: Make beer, adjust production to demand• ALL

– Avoid the costs of excess and insufficient inventory

J. Sterman at MIT http://web.mit.edu/jsterman/www/SDG/beergame.html

Page 13: Thinking Systems Class 10 Matt Cohen, PhD

Beer Game Simulator

Brewery

Wholesaler

Distributor

Retailer

Team 1 Team 2 Team 3 Team 4

OR

DE

RS

EX

CE

SS

/B

AC

KLO

G

Oscillation

Amplification

Lag

Changing Demand

Page 14: Thinking Systems Class 10 Matt Cohen, PhD

Dependence on History: Algae, Nutrients, and Shallow Lakes

• Shallow lakes (< 10 m deep)• Two alternative “states”

– Rooted vegetation (macrophytes)– Algae

• Shifts between the two occur catastrophically, and BOTH can occur under the same environmental conditions

• Where you are depends on where you’ve been

Page 15: Thinking Systems Class 10 Matt Cohen, PhD

Self-Reinforcing Feedbacks in Shallow Lakes

• Rooted Plant State– Plants require clear water– Plants stabilize sediments– Stable sediments keep

water P concentrations low AND limit stirring

– Low P limits algae and high clarity favors rooted plants

• Algae State– Algae makes ooze– Ooze is easily stirred up,

making the water turbid and recycling P

– More P makes algae grow faster AND sediments looser via loss of plants

• Regime shifts due to combined effects:– Too much P (human pollution)– Disturbances (pollution affects vulnerability)

Page 16: Thinking Systems Class 10 Matt Cohen, PhD

Environmental Change and Ecosystem “State” Shifts

Typical Models of Nature

Emerging Model of Many Complex Systems

Scheffer et al. (2001) - Nature

Page 17: Thinking Systems Class 10 Matt Cohen, PhD

Thinking for Managing Complex Systems

• The “state” of a system is controlled by external forces AND internal interactions

• Indirect effects lead to surprising behavior• Fast and slow variables interact to create

instability– Spatial variability (local vs. global variable) also

• Managing for ONE THING often creates bigger problems later (discussion section)