1 of 50

Getting Started on the Right Foot with System Dynamics Modeling

George P. Richardson

Rockefeller College of Public Affairs and Policy

University at Albany, State University of New York

​

GPR@Albany.edu

G. P. Richardson

Teaching Workshop, March 2023

1

Rockefeller College of Public Affairs and Policy

University at Albany

2 of 50

A Teaching Plan that Builds on Success

G. P. Richardson

Teaching Workshop, March 2023

2

Rockefeller College of Public Affairs and Policy

University at Albany

3 of 50

Students come to us …

  • Knowing nothing of feedback loops in human affairs,
  • Knowing little about modeling in science and policy,
  • Inexperienced in systems thinking,
  • Unaware of stocks, unable to think reliably about them,
  • Ignorant or confused about what ‘nonlinear’ means,
  • Inexperienced in computer simulation,
  • Completely ignorant of the “endogenous point of view” of the dynamics of complex systems
  • Furthermore, they have no idea what a system dynamics model actually looks like.

G. P. Richardson

Teaching Workshop, March 2023

3

Rockefeller College of Public Affairs and Policy

University at Albany

4 of 50

What looks like a feedback loop but isn’t?  �What looks like a system dynamics model but isn’t?

​

We are aiming for understandings of our field �that we can present at the very beginning.

So this talk is about what good work in our field actually looks like.

G. P. Richardson

Teaching Workshop, March 2023

4

Rockefeller College of Public Affairs and Policy

University at Albany

5 of 50

Loops:�“Which of these things is not like the others?”

G. P. Richardson

Teaching Workshop, March 2023

5

Rockefeller College of Public Affairs and Policy

University at Albany

6 of 50

These arrows mean ‘and then’�

  • We start with some understandings of the problem and its systemic context,
  • And then we conceptualize (map) the system.
  • Then we build the beginnings of a model, which we then test to understand it.
  • Then that may lead to a new understandings, or to entirely new conceptualization, and then …

This picture is a sequence of steps in the modeling process, not a set of feedback loops.

​

G. P. Richardson

Teaching Workshop, March 2023

6

Rockefeller College of Public Affairs and Policy

University at Albany

7 of 50

Arrows here are flows of material

The words here represent stocks.

​

The arrows represent flows.

​

This is a view of the “carbon cycle.”

​

It’s not a causal diagram. It’s a stock-and-flow diagram.

​

G. P. Richardson

Teaching Workshop, March 2023

7

Rockefeller College of Public Affairs and Policy

University at Albany

8 of 50

Only this one is a feedback loop

It tells a very compelling and important

self-reinforcing story.

G. P. Richardson

Teaching Workshop, March 2023

8

Rockefeller College of Public Affairs and Policy

University at Albany

9 of 50

They look alike, �but only one is a feedback loop

G. P. Richardson

Teaching Workshop, March 2023

9

Rockefeller College of Public Affairs and Policy

University at Albany

10 of 50

A Tool for Thinking: A “Connection Circle”

An actual connection circle on the wall of a second grade classroom in in St. Louis.

G. P. Richardson

Teaching Workshop, March 2023

10

Rockefeller College of Public Affairs and Policy

University at Albany

11 of 50

Resulting Loop Diagram

This CLD was posted alongside the Connection Circle.

Excellent practice to try to pull out one or more well-formed feedback loops from connection circles!

G. P. Richardson

Teaching Workshop, March 2023

11

Rockefeller College of Public Affairs and Policy

University at Albany

12 of 50

The Connection Circle �Helps Uncover Feedback Loops

But by itself a Connection Circle is not a Feedback loop.

G. P. Richardson

Teaching Workshop, March 2023

12

Rockefeller College of Public Affairs and Policy

University at Albany

13 of 50

Adding Helpful Concepts

Water vapor

Clouds

G. P. Richardson

Teaching Workshop, March 2023

13

Rockefeller College of Public Affairs and Policy

University at Albany

14 of 50

Same Loop Rearranged

G. P. Richardson

Teaching Workshop, March 2023

14

Rockefeller College of Public Affairs and Policy

University at Albany

15 of 50

Showing Stocks�An important step here, but still looks reinforcing.

+

+

+

+

+

+

G. P. Richardson

Teaching Workshop, March 2023

15

Rockefeller College of Public Affairs and Policy

University at Albany

16 of 50

Representing the Actions as Flows�

The links are flows, not causal influences.

We wouldn’t call this a “feedback loop” and it’s certainly not reinforcing.

G. P. Richardson

Teaching Workshop, March 2023

16

Rockefeller College of Public Affairs and Policy

University at Albany

17 of 50

Getting started:

A connection circle

Systems thinking: What’s really going on.

G. P. Richardson

Teaching Workshop, March 2023

17

Rockefeller College of Public Affairs and Policy

University at Albany

18 of 50

Either of these is OK as long as �we realize both tell conserved flow stories and neither is a reinforcing feedback loop.

The one on the left probably communicates best.

G. P. Richardson

Teaching Workshop, March 2023

18

Rockefeller College of Public Affairs and Policy

University at Albany

19 of 50

�What does a �Well-Structured System Dynamics Model �actually look like?

We are talking about what strong work in our field actually looks like.

G. P. Richardson

Teaching Workshop, March 2023

19

Rockefeller College of Public Affairs and Policy

University at Albany

20 of 50

Keep in Mind:�Students come to us …

  • Knowing nothing of feedback loops in human affairs,
  • Knowing little about modeling in science and policy,
  • Inexperienced in systems thinking,
  • Unaware of stocks, unable to think reliably about them,
  • Ignorant or confused about what ‘nonlinear’ means,
  • Inexperienced in computer simulation,
  • Completely ignorant of the “endogenous point of view” of the dynamics of complex systems
  • Furthermore, they have no idea what a system dynamics model actually looks like.

G. P. Richardson

Teaching Workshop, March 2023

20

Rockefeller College of Public Affairs and Policy

University at Albany

21 of 50

So we have to show what insightful�Modeling looks like

  • The model I’ll show is about students becoming interested in system dynamics modeling.
  • We won’t show HOW to create it – just what a well-structured model actually looks like.
  • You want your models to look like this one.
  • You want your students’ models to look like this, too.
  • Later I’ll show some pictures your models should not look like.

​

G. P. Richardson

Teaching Workshop, March 2023

21

Rockefeller College of Public Affairs and Policy

University at Albany

22 of 50

A well-structured system dynamics model�

G. P. Richardson

Teaching Workshop, March 2023

22

Rockefeller College of Public Affairs and Policy

University at Albany

23 of 50

Building up the model�

G. P. Richardson

Teaching Workshop, March 2023

23

Rockefeller College of Public Affairs and Policy

University at Albany

24 of 50

Students and Teachers�

G. P. Richardson

Teaching Workshop, March 2023

24

Rockefeller College of Public Affairs and Policy

University at Albany

25 of 50

Teachers and Teacher Effectiveness�

G. P. Richardson

Teaching Workshop, March 2023

25

Rockefeller College of Public Affairs and Policy

University at Albany

26 of 50

The Complete Model�

G. P. Richardson

Teaching Workshop, March 2023

26

Rockefeller College of Public Affairs and Policy

University at Albany

27 of 50

And here’s how it behaves�

G. P. Richardson

Teaching Workshop, March 2023

27

Rockefeller College of Public Affairs and Policy

University at Albany

28 of 50

The story it tells

  • Teachers start teaching a few students System Dynamics.
  • Students like it and�tell others.
  • More students get involved.
  • Too many! They overload the teachers �and WOM turns negative.

G. P. Richardson

Teaching Workshop, March 2023

28

Rockefeller College of Public Affairs and Policy

University at Albany

29 of 50

What Should You Notice

A well-structured model!

  • Pressures for change emerge from stocks.
  • That is to say: Stocks are the sources of dynamics!
  • Flows identify what is changing.
  • Equations can be obvious (and probably should be).
  • “Effective teacher load” lasts over time, so it is modeled as a stock.
  • Here it is modeled as a special kind of stock, a “SMOOTH”, so we don’t bother with the pipe. (The only time a blue arrow goes INTO a stock.)

​

(But the model has minor flaws. We could talk about that.)

G. P. Richardson

Teaching Workshop, March 2023

29

Rockefeller College of Public Affairs and Policy

University at Albany

30 of 50

What do students know after seeing this?

A little about feedback loops in human affairs

    • Initial success can turn bad.

A little about modeling dynamic feedback systems

    • The correct “syntax” of a model diagram.

Some awareness of stocks and their importance

    • Stocks are the sources of pressures for change.

The beginnings of the “endogenous point of view”

    • Nobody told the students to stop coming! The dynamics come from the internal structure of the system.

A good idea of what a well-structured�system dynamics model actually looks like.

G. P. Richardson

Teaching Workshop, March 2023

30

Rockefeller College of Public Affairs and Policy

University at Albany

31 of 50

Do you see another potential stock here?�

G. P. Richardson

Teaching Workshop, March 2023

31

Rockefeller College of Public Affairs and Policy

University at Albany

32 of 50

Adding an adapting stock of Teachers�

(But this model has flaws too. We could talk about that.)

Red graphs original

Blue graphs revised

G. P. Richardson

Teaching Workshop, March 2023

32

Rockefeller College of Public Affairs and Policy

University at Albany

33 of 50

Fixing language�

Students completing course

Length of course

Current Teacher load

G. P. Richardson

Teaching Workshop, March 2023

33

Rockefeller College of Public Affairs and Policy

University at Albany

34 of 50

Another well-structured simple model –�A high school issue: Unwise Drinking

G. P. Richardson

Teaching Workshop, March 2023

34

Rockefeller College of Public Affairs and Policy

University at Albany

35 of 50

Unwise Drinking

G. P. Richardson

Teaching Workshop, March 2023

35

Rockefeller College of Public Affairs and Policy

University at Albany

36 of 50

Unwise Drinking

Alcohol impaired

G. P. Richardson

Teaching Workshop, March 2023

36

Rockefeller College of Public Affairs and Policy

University at Albany

37 of 50

Class Could Talk about (simulate?)�Preventing Unwise Drinking

G. P. Richardson

Teaching Workshop, March 2023

37

Rockefeller College of Public Affairs and Policy

University at Albany

38 of 50

Things to Avoid

“Causal mishmash”

(Unclear arithmetic, unclear causality)

Unused

stock

G. P. Richardson

Teaching Workshop, March 2023

38

Rockefeller College of Public Affairs and Policy

University at Albany

39 of 50

More things to Avoid

Flows flow into Stocks.

They usually don’t “cause” other things!

G. P. Richardson

Teaching Workshop, March 2023

39

Rockefeller College of Public Affairs and Policy

University at Albany

40 of 50

Things to Avoid

Causal Mishmash

Unused Stocks

G. P. Richardson

Teaching Workshop, March 2023

40

Rockefeller College of Public Affairs and Policy

University at Albany

41 of 50

Here the Students got it close to right!

  • Pressures coming from the stock of Tension!
  • Feedback loops from stock-to- flow-to-stock!
  • Stock-and-flow feedback thought is evident!

G. P. Richardson

Teaching Workshop, March 2023

41

Rockefeller College of Public Affairs and Policy

University at Albany

42 of 50

Work of a student who didn’t really know �what a well-structured model looks like

G. P. Richardson

Teaching Workshop, March 2023

42

Rockefeller College of Public Affairs and Policy

University at Albany

43 of 50

So how do you get students to create�something like this?

G. P. Richardson

Teaching Workshop, March 2023

43

Rockefeller College of Public Affairs and Policy

University at Albany

44 of 50

A Teaching Plan that Builds on Success

See Explore

Exercise existing models. Led by the teacher

Copy simple models. Get the copy to run correctly!

Play Experiment with existing models

Correct or improve structure. Find and fix an error

Model a “canned” model description. Turn text into a model

Add structure to an existing model. Extend a model

Think Learn how modelers think, and think like that

Model Conceptualization. A whole other story!

Problem > Graphs over time > Dynamic hypothesis > Arithmetic > …

Do I can do this now. Let me try.

Problems with familiar dynamics

& structure

Personally chosen problems

G. P. Richardson

Teaching Workshop, March 2023

44

Rockefeller College of Public Affairs and Policy

University at Albany

45 of 50

Benefits of Copying Models

  • Focusing on an existing, well-formulated piece of model structure
  • Attention to details of formulation and equation-writing before modeling skills are developed
  • Focusing on finding errors and making a model work, before modeling skills are developed
  • Students grow in confidence in their abilities to learn to be modelers

​

G. P. Richardson

Teaching Workshop, March 2023

45

Rockefeller College of Public Affairs and Policy

University at Albany

46 of 50

Benefits of Adding Structure

  • Beginning with a model that makes sense and runs
  • Adding a few equations that have to be created
  • Quick simulation-based feedback (“Ah!” or “Oops!”)
  • Taking small steps toward model building
  • Working toward starting with a blank page
  • Growing in capabilities and confidence
  • Understanding iterative modeling

G. P. Richardson

Teaching Workshop, March 2023

46

Rockefeller College of Public Affairs and Policy

University at Albany

47 of 50

More Benefits of Exercises �Copying Models and Adding Structure

  • Enables beginning simply
  • Builds understanding
  • Carries understanding from one iteration to the next
  • Familiarizes with core models in the field
  • Accumulates confidence in modeling at each stage
  • Grows understandings of complex structure and behavior
  • Emphasizes always having a running model!

​

G. P. Richardson

Teaching Workshop, March 2023

47

Rockefeller College of Public Affairs and Policy

University at Albany

48 of 50

A Teaching Plan that Builds on Success

See Explore

Exercise existing models.

Copy simple models.

Play Experiment with existing models

Correct or improve structure.

Model a “canned” model description.

Add structure to an existing model.

Think Learn how modelers think, and think like that

Model Conceptualization. A whole other story!

Problem > Graphs over time > Dynamic hypothesis > Arithmetic > …

Do I can do this now. Let me try.

Problems with familiar dynamics

& structure

Personally chosen problems

G. P. Richardson

Teaching Workshop, March 2023

48

Rockefeller College of Public Affairs and Policy

University at Albany

49 of 50

Questions?�

I’ll stay as long as you want.

G. P. Richardson

Teaching Workshop, March 2023

49

Rockefeller College of Public Affairs and Policy

University at Albany

50 of 50

For more on this approach

​

“Model teaching”

Notes I, II, & III

System Dynamics Review (2014 & 2015)

​

Notes and Insights, beginning with issue 30, 81–88 (2014)

​

G. P. Richardson

Teaching Workshop, March 2023

50

Rockefeller College of Public Affairs and Policy

University at Albany